From e74189cc19205cf6652eb1bc46eecabdbe3f6afa Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 22 Sep 2024 20:23:56 +0200 Subject: [PATCH 001/125] (feat): Add DX11 Win32 Backend to CMakeLists --- Backends/CMakeLists.txt | 11 +++++++++++ CMake/OptionsLists.cmake | 1 + 2 files changed, 12 insertions(+) diff --git a/Backends/CMakeLists.txt b/Backends/CMakeLists.txt index 49d6469ad..d4edcb84b 100644 --- a/Backends/CMakeLists.txt +++ b/Backends/CMakeLists.txt @@ -34,6 +34,17 @@ target_sources(rmlui_backend_Win32_VK INTERFACE ) target_link_libraries(rmlui_backend_Win32_VK INTERFACE rmlui_backend_common_headers) +add_library(rmlui_backend_Win32_DX11 INTERFACE) +target_sources(rmlui_backend_Win32_DX11 INTERFACE + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Platform_Win32.cpp" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Renderer_DX11.cpp" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Backend_Win32_DX11.cpp" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Platform_Win32.h" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Renderer_DX11.h" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Include_Windows.h" +) +target_link_libraries(rmlui_backend_Win32_DX11 INTERFACE rmlui_backend_common_headers d3d11.lib dxgi.lib ) + add_library(rmlui_backend_X11_GL2 INTERFACE) target_sources(rmlui_backend_X11_GL2 INTERFACE "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Platform_X11.cpp" diff --git a/CMake/OptionsLists.cmake b/CMake/OptionsLists.cmake index 67a437246..90d0d8a9e 100644 --- a/CMake/OptionsLists.cmake +++ b/CMake/OptionsLists.cmake @@ -7,6 +7,7 @@ set(RMLUI_BACKEND_OPTIONS "native" "Win32_GL2" "Win32_VK" + "Win32_DX11" "X11_GL2" "SDL_GL2" "SDL_GL3" From 5be2f24f6b120f32dfd8bb17ea4040b3f8811747 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 22 Sep 2024 20:24:19 +0200 Subject: [PATCH 002/125] (feat): Implement stub DX11 backend. Not usable in current state. --- Backends/RmlUi_Backend_Win32_DX11.cpp | 486 ++++++++++++++++++++++++++ Backends/RmlUi_Renderer_DX11.cpp | 249 +++++++++++++ Backends/RmlUi_Renderer_DX11.h | 106 ++++++ 3 files changed, 841 insertions(+) create mode 100644 Backends/RmlUi_Backend_Win32_DX11.cpp create mode 100644 Backends/RmlUi_Renderer_DX11.cpp create mode 100644 Backends/RmlUi_Renderer_DX11.h diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp new file mode 100644 index 000000000..bdcc885df --- /dev/null +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -0,0 +1,486 @@ +/* +* This source file is part of RmlUi, the HTML/CSS Interface Middleware +* +* For the latest information, see http://github.com/mikke89/RmlUi +* +* Copyright (c) 2008-2010 CodePoint Ltd, Shift Technology Ltd +* Copyright (c) 2019-2023 The RmlUi Team, and contributors +* +* Permission is hereby granted, free of charge, to any person obtaining a copy +* of this software and associated documentation files (the "Software"), to deal +* in the Software without restriction, including without limitation the rights +* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +* copies of the Software, and to permit persons to whom the Software is +* furnished to do so, subject to the following conditions: +* +* The above copyright notice and this permission notice shall be included in +* all copies or substantial portions of the Software. +* +* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +* THE SOFTWARE. +* +*/ + +#include "RmlUi_Backend.h" +#include "RmlUi_Include_Windows.h" +#include "RmlUi_Platform_Win32.h" +#include "RmlUi_Renderer_DX11.h" +#include +#include +#include +#include +#include + +/** +High DPI support using Windows Per Monitor V2 DPI awareness. + +Requires Windows 10, version 1703. +*/ +#ifndef DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2 +#define DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2 ((HANDLE)-4) +#endif +#ifndef WM_DPICHANGED +#define WM_DPICHANGED 0x02E0 +#endif + +// Declare pointers to the DPI aware Windows API functions. +using ProcSetProcessDpiAwarenessContext = BOOL(WINAPI*)(HANDLE value); +using ProcGetDpiForWindow = UINT(WINAPI*)(HWND hwnd); +using ProcAdjustWindowRectExForDpi = BOOL(WINAPI*)(LPRECT lpRect, DWORD dwStyle, BOOL bMenu, DWORD dwExStyle, UINT dpi); + +static bool has_dpi_support = false; +static ProcSetProcessDpiAwarenessContext procSetProcessDpiAwarenessContext = NULL; +static ProcGetDpiForWindow procGetDpiForWindow = NULL; +static ProcAdjustWindowRectExForDpi procAdjustWindowRectExForDpi = NULL; + +// Make ourselves DPI aware on supported Windows versions. +static void InitializeDpiSupport() +{ + // Cast function pointers to void* first for MinGW not to emit errors. + procSetProcessDpiAwarenessContext = + (ProcSetProcessDpiAwarenessContext)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "SetProcessDpiAwarenessContext"); + procGetDpiForWindow = (ProcGetDpiForWindow)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "GetDpiForWindow"); + procAdjustWindowRectExForDpi = + (ProcAdjustWindowRectExForDpi)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "AdjustWindowRectExForDpi"); + + if (!has_dpi_support && procSetProcessDpiAwarenessContext != NULL && procGetDpiForWindow != NULL && procAdjustWindowRectExForDpi != NULL) + { + // Activate Per Monitor V2. + if (procSetProcessDpiAwarenessContext(DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2)) + has_dpi_support = true; + } +} + +static UINT GetWindowDpi(HWND window_handle) +{ + if (has_dpi_support) + { + UINT dpi = procGetDpiForWindow(window_handle); + if (dpi != 0) + return dpi; + } + return USER_DEFAULT_SCREEN_DPI; +} + +static float GetDensityIndependentPixelRatio(HWND window_handle) +{ + return float(GetWindowDpi(window_handle)) / float(USER_DEFAULT_SCREEN_DPI); +} + +// Create the window but don't show it yet. Returns the pixel size of the window, which may be different than the passed size due to DPI settings. +static HWND InitializeWindow(HINSTANCE instance_handle, const std::wstring& name, int& inout_width, int& inout_height, bool allow_resize); + +// D3D Creation / Cleanup functions +static bool CreateDeviceD3D(HWND hwnd); +static void CleanupDeviceD3D(); +static void CreateRenderTarget(); +static void CleanupRenderTarget(); + +/** +Global data used by this backend. + +Lifetime governed by the calls to Backend::Initialize() and Backend::Shutdown(). +*/ +struct BackendData { + SystemInterface_Win32 system_interface; + RenderInterface_DX11 render_interface; + TextInputMethodEditor_Win32 text_input_method_editor; + + HINSTANCE instance_handle = nullptr; + std::wstring instance_name; + HWND window_handle = nullptr; + + // D3D11 resources + struct { + ID3D11Device* pd3dDevice = nullptr; + ID3D11DeviceContext* pd3dDeviceContext = nullptr; + IDXGISwapChain* pSwapChain = nullptr; + bool swapchain_cccluded = false; + ID3D11RenderTargetView* pMainRenderTargetView = nullptr; + } d3d_resources; + + bool context_dimensions_dirty = true; + Rml::Vector2i window_dimensions; + bool running = true; + + // Arguments set during event processing and nulled otherwise. + Rml::Context* context = nullptr; + KeyDownCallback key_down_callback = nullptr; +}; +static Rml::UniquePtr data; + +bool Backend::Initialize(const char* window_name, int width, int height, bool allow_resize) +{ + RMLUI_ASSERT(!data); + + const std::wstring name = RmlWin32::ConvertToUTF16(Rml::String(window_name)); + + data = Rml::MakeUnique(); + data->instance_handle = GetModuleHandle(nullptr); + data->instance_name = name.data(); + + InitializeDpiSupport(); + + // Initialize the window but don't show it yet. + HWND window_handle = InitializeWindow(data->instance_handle, name, width, height, allow_resize); + if (!window_handle) + return false; + + if (!CreateDeviceD3D(window_handle)) { + ::CloseWindow(window_handle); + return false; + } + + data->render_interface.Init( + data->d3d_resources.pd3dDevice, + data->d3d_resources.pd3dDeviceContext, + data->d3d_resources.pSwapChain, + data->d3d_resources.pMainRenderTargetView); + + data->window_handle = window_handle; + data->system_interface.SetWindow(window_handle); + + // Now we are ready to show the window. + ::ShowWindow(window_handle, SW_SHOW); + ::SetForegroundWindow(window_handle); + ::SetFocus(window_handle); + + // Provide a backend-specific text input handler to manage the IME. + Rml::SetTextInputHandler(&data->text_input_method_editor); + + return true; +} + +void Backend::Shutdown() +{ + RMLUI_ASSERT(data); + + // As we forcefully override the global text input handler, we must reset it before the data is destroyed to avoid any potential use-after-free. + if (Rml::GetTextInputHandler() == &data->text_input_method_editor) + Rml::SetTextInputHandler(nullptr); + + CleanupDeviceD3D(); + + ::DestroyWindow(data->window_handle); + ::UnregisterClassW((LPCWSTR)data->instance_name.data(), data->instance_handle); + + data.reset(); +} + +Rml::SystemInterface* Backend::GetSystemInterface() +{ + RMLUI_ASSERT(data); + return &data->system_interface; +} + +Rml::RenderInterface* Backend::GetRenderInterface() +{ + RMLUI_ASSERT(data); + return &data->render_interface; +} + +static bool NextEvent(MSG& message, UINT timeout) +{ + if (timeout != 0) + { + UINT_PTR timer_id = SetTimer(NULL, 0, timeout, NULL); + BOOL res = GetMessage(&message, NULL, 0, 0); + KillTimer(NULL, timer_id); + if (message.message != WM_TIMER || message.hwnd != nullptr || message.wParam != timer_id) + return res; + } + return PeekMessage(&message, nullptr, 0, 0, PM_REMOVE); +} + +bool Backend::ProcessEvents(Rml::Context* context, KeyDownCallback key_down_callback, bool power_save) +{ + RMLUI_ASSERT(data && context); + + // The initial window size may have been affected by system DPI settings, apply the actual pixel size and dp-ratio to the context. + if (data->context_dimensions_dirty) + { + data->context_dimensions_dirty = false; + const float dp_ratio = GetDensityIndependentPixelRatio(data->window_handle); + context->SetDimensions(data->window_dimensions); + context->SetDensityIndependentPixelRatio(dp_ratio); + } + + data->context = context; + data->key_down_callback = key_down_callback; + + MSG message; + bool has_message = NextEvent(message, power_save ? static_cast(Rml::Math::Min(context->GetNextUpdateDelay(), 10.0) * 1000.0) : 0); + while (has_message) + { + // Dispatch the message to our local event handler below. + TranslateMessage(&message); + DispatchMessage(&message); + + has_message = NextEvent(message, 0); + } + + data->context = nullptr; + data->key_down_callback = nullptr; + + const bool result = data->running; + data->running = true; + return result; +} + +void Backend::RequestExit() +{ + RMLUI_ASSERT(data); + data->running = false; +} + +void Backend::BeginFrame() +{ + RMLUI_ASSERT(data); + // @TODO: Check if should resize, if so, re-create swapchain + data->render_interface.BeginFrame(); +} + +void Backend::PresentFrame() +{ + RMLUI_ASSERT(data); + data->render_interface.EndFrame(); + + // Present + HRESULT hr = data->d3d_resources.pSwapChain->Present(1, 0); // Present with vsync + //HRESULT hr = g_pSwapChain->Present(0, 0); // Present without vsync + data->d3d_resources.swapchain_cccluded = (hr == DXGI_STATUS_OCCLUDED); + + // Optional, used to mark frames during performance profiling. + RMLUI_FrameMark; +} + +// Local event handler for window and input events. +static LRESULT CALLBACK WindowProcedureHandler(HWND window_handle, UINT message, WPARAM w_param, LPARAM l_param) +{ + RMLUI_ASSERT(data); + + switch (message) + { + case WM_CLOSE: + { + data->running = false; + return 0; + } + break; + case WM_SIZE: + { + const int width = LOWORD(l_param); + const int height = HIWORD(l_param); + data->window_dimensions.x = width; + data->window_dimensions.y = height; + data->render_interface.SetViewport(width, height); + if (data->context) + data->context->SetDimensions(data->window_dimensions); + return 0; + } + break; + case WM_DPICHANGED: + { + RECT* new_pos = (RECT*)l_param; + SetWindowPos(window_handle, NULL, new_pos->left, new_pos->top, new_pos->right - new_pos->left, new_pos->bottom - new_pos->top, + SWP_NOZORDER | SWP_NOACTIVATE); + if (data->context && has_dpi_support) + data->context->SetDensityIndependentPixelRatio(GetDensityIndependentPixelRatio(window_handle)); + return 0; + } + break; + case WM_KEYDOWN: + { + // Override the default key event callback to add global shortcuts for the samples. + Rml::Context* context = data->context; + KeyDownCallback key_down_callback = data->key_down_callback; + + const Rml::Input::KeyIdentifier rml_key = RmlWin32::ConvertKey((int)w_param); + const int rml_modifier = RmlWin32::GetKeyModifierState(); + const float native_dp_ratio = GetDensityIndependentPixelRatio(window_handle); + + // See if we have any global shortcuts that take priority over the context. + if (key_down_callback && !key_down_callback(context, rml_key, rml_modifier, native_dp_ratio, true)) + return 0; + // Otherwise, hand the event over to the context by calling the input handler as normal. + if (!RmlWin32::WindowProcedure(context, data->text_input_method_editor, window_handle, message, w_param, l_param)) + return 0; + // The key was not consumed by the context either, try keyboard shortcuts of lower priority. + if (key_down_callback && !key_down_callback(context, rml_key, rml_modifier, native_dp_ratio, false)) + return 0; + return 0; + } + break; + default: + { + // Submit it to the platform handler for default input handling. + if (!RmlWin32::WindowProcedure(data->context, data->text_input_method_editor, window_handle, message, w_param, l_param)) + return 0; + } + break; + } + + // All unhandled messages go to DefWindowProc. + return DefWindowProc(window_handle, message, w_param, l_param); +} + +static HWND InitializeWindow(HINSTANCE instance_handle, const std::wstring& name, int& inout_width, int& inout_height, bool allow_resize) +{ + // Fill out the window class struct. + WNDCLASSW window_class; + ZeroMemory(&window_class, sizeof(window_class)); + window_class.style = CS_HREDRAW | CS_VREDRAW | CS_OWNDC; + window_class.lpfnWndProc = &WindowProcedureHandler; // Attach our local event handler. + window_class.cbClsExtra = 0; + window_class.cbWndExtra = 0; + window_class.hInstance = instance_handle; + window_class.hIcon = LoadIcon(nullptr, IDI_WINLOGO); + window_class.hCursor = LoadCursor(nullptr, IDC_ARROW); + window_class.hbrBackground = nullptr; + window_class.lpszMenuName = nullptr; + window_class.lpszClassName = name.data(); + + if (!RegisterClassW(&window_class)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Failed to register window class"); + return nullptr; + } + + HWND window_handle = CreateWindowExW(WS_EX_APPWINDOW | WS_EX_WINDOWEDGE, + name.data(), // Window class name. + name.data(), WS_CLIPSIBLINGS | WS_CLIPCHILDREN | WS_OVERLAPPEDWINDOW, 0, 0, // Window position. + 0, 0, // Window size. + nullptr, nullptr, instance_handle, nullptr); + + if (!window_handle) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Failed to create window"); + return nullptr; + } + + UINT window_dpi = GetWindowDpi(window_handle); + inout_width = (inout_width * (int)window_dpi) / USER_DEFAULT_SCREEN_DPI; + inout_height = (inout_height * (int)window_dpi) / USER_DEFAULT_SCREEN_DPI; + + DWORD style = (allow_resize ? WS_OVERLAPPEDWINDOW : (WS_OVERLAPPEDWINDOW & ~WS_SIZEBOX & ~WS_MAXIMIZEBOX)); + DWORD extended_style = WS_EX_APPWINDOW | WS_EX_WINDOWEDGE; + + // Adjust the window size to take the edges into account. + RECT window_rect; + window_rect.top = 0; + window_rect.left = 0; + window_rect.right = inout_width; + window_rect.bottom = inout_height; + if (has_dpi_support) + procAdjustWindowRectExForDpi(&window_rect, style, FALSE, extended_style, window_dpi); + else + AdjustWindowRectEx(&window_rect, style, FALSE, extended_style); + + SetWindowLong(window_handle, GWL_EXSTYLE, extended_style); + SetWindowLong(window_handle, GWL_STYLE, style); + + // Resize the window and center it on the screen. + Rml::Vector2i screen_size = {GetSystemMetrics(SM_CXSCREEN), GetSystemMetrics(SM_CYSCREEN)}; + Rml::Vector2i window_size = {int(window_rect.right - window_rect.left), int(window_rect.bottom - window_rect.top)}; + Rml::Vector2i window_pos = Rml::Math::Max((screen_size - window_size) / 2, Rml::Vector2i(0)); + + SetWindowPos(window_handle, HWND_TOP, window_pos.x, window_pos.y, window_size.x, window_size.y, SWP_NOACTIVATE); + + return window_handle; +} + +static bool CreateDeviceD3D(HWND hwnd) { + // Setup swap chain + DXGI_SWAP_CHAIN_DESC sd; + ZeroMemory(&sd, sizeof(sd)); + sd.BufferCount = 2; + sd.BufferDesc.Width = 0; + sd.BufferDesc.Height = 0; + sd.BufferDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM; + sd.BufferDesc.RefreshRate.Numerator = 60; + sd.BufferDesc.RefreshRate.Denominator = 1; + sd.Flags = DXGI_SWAP_CHAIN_FLAG_ALLOW_MODE_SWITCH; + sd.BufferUsage = DXGI_USAGE_RENDER_TARGET_OUTPUT; + sd.OutputWindow = hwnd; + sd.SampleDesc.Count = 1; + sd.SampleDesc.Quality = 0; + sd.Windowed = TRUE; + sd.SwapEffect = DXGI_SWAP_EFFECT_DISCARD; + + UINT createDeviceFlags = 0; + //createDeviceFlags |= D3D11_CREATE_DEVICE_DEBUG; + D3D_FEATURE_LEVEL featureLevel; + const D3D_FEATURE_LEVEL featureLevelArray[2] = { D3D_FEATURE_LEVEL_11_0, D3D_FEATURE_LEVEL_10_0, }; + HRESULT res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_HARDWARE, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, + &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); + if (res == DXGI_ERROR_UNSUPPORTED) // Try high-performance WARP software driver if hardware is not available. + res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_WARP, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, + &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); + if (res != S_OK) + return false; + + CreateRenderTarget(); + return true; +} + +static void CleanupDeviceD3D() { + + CleanupRenderTarget(); + if (data->d3d_resources.pSwapChain) + { + data->d3d_resources.pSwapChain->Release(); + data->d3d_resources.pSwapChain = nullptr; + } + if (data->d3d_resources.pd3dDeviceContext) + { + data->d3d_resources.pd3dDeviceContext->Release(); + data->d3d_resources.pd3dDeviceContext = nullptr; + } + if (data->d3d_resources.pd3dDevice) + { + data->d3d_resources.pd3dDevice->Release(); + data->d3d_resources.pd3dDevice = nullptr; + } +} + +static void CreateRenderTarget() +{ + ID3D11Texture2D* pBackBuffer = nullptr; + data->d3d_resources.pSwapChain->GetBuffer(0, IID_PPV_ARGS(&pBackBuffer)); + data->d3d_resources.pd3dDevice->CreateRenderTargetView(pBackBuffer, nullptr, &data->d3d_resources.pMainRenderTargetView); + pBackBuffer->Release(); +} + +static void CleanupRenderTarget() +{ + if (data->d3d_resources.pMainRenderTargetView) + { + data->d3d_resources.pMainRenderTargetView->Release(); + data->d3d_resources.pMainRenderTargetView = nullptr; + } +} diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp new file mode 100644 index 000000000..39f05af88 --- /dev/null +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -0,0 +1,249 @@ +/* + * This source file is part of RmlUi, the HTML/CSS Interface Middleware + * + * For the latest information, see http://github.com/mikke89/RmlUi + * + * Copyright (c) 2008-2010 CodePoint Ltd, Shift Technology Ltd + * Copyright (c) 2019-2023 The RmlUi Team, and contributors + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + * + */ + +#include + +#if defined RMLUI_PLATFORM_WIN32 + + #include "RmlUi_Renderer_DX11.h" + #include + #include + #include + + #include "RmlUi_Include_Windows.h" + +RenderInterface_DX11::RenderInterface_DX11() { + +} + +void RenderInterface_DX11::Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain, + ID3D11RenderTargetView* mainRenderTargetView) +{ + RMLUI_ASSERTMSG(pd3dDevice, "pd3dDevice cannot be nullptr!"); + RMLUI_ASSERTMSG(pSwapChain, "pSwapChain cannot be nullptr!"); + RMLUI_ASSERTMSG(pd3dDeviceContext, "pd3dDeviceContext cannot be nullptr!"); + RMLUI_ASSERTMSG(mainRenderTargetView, "mainRenderTargetView cannot be nullptr!"); + + // Assign D3D resources + m_d3dDevice = pd3dDevice; + m_d3dContext = pd3dDeviceContext; + m_swapChain = pSwapChain; + m_mainRenderTargetView = mainRenderTargetView; + + // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. + // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). + if (!m_blendState) + { + D3D11_BLEND_DESC blendDesc; + ZeroMemory(&blendDesc, sizeof(blendDesc)); + blendDesc.AlphaToCoverageEnable = FALSE; + blendDesc.IndependentBlendEnable = FALSE; + blendDesc.RenderTarget[0].BlendEnable = TRUE; + blendDesc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE; + blendDesc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA; + blendDesc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; + blendDesc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ONE; + blendDesc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA; + blendDesc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; + blendDesc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; + HRESULT result = m_d3dDevice->CreateBlendState(&blendDesc, &m_blendState); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device1::CreateBlendState (%d)", result); + } + } +} + +void RenderInterface_DX11::Cleanup() { + m_blendState->Release(); +} + +void RenderInterface_DX11::BeginFrame() { + RMLUI_ASSERTMSG(m_d3dContext, "d3dContext cannot be nullptr!"); + RMLUI_ASSERTMSG(m_d3dDevice, "d3dDevice cannot be nullptr!"); + + D3D11_VIEWPORT d3dviewport; + d3dviewport.TopLeftX = 0; + d3dviewport.TopLeftY = 0; + d3dviewport.Width = m_width; + d3dviewport.Height = m_height; + d3dviewport.MinDepth = 0.0f; + d3dviewport.MaxDepth = 1.0f; + m_d3dContext->RSSetViewports(1, &d3dviewport); + Clear(); + SetBlendState(m_blendState); +} + +void RenderInterface_DX11::EndFrame() { + +} + + +Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span vertices, Rml::Span indices) { + return Rml::CompiledGeometryHandle(0); +} + +void RenderInterface_DX11::ReleaseGeometry(Rml::CompiledGeometryHandle geometry) { + +} + +void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) { + +} + +// Set to byte packing, or the compiler will expand our struct, which means it won't read correctly from file +#pragma pack(1) +struct TGAHeader { + char idLength; + char colourMapType; + char dataType; + short int colourMapOrigin; + short int colourMapLength; + char colourMapDepth; + short int xOrigin; + short int yOrigin; + short int width; + short int height; + char bitsPerPixel; + char imageDescriptor; +}; +// Restore packing +#pragma pack() + + +Rml::TextureHandle RenderInterface_DX11::LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) { + Rml::FileInterface* file_interface = Rml::GetFileInterface(); + Rml::FileHandle file_handle = file_interface->Open(source); + if (!file_handle) + { + return false; + } + + file_interface->Seek(file_handle, 0, SEEK_END); + size_t buffer_size = file_interface->Tell(file_handle); + file_interface->Seek(file_handle, 0, SEEK_SET); + + if (buffer_size <= sizeof(TGAHeader)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Texture file size is smaller than TGAHeader, file is not a valid TGA image."); + file_interface->Close(file_handle); + return false; + } + + using Rml::byte; + Rml::UniquePtr buffer(new byte[buffer_size]); + file_interface->Read(buffer.get(), buffer_size, file_handle); + file_interface->Close(file_handle); + + TGAHeader header; + memcpy(&header, buffer.get(), sizeof(TGAHeader)); + + int color_mode = header.bitsPerPixel / 8; + const size_t image_size = header.width * header.height * 4; // We always make 32bit textures + + if (header.dataType != 2) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Only 24/32bit uncompressed TGAs are supported."); + return false; + } + + // Ensure we have at least 3 colors + if (color_mode < 3) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Only 24 and 32bit textures are supported."); + return false; + } + + const byte* image_src = buffer.get() + sizeof(TGAHeader); + Rml::UniquePtr image_dest_buffer(new byte[image_size]); + byte* image_dest = image_dest_buffer.get(); + + // Targa is BGR, swap to RGB, flip Y axis, and convert to premultiplied alpha. + for (long y = 0; y < header.height; y++) + { + long read_index = y * header.width * color_mode; + long write_index = ((header.imageDescriptor & 32) != 0) ? read_index : (header.height - y - 1) * header.width * 4; + for (long x = 0; x < header.width; x++) + { + image_dest[write_index] = image_src[read_index + 2]; + image_dest[write_index + 1] = image_src[read_index + 1]; + image_dest[write_index + 2] = image_src[read_index]; + if (color_mode == 4) + { + const byte alpha = image_src[read_index + 3]; + for (size_t j = 0; j < 3; j++) + image_dest[write_index + j] = byte((image_dest[write_index + j] * alpha) / 255); + image_dest[write_index + 3] = alpha; + } + else + image_dest[write_index + 3] = 255; + + write_index += 4; + read_index += color_mode; + } + } + + texture_dimensions.x = header.width; + texture_dimensions.y = header.height; + + return GenerateTexture({image_dest, image_size}, texture_dimensions); +} + +Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) { + return Rml::TextureHandle(0); +} + +void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) { + +} + +void RenderInterface_DX11::EnableScissorRegion(bool enable) { + +} + +void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) { +} + +void RenderInterface_DX11::SetViewport(const int width, const int height) { + m_width = width; + m_height = height; +} + +void RenderInterface_DX11::Clear() { + float clearColor[4] = { 0.0f, 0.0f, 0.0f, 1.0f }; + m_d3dContext->ClearRenderTargetView(m_mainRenderTargetView, clearColor); +} + +void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) { + if (blendState != m_currentBlendState) + { + m_d3dContext->OMSetBlendState(blendState, 0, 0xFFFFFFFF); + m_currentBlendState = blendState; + } +} +#endif // RMLUI_PLATFORM_WIN32 \ No newline at end of file diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h new file mode 100644 index 000000000..c889193ae --- /dev/null +++ b/Backends/RmlUi_Renderer_DX11.h @@ -0,0 +1,106 @@ +/* +* This source file is part of RmlUi, the HTML/CSS Interface Middleware +* +* For the latest information, see http://github.com/mikke89/RmlUi +* +* Copyright (c) 2008-2010 CodePoint Ltd, Shift Technology Ltd +* Copyright (c) 2019-2023 The RmlUi Team, and contributors +* +* Permission is hereby granted, free of charge, to any person obtaining a copy +* of this software and associated documentation files (the "Software"), to deal +* in the Software without restriction, including without limitation the rights +* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +* copies of the Software, and to permit persons to whom the Software is +* furnished to do so, subject to the following conditions: +* +* The above copyright notice and this permission notice shall be included in +* all copies or substantial portions of the Software. +* +* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +* THE SOFTWARE. +* +*/ + +#ifndef RMLUI_BACKENDS_RENDERER_DX11_H +#define RMLUI_BACKENDS_RENDERER_DX11_H + + #include + + #if defined RMLUI_PLATFORM_WIN32 + #include + + #include + #include + +class RenderInterface_DX11 : public Rml::RenderInterface { +public: + RenderInterface_DX11(); + + // Resource initialisation and cleanup + void Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain, ID3D11RenderTargetView* mainRenderTargetView); + void Cleanup (); + + // Sets up DirectX11 states for taking rendering commands from RmlUi. + void BeginFrame(); + void EndFrame(); + + // -- Inherited from Rml::RenderInterface -- + + Rml::CompiledGeometryHandle CompileGeometry(Rml::Span vertices, Rml::Span indices) override; + void ReleaseGeometry(Rml::CompiledGeometryHandle geometry) override; + void RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) override; + + Rml::TextureHandle LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) override; + Rml::TextureHandle GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) override; + void ReleaseTexture(Rml::TextureHandle texture_handle) override; + + void EnableScissorRegion(bool enable) override; + void SetScissorRegion(Rml::Rectanglei region) override; + + void SetViewport(const int width, const int height); + void Clear(); + +private: + void SetBlendState(ID3D11BlendState* blendState); + +private: + // @TODO: Replace with vertex / index buffer pairs + struct GeometryView { + Rml::Span vertices; + Rml::Span indices; + }; + + // D3D11 core resources + ID3D11Device* m_d3dDevice = nullptr; + ID3D11DeviceContext* m_d3dContext = nullptr; + IDXGISwapChain* m_swapChain = nullptr; + ID3D11RenderTargetView* m_mainRenderTargetView = nullptr; + + // Viewport dimensions + int m_width = 0; + int m_height = 0; + + // SDL_Renderer* renderer; + ID3D11BlendState* m_blendState = nullptr; + // SDL_Rect rect_scissor = {}; + bool scissor_region_enabled = false; + + // D3D11 state + ID3D11BlendState* m_currentBlendState = nullptr; + + struct D3D11State { + + ID3D11BlendState* m_previousBlendState = nullptr; + }; + + D3D11State m_previousD3DState; +}; + +#endif // RMLUI_PLATFORM_WIN32 + +#endif // RMLUI_BACKENDS_RENDERER_DX11_H From e1da400c98d25f6a4f3b09c4e7f4cf7e9e48a2c2 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 01:08:43 +0200 Subject: [PATCH 003/125] (chore): Use spaces instead of tabs --- Backends/RmlUi_Renderer_DX11.cpp | 63 +++++++++++++++++--------------- Backends/RmlUi_Renderer_DX11.h | 46 +++++++++++------------ 2 files changed, 56 insertions(+), 53 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 39f05af88..85f4f9a3e 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -30,12 +30,12 @@ #if defined RMLUI_PLATFORM_WIN32 - #include "RmlUi_Renderer_DX11.h" - #include - #include - #include + #include "RmlUi_Renderer_DX11.h" + #include + #include + #include - #include "RmlUi_Include_Windows.h" + #include "RmlUi_Include_Windows.h" RenderInterface_DX11::RenderInterface_DX11() { @@ -80,7 +80,7 @@ void RenderInterface_DX11::Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* p } void RenderInterface_DX11::Cleanup() { - m_blendState->Release(); + m_blendState->Release(); } void RenderInterface_DX11::BeginFrame() { @@ -105,7 +105,7 @@ void RenderInterface_DX11::EndFrame() { Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span vertices, Rml::Span indices) { - return Rml::CompiledGeometryHandle(0); + return Rml::CompiledGeometryHandle(0); } void RenderInterface_DX11::ReleaseGeometry(Rml::CompiledGeometryHandle geometry) { @@ -119,18 +119,18 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm // Set to byte packing, or the compiler will expand our struct, which means it won't read correctly from file #pragma pack(1) struct TGAHeader { - char idLength; - char colourMapType; - char dataType; - short int colourMapOrigin; - short int colourMapLength; - char colourMapDepth; - short int xOrigin; - short int yOrigin; - short int width; - short int height; - char bitsPerPixel; - char imageDescriptor; + char idLength; + char colourMapType; + char dataType; + short int colourMapOrigin; + short int colourMapLength; + char colourMapDepth; + short int xOrigin; + short int yOrigin; + short int width; + short int height; + char bitsPerPixel; + char imageDescriptor; }; // Restore packing #pragma pack() @@ -229,21 +229,24 @@ void RenderInterface_DX11::EnableScissorRegion(bool enable) { void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) { } -void RenderInterface_DX11::SetViewport(const int width, const int height) { - m_width = width; - m_height = height; +void RenderInterface_DX11::SetViewport(const int width, const int height) +{ + m_width = width; + m_height = height; } -void RenderInterface_DX11::Clear() { - float clearColor[4] = { 0.0f, 0.0f, 0.0f, 1.0f }; - m_d3dContext->ClearRenderTargetView(m_mainRenderTargetView, clearColor); +void RenderInterface_DX11::Clear() +{ + float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; + m_d3dContext->ClearRenderTargetView(m_boundRenderTarget, clearColor); } void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) { - if (blendState != m_currentBlendState) - { - m_d3dContext->OMSetBlendState(blendState, 0, 0xFFFFFFFF); - m_currentBlendState = blendState; - } + if (blendState != m_currentBlendState) + { + m_d3dContext->OMSetBlendState(blendState, 0, 0xFFFFFFFF); + m_currentBlendState = blendState; + } } + #endif // RMLUI_PLATFORM_WIN32 \ No newline at end of file diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index c889193ae..6d2d7bfba 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -29,44 +29,44 @@ #ifndef RMLUI_BACKENDS_RENDERER_DX11_H #define RMLUI_BACKENDS_RENDERER_DX11_H - #include + #include - #if defined RMLUI_PLATFORM_WIN32 - #include + #if defined RMLUI_PLATFORM_WIN32 + #include - #include - #include + #include + #include class RenderInterface_DX11 : public Rml::RenderInterface { public: - RenderInterface_DX11(); - - // Resource initialisation and cleanup void Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain, ID3D11RenderTargetView* mainRenderTargetView); - void Cleanup (); + RenderInterface_DX11(); + + // Resource initialisation and cleanup + void Cleanup (); - // Sets up DirectX11 states for taking rendering commands from RmlUi. void BeginFrame(); - void EndFrame(); + // Sets up DirectX11 states for taking rendering commands from RmlUi. + void EndFrame(); - // -- Inherited from Rml::RenderInterface -- + // -- Inherited from Rml::RenderInterface -- - Rml::CompiledGeometryHandle CompileGeometry(Rml::Span vertices, Rml::Span indices) override; - void ReleaseGeometry(Rml::CompiledGeometryHandle geometry) override; - void RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) override; + Rml::CompiledGeometryHandle CompileGeometry(Rml::Span vertices, Rml::Span indices) override; + void ReleaseGeometry(Rml::CompiledGeometryHandle geometry) override; + void RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) override; - Rml::TextureHandle LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) override; - Rml::TextureHandle GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) override; - void ReleaseTexture(Rml::TextureHandle texture_handle) override; + Rml::TextureHandle LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) override; + Rml::TextureHandle GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) override; + void ReleaseTexture(Rml::TextureHandle texture_handle) override; - void EnableScissorRegion(bool enable) override; - void SetScissorRegion(Rml::Rectanglei region) override; + void EnableScissorRegion(bool enable) override; + void SetScissorRegion(Rml::Rectanglei region) override; - void SetViewport(const int width, const int height); - void Clear(); + void SetViewport(const int width, const int height); + void Clear(); private: - void SetBlendState(ID3D11BlendState* blendState); + void SetBlendState(ID3D11BlendState* blendState); private: // @TODO: Replace with vertex / index buffer pairs From 3dbc64b39ad4402a0123db4fbe383f11e4a8265b Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 01:10:26 +0200 Subject: [PATCH 004/125] (feat): Make BeginFrame function take a rendetargetview This allows users to render to a render target which may either be the swapchain (in most cases it will be) or a render texture which may then be used elsewhere (eg in a 3D scene). --- Backends/RmlUi_Backend_Win32_DX11.cpp | 2 +- Backends/RmlUi_Renderer_DX11.cpp | 150 +++++++++++++++++--------- Backends/RmlUi_Renderer_DX11.h | 65 +++++------ 3 files changed, 131 insertions(+), 86 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index bdcc885df..a944a2248 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -262,7 +262,7 @@ void Backend::BeginFrame() { RMLUI_ASSERT(data); // @TODO: Check if should resize, if so, re-create swapchain - data->render_interface.BeginFrame(); + data->render_interface.BeginFrame(data->d3d_resources.pMainRenderTargetView); } void Backend::PresentFrame() diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 85f4f9a3e..0dd68da05 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -37,70 +37,114 @@ #include "RmlUi_Include_Windows.h" -RenderInterface_DX11::RenderInterface_DX11() { +RenderInterface_DX11::RenderInterface_DX11() {} -} - -void RenderInterface_DX11::Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain, - ID3D11RenderTargetView* mainRenderTargetView) +void RenderInterface_DX11::Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain) { - RMLUI_ASSERTMSG(pd3dDevice, "pd3dDevice cannot be nullptr!"); - RMLUI_ASSERTMSG(pSwapChain, "pSwapChain cannot be nullptr!"); - RMLUI_ASSERTMSG(pd3dDeviceContext, "pd3dDeviceContext cannot be nullptr!"); - RMLUI_ASSERTMSG(mainRenderTargetView, "mainRenderTargetView cannot be nullptr!"); - - // Assign D3D resources - m_d3dDevice = pd3dDevice; - m_d3dContext = pd3dDeviceContext; - m_swapChain = pSwapChain; - m_mainRenderTargetView = mainRenderTargetView; - - // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. - // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). - if (!m_blendState) - { - D3D11_BLEND_DESC blendDesc; - ZeroMemory(&blendDesc, sizeof(blendDesc)); - blendDesc.AlphaToCoverageEnable = FALSE; - blendDesc.IndependentBlendEnable = FALSE; - blendDesc.RenderTarget[0].BlendEnable = TRUE; - blendDesc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE; - blendDesc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA; - blendDesc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; - blendDesc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ONE; - blendDesc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA; - blendDesc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; - blendDesc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; - HRESULT result = m_d3dDevice->CreateBlendState(&blendDesc, &m_blendState); - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device1::CreateBlendState (%d)", result); - } - } + RMLUI_ASSERTMSG(pd3dDevice, "pd3dDevice cannot be nullptr!"); + RMLUI_ASSERTMSG(pSwapChain, "pSwapChain cannot be nullptr!"); + RMLUI_ASSERTMSG(pd3dDeviceContext, "pd3dDeviceContext cannot be nullptr!"); + + // Assign D3D resources + m_d3dDevice = pd3dDevice; + m_d3dContext = pd3dDeviceContext; + m_swapChain = pSwapChain; + + // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. + // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). + if (!m_blendState) + { + D3D11_BLEND_DESC blendDesc; + ZeroMemory(&blendDesc, sizeof(blendDesc)); + blendDesc.AlphaToCoverageEnable = FALSE; + blendDesc.IndependentBlendEnable = FALSE; + blendDesc.RenderTarget[0].BlendEnable = TRUE; + blendDesc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE; + blendDesc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA; + blendDesc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; + blendDesc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ONE; + blendDesc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA; + blendDesc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; + blendDesc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; + HRESULT result = m_d3dDevice->CreateBlendState(&blendDesc, &m_blendState); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); + return; + } + } + + // Scissor regions require a rasterizer state. Cache one for scissor on and off + { + D3D11_RASTERIZER_DESC rasterizerDesc; + rasterizerDesc.FillMode = D3D11_FILL_SOLID; + rasterizerDesc.CullMode = D3D11_CULL_BACK; + rasterizerDesc.FrontCounterClockwise = false; + rasterizerDesc.DepthBias = 0; + rasterizerDesc.SlopeScaledDepthBias = 0.0f; + rasterizerDesc.DepthBiasClamp = 0.0f; + rasterizerDesc.DepthClipEnable = true; + rasterizerDesc.ScissorEnable = true; + rasterizerDesc.MultisampleEnable = MSAA_SAMPLES > 1; + rasterizerDesc.AntialiasedLineEnable = MSAA_SAMPLES > 1; + + HRESULT result = m_d3dDevice->CreateRasterizerState(&rasterizerDesc, &m_rasterizerState_scissorEnabled); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRasterizerState (scissor: enabled) (%d)", result); + return; + } + + rasterizerDesc.ScissorEnable = false; + + HRESULT result = m_d3dDevice->CreateRasterizerState(&rasterizerDesc, &m_rasterizerState_scissorDisabled); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRasterizerState (scissor: disabled) (%d)", result); + return; + } + } + + // Load shaders + + // @TODO: + + // Create vertex layout. This will be constant to avoid copying to an intermediate struct. + + // @TODO: + + // Create constant buffers. This is so that we can bind uniforms such as translation and color to the shaders. + + // @TODO: } void RenderInterface_DX11::Cleanup() { m_blendState->Release(); } -void RenderInterface_DX11::BeginFrame() { - RMLUI_ASSERTMSG(m_d3dContext, "d3dContext cannot be nullptr!"); - RMLUI_ASSERTMSG(m_d3dDevice, "d3dDevice cannot be nullptr!"); - - D3D11_VIEWPORT d3dviewport; - d3dviewport.TopLeftX = 0; - d3dviewport.TopLeftY = 0; - d3dviewport.Width = m_width; - d3dviewport.Height = m_height; - d3dviewport.MinDepth = 0.0f; - d3dviewport.MaxDepth = 1.0f; - m_d3dContext->RSSetViewports(1, &d3dviewport); - Clear(); - SetBlendState(m_blendState); +void RenderInterface_DX11::BeginFrame(ID3D11RenderTargetView* renderTargetView) { + RMLUI_ASSERTMSG(renderTargetView, "renderTargetView cannot be nullptr!"); + RMLUI_ASSERTMSG(m_d3dContext, "d3dContext cannot be nullptr!"); + RMLUI_ASSERTMSG(m_d3dDevice, "d3dDevice cannot be nullptr!"); + + m_boundRenderTarget = renderTargetView; + + D3D11_VIEWPORT d3dviewport; + d3dviewport.TopLeftX = 0; + d3dviewport.TopLeftY = 0; + d3dviewport.Width = m_width; + d3dviewport.Height = m_height; + d3dviewport.MinDepth = 0.0f; + d3dviewport.MaxDepth = 1.0f; + m_d3dContext->RSSetViewports(1, &d3dviewport); + Clear(); + SetBlendState(m_blendState); } void RenderInterface_DX11::EndFrame() { - + // @TODO: Compositing + // @TODO: Layer stack + m_boundRenderTarget = nullptr; } diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 6d2d7bfba..14b1a6524 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -39,14 +39,14 @@ class RenderInterface_DX11 : public Rml::RenderInterface { public: - void Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain, ID3D11RenderTargetView* mainRenderTargetView); RenderInterface_DX11(); // Resource initialisation and cleanup + void Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain); void Cleanup (); - void BeginFrame(); // Sets up DirectX11 states for taking rendering commands from RmlUi. + void BeginFrame(ID3D11RenderTargetView* renderTargetView); void EndFrame(); // -- Inherited from Rml::RenderInterface -- @@ -69,36 +69,37 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void SetBlendState(ID3D11BlendState* blendState); private: - // @TODO: Replace with vertex / index buffer pairs - struct GeometryView { - Rml::Span vertices; - Rml::Span indices; - }; - - // D3D11 core resources - ID3D11Device* m_d3dDevice = nullptr; - ID3D11DeviceContext* m_d3dContext = nullptr; - IDXGISwapChain* m_swapChain = nullptr; - ID3D11RenderTargetView* m_mainRenderTargetView = nullptr; - - // Viewport dimensions - int m_width = 0; - int m_height = 0; - - // SDL_Renderer* renderer; - ID3D11BlendState* m_blendState = nullptr; - // SDL_Rect rect_scissor = {}; - bool scissor_region_enabled = false; - - // D3D11 state - ID3D11BlendState* m_currentBlendState = nullptr; - - struct D3D11State { - - ID3D11BlendState* m_previousBlendState = nullptr; - }; - - D3D11State m_previousD3DState; + // @TODO: Replace with vertex / index buffer pairs + struct GeometryView { + Rml::Span vertices; + Rml::Span indices; + }; + + // D3D11 core resources + ID3D11Device* m_d3dDevice = nullptr; + ID3D11DeviceContext* m_d3dContext = nullptr; + IDXGISwapChain* m_swapChain = nullptr; + ID3D11RenderTargetView* m_boundRenderTarget = nullptr; + ID3D11RasterizerState* m_rasterizerState_scissorEnabled = nullptr; + ID3D11RasterizerState* m_rasterizerState_scissorDisabled = nullptr; + + // Viewport dimensions + int m_width = 0; + int m_height = 0; + + ID3D11BlendState* m_blendState = nullptr; + D3D11_RECT m_rect_scissor = {}; + bool m_scissor_region_enabled = false; + + // D3D11 state + ID3D11BlendState* m_currentBlendState = nullptr; + + struct D3D11State { + + ID3D11BlendState* m_previousBlendState = nullptr; + }; + + D3D11State m_previousD3DState; }; #endif // RMLUI_PLATFORM_WIN32 From 89b8376edffc18f7d947eba15b3fbe53c26a80e8 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 01:11:55 +0200 Subject: [PATCH 005/125] (feat): Allow users to overload the default image loader We should provide a image loader (in this case the uncompressed TGA loader), but should also allow users using this renderer backend to overload it with their own one. E.g. a user may have libpng in their project and they want to hook libpng into the load image function to load PNGs. --- Backends/RmlUi_Renderer_DX11.cpp | 176 ++++++++++++++++++------------- Backends/RmlUi_Renderer_DX11.h | 22 ++++ 2 files changed, 122 insertions(+), 76 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 0dd68da05..9f8752b2a 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -180,82 +180,106 @@ struct TGAHeader { #pragma pack() -Rml::TextureHandle RenderInterface_DX11::LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) { - Rml::FileInterface* file_interface = Rml::GetFileInterface(); - Rml::FileHandle file_handle = file_interface->Open(source); - if (!file_handle) - { - return false; - } - - file_interface->Seek(file_handle, 0, SEEK_END); - size_t buffer_size = file_interface->Tell(file_handle); - file_interface->Seek(file_handle, 0, SEEK_SET); - - if (buffer_size <= sizeof(TGAHeader)) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "Texture file size is smaller than TGAHeader, file is not a valid TGA image."); - file_interface->Close(file_handle); - return false; - } - - using Rml::byte; - Rml::UniquePtr buffer(new byte[buffer_size]); - file_interface->Read(buffer.get(), buffer_size, file_handle); - file_interface->Close(file_handle); - - TGAHeader header; - memcpy(&header, buffer.get(), sizeof(TGAHeader)); - - int color_mode = header.bitsPerPixel / 8; - const size_t image_size = header.width * header.height * 4; // We always make 32bit textures - - if (header.dataType != 2) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "Only 24/32bit uncompressed TGAs are supported."); - return false; - } - - // Ensure we have at least 3 colors - if (color_mode < 3) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "Only 24 and 32bit textures are supported."); - return false; - } - - const byte* image_src = buffer.get() + sizeof(TGAHeader); - Rml::UniquePtr image_dest_buffer(new byte[image_size]); - byte* image_dest = image_dest_buffer.get(); - - // Targa is BGR, swap to RGB, flip Y axis, and convert to premultiplied alpha. - for (long y = 0; y < header.height; y++) - { - long read_index = y * header.width * color_mode; - long write_index = ((header.imageDescriptor & 32) != 0) ? read_index : (header.height - y - 1) * header.width * 4; - for (long x = 0; x < header.width; x++) - { - image_dest[write_index] = image_src[read_index + 2]; - image_dest[write_index + 1] = image_src[read_index + 1]; - image_dest[write_index + 2] = image_src[read_index]; - if (color_mode == 4) - { - const byte alpha = image_src[read_index + 3]; - for (size_t j = 0; j < 3; j++) - image_dest[write_index + j] = byte((image_dest[write_index + j] * alpha) / 255); - image_dest[write_index + 3] = alpha; - } - else - image_dest[write_index + 3] = 255; - - write_index += 4; - read_index += color_mode; - } - } - - texture_dimensions.x = header.width; - texture_dimensions.y = header.height; - - return GenerateTexture({image_dest, image_size}, texture_dimensions); +Rml::TextureHandle RenderInterface_DX11::LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) +{ + // Use the user provided image loading function if it's provided, else fallback to the included TGA one + if (LoadTextureFromFileRaw != nullptr && FreeTextureFromFileRaw != nullptr) + { + int texture_width = 0, texture_height = 0; + size_t image_size_bytes = 0; + uint8_t* texture_data = nullptr; + LoadTextureFromFileRaw(source, &texture_width, &texture_height, &texture_data, &image_size_bytes); + + if (texture_data != nullptr) + { + texture_dimensions.x = texture_width; + texture_dimensions.y = texture_height; + + Rml::TextureHandle handle = GenerateTexture({texture_data, image_size_bytes}, texture_dimensions); + + FreeTextureFromFileRaw(texture_data); + return handle; + } + + // Image must be invalid if the file failed to load. Return invalid handle + return Rml::TextureHandle(0); + } + + Rml::FileInterface* file_interface = Rml::GetFileInterface(); + Rml::FileHandle file_handle = file_interface->Open(source); + if (!file_handle) + { + return false; + } + + file_interface->Seek(file_handle, 0, SEEK_END); + size_t buffer_size = file_interface->Tell(file_handle); + file_interface->Seek(file_handle, 0, SEEK_SET); + + if (buffer_size <= sizeof(TGAHeader)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Texture file size is smaller than TGAHeader, file is not a valid TGA image."); + file_interface->Close(file_handle); + return false; + } + + using Rml::byte; + Rml::UniquePtr buffer(new byte[buffer_size]); + file_interface->Read(buffer.get(), buffer_size, file_handle); + file_interface->Close(file_handle); + + TGAHeader header; + memcpy(&header, buffer.get(), sizeof(TGAHeader)); + + int color_mode = header.bitsPerPixel / 8; + const size_t image_size = header.width * header.height * 4; // We always make 32bit textures + + if (header.dataType != 2) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Only 24/32bit uncompressed TGAs are supported."); + return false; + } + + // Ensure we have at least 3 colors + if (color_mode < 3) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Only 24 and 32bit textures are supported."); + return false; + } + + const byte* image_src = buffer.get() + sizeof(TGAHeader); + Rml::UniquePtr image_dest_buffer(new byte[image_size]); + byte* image_dest = image_dest_buffer.get(); + + // Targa is BGR, swap to RGB, flip Y axis, and convert to premultiplied alpha. + for (long y = 0; y < header.height; y++) + { + long read_index = y * header.width * color_mode; + long write_index = ((header.imageDescriptor & 32) != 0) ? read_index : (header.height - y - 1) * header.width * 4; + for (long x = 0; x < header.width; x++) + { + image_dest[write_index] = image_src[read_index + 2]; + image_dest[write_index + 1] = image_src[read_index + 1]; + image_dest[write_index + 2] = image_src[read_index]; + if (color_mode == 4) + { + const byte alpha = image_src[read_index + 3]; + for (size_t j = 0; j < 3; j++) + image_dest[write_index + j] = byte((image_dest[write_index + j] * alpha) / 255); + image_dest[write_index + 3] = alpha; + } + else + image_dest[write_index + 3] = 255; + + write_index += 4; + read_index += color_mode; + } + } + + texture_dimensions.x = header.width; + texture_dimensions.y = header.height; + + return GenerateTexture({image_dest, image_size}, texture_dimensions); } Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) { diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 14b1a6524..675649ea2 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -37,6 +37,16 @@ #include #include +// Allow the user to override the number of MSAA samples +#ifndef MSAA_SAMPLES +#define MSAA_SAMPLES 2 +#endif + +// Use these typedefs to overload the default implementation with another image loader so that you can handle more than uncompressed TGA if you wish. +// Must define both the load and free functions. See RenderInterface_DX11:: +typedef void (*pfnLoadTextureRaw)(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize); +typedef void (*pfnFreeTextureRaw)(uint8_t* pData); + class RenderInterface_DX11 : public Rml::RenderInterface { public: RenderInterface_DX11(); @@ -65,6 +75,18 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void SetViewport(const int width, const int height); void Clear(); +public: + /// Called by the renderer when it wants to load a texture from disk. + /// @param[in] pFilename A Rml string containing the file name. + /// @param[out] pWidth The width of the texture read from disk. + /// @param[out] pHeight The height of the texture read from disk. + /// @param[out] pData A pointer to an RGBA byte array of texture data, which will then be used to generate a texture. Set to NULL to indicate that the file failed to load for whatever reason. + /// @param[out] pDataSize A pointer to a size_t storing how many bytes is in pData. + pfnLoadTextureRaw LoadTextureFromFileRaw = nullptr; + /// Called by the renderer when it wants to free a texture from disk. Always called after a successful load from LoadTextureFromFileRaw. + /// @param[in] pData A pointer to an RGBA byte array of texture data to free. + pfnFreeTextureRaw FreeTextureFromFileRaw = nullptr; + private: void SetBlendState(ID3D11BlendState* blendState); From 91e517f7caedd0caaefce92bd2d9df464e0515dd Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 01:12:20 +0200 Subject: [PATCH 006/125] (feat): Scissor region --- Backends/RmlUi_Renderer_DX11.cpp | 32 ++++++++++++++++++++++++++++---- 1 file changed, 28 insertions(+), 4 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 9f8752b2a..08b6c2cc5 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -282,19 +282,43 @@ Rml::TextureHandle RenderInterface_DX11::LoadTexture(Rml::Vector2i& texture_dime return GenerateTexture({image_dest, image_size}, texture_dimensions); } -Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) { - return Rml::TextureHandle(0); +Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) +{ + return Rml::TextureHandle(0); } -void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) { +void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) +{ } void RenderInterface_DX11::EnableScissorRegion(bool enable) { + if (enable != m_scissor_region_enabled) + { + if (enable) + { + m_d3dContext->RSSetState(m_rasterizerState_scissorEnabled); + } + else + { + m_d3dContext->RSSetState(m_rasterizerState_scissorDisabled); + } + m_scissor_region_enabled = enable; + } } -void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) { +void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) +{ + m_rect_scissor.left = region.Left(); + m_rect_scissor.top = region.Top(); + m_rect_scissor.bottom = region.Bottom(); + m_rect_scissor.right = region.Right(); + + if (m_scissor_region_enabled) + { + m_d3dContext->RSSetScissorRects(1, &m_rect_scissor); + } } void RenderInterface_DX11::SetViewport(const int width, const int height) From c17d7b2707f850d8cf3f69ec5b015d6df4b6a15a Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 01:13:45 +0200 Subject: [PATCH 007/125] (fix): Incorrect arguments into Init --- Backends/RmlUi_Backend_Win32_DX11.cpp | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index a944a2248..52704bc55 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -159,8 +159,7 @@ bool Backend::Initialize(const char* window_name, int width, int height, bool al data->render_interface.Init( data->d3d_resources.pd3dDevice, data->d3d_resources.pd3dDeviceContext, - data->d3d_resources.pSwapChain, - data->d3d_resources.pMainRenderTargetView); + data->d3d_resources.pSwapChain); data->window_handle = window_handle; data->system_interface.SetWindow(window_handle); From 651c3ba5c023b042fe0f0dff39b769e9ac3dcc81 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 01:14:01 +0200 Subject: [PATCH 008/125] (fix): Fix double declaration --- Backends/RmlUi_Renderer_DX11.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 08b6c2cc5..81171997a 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -97,7 +97,7 @@ void RenderInterface_DX11::Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* p rasterizerDesc.ScissorEnable = false; - HRESULT result = m_d3dDevice->CreateRasterizerState(&rasterizerDesc, &m_rasterizerState_scissorDisabled); + result = m_d3dDevice->CreateRasterizerState(&rasterizerDesc, &m_rasterizerState_scissorDisabled); if (FAILED(result)) { Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRasterizerState (scissor: disabled) (%d)", result); From 5b744c0089d84041c5d2f033a12dd5a916db5403 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:19:50 +0200 Subject: [PATCH 009/125] (chore): Also require swapchain on begin frame --- Backends/RmlUi_Backend_Win32_DX11.cpp | 7 ++----- Backends/RmlUi_Renderer_DX11.cpp | 3 ++- Backends/RmlUi_Renderer_DX11.h | 4 ++-- 3 files changed, 6 insertions(+), 8 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 52704bc55..3d643671f 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -61,19 +61,16 @@ static ProcAdjustWindowRectExForDpi procAdjustWindowRectExForDpi = NULL; // Make ourselves DPI aware on supported Windows versions. static void InitializeDpiSupport() { - // Cast function pointers to void* first for MinGW not to emit errors. - procSetProcessDpiAwarenessContext = - (ProcSetProcessDpiAwarenessContext)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "SetProcessDpiAwarenessContext"); - procGetDpiForWindow = (ProcGetDpiForWindow)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "GetDpiForWindow"); procAdjustWindowRectExForDpi = (ProcAdjustWindowRectExForDpi)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "AdjustWindowRectExForDpi"); if (!has_dpi_support && procSetProcessDpiAwarenessContext != NULL && procGetDpiForWindow != NULL && procAdjustWindowRectExForDpi != NULL) { - // Activate Per Monitor V2. if (procSetProcessDpiAwarenessContext(DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2)) has_dpi_support = true; } + procAdjustWindowRectExForDpi = + (ProcAdjustWindowRectExForDpi)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "AdjustWindowRectExForDpi"); } static UINT GetWindowDpi(HWND window_handle) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 81171997a..93aa27b37 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -39,11 +39,11 @@ RenderInterface_DX11::RenderInterface_DX11() {} -void RenderInterface_DX11::Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain) { RMLUI_ASSERTMSG(pd3dDevice, "pd3dDevice cannot be nullptr!"); RMLUI_ASSERTMSG(pSwapChain, "pSwapChain cannot be nullptr!"); RMLUI_ASSERTMSG(pd3dDeviceContext, "pd3dDeviceContext cannot be nullptr!"); +void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context) // Assign D3D resources m_d3dDevice = pd3dDevice; @@ -126,6 +126,7 @@ void RenderInterface_DX11::BeginFrame(ID3D11RenderTargetView* renderTargetView) RMLUI_ASSERTMSG(renderTargetView, "renderTargetView cannot be nullptr!"); RMLUI_ASSERTMSG(m_d3dContext, "d3dContext cannot be nullptr!"); RMLUI_ASSERTMSG(m_d3dDevice, "d3dDevice cannot be nullptr!"); +void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) m_boundRenderTarget = renderTargetView; diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 675649ea2..889026f6d 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -52,11 +52,11 @@ class RenderInterface_DX11 : public Rml::RenderInterface { RenderInterface_DX11(); // Resource initialisation and cleanup - void Init(ID3D11Device* pd3dDevice, ID3D11DeviceContext* pd3dDeviceContext, IDXGISwapChain* pSwapChain); + void Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context); void Cleanup (); // Sets up DirectX11 states for taking rendering commands from RmlUi. - void BeginFrame(ID3D11RenderTargetView* renderTargetView); + void BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view); void EndFrame(); // -- Inherited from Rml::RenderInterface -- From ccb940c29483d24804fd0c1ed172260c5d5b88d9 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:21:29 +0200 Subject: [PATCH 010/125] (chore): Use spaces not tabs --- Backends/RmlUi_Backend_Win32_DX11.cpp | 566 +++++++++++++------------- 1 file changed, 283 insertions(+), 283 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 3d643671f..03036e84a 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -61,32 +61,35 @@ static ProcAdjustWindowRectExForDpi procAdjustWindowRectExForDpi = NULL; // Make ourselves DPI aware on supported Windows versions. static void InitializeDpiSupport() { - procAdjustWindowRectExForDpi = - (ProcAdjustWindowRectExForDpi)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "AdjustWindowRectExForDpi"); - - if (!has_dpi_support && procSetProcessDpiAwarenessContext != NULL && procGetDpiForWindow != NULL && procAdjustWindowRectExForDpi != NULL) - { - if (procSetProcessDpiAwarenessContext(DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2)) - has_dpi_support = true; - } + // Cast function pointers to void* first for MinGW not to emit errors. + procSetProcessDpiAwarenessContext = + (ProcSetProcessDpiAwarenessContext)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "SetProcessDpiAwarenessContext"); + procGetDpiForWindow = (ProcGetDpiForWindow)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "GetDpiForWindow"); procAdjustWindowRectExForDpi = (ProcAdjustWindowRectExForDpi)(void*)GetProcAddress(GetModuleHandle(TEXT("User32.dll")), "AdjustWindowRectExForDpi"); + + if (!has_dpi_support && procSetProcessDpiAwarenessContext != NULL && procGetDpiForWindow != NULL && procAdjustWindowRectExForDpi != NULL) + { + // Activate Per Monitor V2. + if (procSetProcessDpiAwarenessContext(DPI_AWARENESS_CONTEXT_PER_MONITOR_AWARE_V2)) + has_dpi_support = true; + } } static UINT GetWindowDpi(HWND window_handle) { - if (has_dpi_support) - { - UINT dpi = procGetDpiForWindow(window_handle); - if (dpi != 0) - return dpi; - } - return USER_DEFAULT_SCREEN_DPI; + if (has_dpi_support) + { + UINT dpi = procGetDpiForWindow(window_handle); + if (dpi != 0) + return dpi; + } + return USER_DEFAULT_SCREEN_DPI; } static float GetDensityIndependentPixelRatio(HWND window_handle) { - return float(GetWindowDpi(window_handle)) / float(USER_DEFAULT_SCREEN_DPI); + return float(GetWindowDpi(window_handle)) / float(USER_DEFAULT_SCREEN_DPI); } // Create the window but don't show it yet. Returns the pixel size of the window, which may be different than the passed size due to DPI settings. @@ -104,314 +107,311 @@ Global data used by this backend. Lifetime governed by the calls to Backend::Initialize() and Backend::Shutdown(). */ struct BackendData { - SystemInterface_Win32 system_interface; - RenderInterface_DX11 render_interface; - TextInputMethodEditor_Win32 text_input_method_editor; - - HINSTANCE instance_handle = nullptr; - std::wstring instance_name; - HWND window_handle = nullptr; - - // D3D11 resources - struct { - ID3D11Device* pd3dDevice = nullptr; - ID3D11DeviceContext* pd3dDeviceContext = nullptr; - IDXGISwapChain* pSwapChain = nullptr; - bool swapchain_cccluded = false; - ID3D11RenderTargetView* pMainRenderTargetView = nullptr; - } d3d_resources; - - bool context_dimensions_dirty = true; - Rml::Vector2i window_dimensions; - bool running = true; - - // Arguments set during event processing and nulled otherwise. - Rml::Context* context = nullptr; - KeyDownCallback key_down_callback = nullptr; + SystemInterface_Win32 system_interface; + RenderInterface_DX11 render_interface; + TextInputMethodEditor_Win32 text_input_method_editor; + + HINSTANCE instance_handle = nullptr; + std::wstring instance_name; + HWND window_handle = nullptr; + + // D3D11 resources + struct { + ID3D11Device* pd3dDevice = nullptr; + ID3D11DeviceContext* pd3dDeviceContext = nullptr; + IDXGISwapChain* pSwapChain = nullptr; + bool swapchain_cccluded = false; + ID3D11RenderTargetView* pMainRenderTargetView = nullptr; + } d3d_resources; + + bool context_dimensions_dirty = true; + Rml::Vector2i window_dimensions; + bool running = true; + + // Arguments set during event processing and nulled otherwise. + Rml::Context* context = nullptr; + KeyDownCallback key_down_callback = nullptr; }; static Rml::UniquePtr data; bool Backend::Initialize(const char* window_name, int width, int height, bool allow_resize) { - RMLUI_ASSERT(!data); + RMLUI_ASSERT(!data); - const std::wstring name = RmlWin32::ConvertToUTF16(Rml::String(window_name)); + const std::wstring name = RmlWin32::ConvertToUTF16(Rml::String(window_name)); - data = Rml::MakeUnique(); - data->instance_handle = GetModuleHandle(nullptr); - data->instance_name = name.data(); + data = Rml::MakeUnique(); + data->instance_handle = GetModuleHandle(nullptr); + data->instance_name = name.data(); - InitializeDpiSupport(); + InitializeDpiSupport(); - // Initialize the window but don't show it yet. - HWND window_handle = InitializeWindow(data->instance_handle, name, width, height, allow_resize); - if (!window_handle) - return false; + // Initialize the window but don't show it yet. + HWND window_handle = InitializeWindow(data->instance_handle, name, width, height, allow_resize); + if (!window_handle) + return false; - if (!CreateDeviceD3D(window_handle)) { - ::CloseWindow(window_handle); - return false; - } + if (!CreateDeviceD3D(window_handle)) { + ::CloseWindow(window_handle); + return false; + } - data->render_interface.Init( - data->d3d_resources.pd3dDevice, - data->d3d_resources.pd3dDeviceContext, - data->d3d_resources.pSwapChain); + data->render_interface.Init(data->d3d_resources.pd3dDevice, data->d3d_resources.pd3dDeviceContext); - data->window_handle = window_handle; - data->system_interface.SetWindow(window_handle); + data->window_handle = window_handle; + data->system_interface.SetWindow(window_handle); - // Now we are ready to show the window. - ::ShowWindow(window_handle, SW_SHOW); - ::SetForegroundWindow(window_handle); - ::SetFocus(window_handle); + // Now we are ready to show the window. + ::ShowWindow(window_handle, SW_SHOW); + ::SetForegroundWindow(window_handle); + ::SetFocus(window_handle); - // Provide a backend-specific text input handler to manage the IME. - Rml::SetTextInputHandler(&data->text_input_method_editor); + // Provide a backend-specific text input handler to manage the IME. + Rml::SetTextInputHandler(&data->text_input_method_editor); - return true; + return true; } void Backend::Shutdown() { - RMLUI_ASSERT(data); + RMLUI_ASSERT(data); - // As we forcefully override the global text input handler, we must reset it before the data is destroyed to avoid any potential use-after-free. - if (Rml::GetTextInputHandler() == &data->text_input_method_editor) - Rml::SetTextInputHandler(nullptr); + // As we forcefully override the global text input handler, we must reset it before the data is destroyed to avoid any potential use-after-free. + if (Rml::GetTextInputHandler() == &data->text_input_method_editor) + Rml::SetTextInputHandler(nullptr); CleanupDeviceD3D(); - ::DestroyWindow(data->window_handle); - ::UnregisterClassW((LPCWSTR)data->instance_name.data(), data->instance_handle); + ::DestroyWindow(data->window_handle); + ::UnregisterClassW((LPCWSTR)data->instance_name.data(), data->instance_handle); - data.reset(); + data.reset(); } Rml::SystemInterface* Backend::GetSystemInterface() { - RMLUI_ASSERT(data); - return &data->system_interface; + RMLUI_ASSERT(data); + return &data->system_interface; } Rml::RenderInterface* Backend::GetRenderInterface() { - RMLUI_ASSERT(data); - return &data->render_interface; + RMLUI_ASSERT(data); + return &data->render_interface; } static bool NextEvent(MSG& message, UINT timeout) { - if (timeout != 0) - { - UINT_PTR timer_id = SetTimer(NULL, 0, timeout, NULL); - BOOL res = GetMessage(&message, NULL, 0, 0); - KillTimer(NULL, timer_id); - if (message.message != WM_TIMER || message.hwnd != nullptr || message.wParam != timer_id) - return res; - } - return PeekMessage(&message, nullptr, 0, 0, PM_REMOVE); + if (timeout != 0) + { + UINT_PTR timer_id = SetTimer(NULL, 0, timeout, NULL); + BOOL res = GetMessage(&message, NULL, 0, 0); + KillTimer(NULL, timer_id); + if (message.message != WM_TIMER || message.hwnd != nullptr || message.wParam != timer_id) + return res; + } + return PeekMessage(&message, nullptr, 0, 0, PM_REMOVE); } bool Backend::ProcessEvents(Rml::Context* context, KeyDownCallback key_down_callback, bool power_save) { - RMLUI_ASSERT(data && context); - - // The initial window size may have been affected by system DPI settings, apply the actual pixel size and dp-ratio to the context. - if (data->context_dimensions_dirty) - { - data->context_dimensions_dirty = false; - const float dp_ratio = GetDensityIndependentPixelRatio(data->window_handle); - context->SetDimensions(data->window_dimensions); - context->SetDensityIndependentPixelRatio(dp_ratio); - } - - data->context = context; - data->key_down_callback = key_down_callback; - - MSG message; - bool has_message = NextEvent(message, power_save ? static_cast(Rml::Math::Min(context->GetNextUpdateDelay(), 10.0) * 1000.0) : 0); - while (has_message) - { - // Dispatch the message to our local event handler below. - TranslateMessage(&message); - DispatchMessage(&message); - - has_message = NextEvent(message, 0); - } - - data->context = nullptr; - data->key_down_callback = nullptr; - - const bool result = data->running; - data->running = true; - return result; + RMLUI_ASSERT(data && context); + + // The initial window size may have been affected by system DPI settings, apply the actual pixel size and dp-ratio to the context. + if (data->context_dimensions_dirty) + { + data->context_dimensions_dirty = false; + const float dp_ratio = GetDensityIndependentPixelRatio(data->window_handle); + context->SetDimensions(data->window_dimensions); + context->SetDensityIndependentPixelRatio(dp_ratio); + } + + data->context = context; + data->key_down_callback = key_down_callback; + + MSG message; + bool has_message = NextEvent(message, power_save ? static_cast(Rml::Math::Min(context->GetNextUpdateDelay(), 10.0) * 1000.0) : 0); + while (has_message) + { + // Dispatch the message to our local event handler below. + TranslateMessage(&message); + DispatchMessage(&message); + + has_message = NextEvent(message, 0); + } + + data->context = nullptr; + data->key_down_callback = nullptr; + + const bool result = data->running; + data->running = true; + return result; } void Backend::RequestExit() { - RMLUI_ASSERT(data); - data->running = false; + RMLUI_ASSERT(data); + data->running = false; } void Backend::BeginFrame() { - RMLUI_ASSERT(data); - // @TODO: Check if should resize, if so, re-create swapchain - data->render_interface.BeginFrame(data->d3d_resources.pMainRenderTargetView); + RMLUI_ASSERT(data); + // @TODO: Check if should resize, if so, re-create swapchain + data->render_interface.BeginFrame(data->d3d_resources.pSwapChain, data->d3d_resources.pMainRenderTargetView); } void Backend::PresentFrame() { - RMLUI_ASSERT(data); - data->render_interface.EndFrame(); + RMLUI_ASSERT(data); + data->render_interface.EndFrame(); - // Present - HRESULT hr = data->d3d_resources.pSwapChain->Present(1, 0); // Present with vsync - //HRESULT hr = g_pSwapChain->Present(0, 0); // Present without vsync - data->d3d_resources.swapchain_cccluded = (hr == DXGI_STATUS_OCCLUDED); + // Present + HRESULT hr = data->d3d_resources.pSwapChain->Present(1, 0); // Present with vsync + //HRESULT hr = g_pSwapChain->Present(0, 0); // Present without vsync + data->d3d_resources.swapchain_cccluded = (hr == DXGI_STATUS_OCCLUDED); - // Optional, used to mark frames during performance profiling. - RMLUI_FrameMark; + // Optional, used to mark frames during performance profiling. + RMLUI_FrameMark; } // Local event handler for window and input events. static LRESULT CALLBACK WindowProcedureHandler(HWND window_handle, UINT message, WPARAM w_param, LPARAM l_param) { - RMLUI_ASSERT(data); - - switch (message) - { - case WM_CLOSE: - { - data->running = false; - return 0; - } - break; - case WM_SIZE: - { - const int width = LOWORD(l_param); - const int height = HIWORD(l_param); - data->window_dimensions.x = width; - data->window_dimensions.y = height; - data->render_interface.SetViewport(width, height); - if (data->context) - data->context->SetDimensions(data->window_dimensions); - return 0; - } - break; - case WM_DPICHANGED: - { - RECT* new_pos = (RECT*)l_param; - SetWindowPos(window_handle, NULL, new_pos->left, new_pos->top, new_pos->right - new_pos->left, new_pos->bottom - new_pos->top, - SWP_NOZORDER | SWP_NOACTIVATE); - if (data->context && has_dpi_support) - data->context->SetDensityIndependentPixelRatio(GetDensityIndependentPixelRatio(window_handle)); - return 0; - } - break; - case WM_KEYDOWN: - { - // Override the default key event callback to add global shortcuts for the samples. - Rml::Context* context = data->context; - KeyDownCallback key_down_callback = data->key_down_callback; - - const Rml::Input::KeyIdentifier rml_key = RmlWin32::ConvertKey((int)w_param); - const int rml_modifier = RmlWin32::GetKeyModifierState(); - const float native_dp_ratio = GetDensityIndependentPixelRatio(window_handle); - - // See if we have any global shortcuts that take priority over the context. - if (key_down_callback && !key_down_callback(context, rml_key, rml_modifier, native_dp_ratio, true)) - return 0; - // Otherwise, hand the event over to the context by calling the input handler as normal. - if (!RmlWin32::WindowProcedure(context, data->text_input_method_editor, window_handle, message, w_param, l_param)) - return 0; - // The key was not consumed by the context either, try keyboard shortcuts of lower priority. - if (key_down_callback && !key_down_callback(context, rml_key, rml_modifier, native_dp_ratio, false)) - return 0; - return 0; - } - break; - default: - { - // Submit it to the platform handler for default input handling. - if (!RmlWin32::WindowProcedure(data->context, data->text_input_method_editor, window_handle, message, w_param, l_param)) - return 0; - } - break; - } - - // All unhandled messages go to DefWindowProc. - return DefWindowProc(window_handle, message, w_param, l_param); + RMLUI_ASSERT(data); + + switch (message) + { + case WM_CLOSE: + { + data->running = false; + return 0; + } + break; + case WM_SIZE: + { + const int width = LOWORD(l_param); + const int height = HIWORD(l_param); + data->window_dimensions.x = width; + data->window_dimensions.y = height; + data->render_interface.SetViewport(width, height); + if (data->context) + data->context->SetDimensions(data->window_dimensions); + return 0; + } + break; + case WM_DPICHANGED: + { + RECT* new_pos = (RECT*)l_param; + SetWindowPos(window_handle, NULL, new_pos->left, new_pos->top, new_pos->right - new_pos->left, new_pos->bottom - new_pos->top, + SWP_NOZORDER | SWP_NOACTIVATE); + if (data->context && has_dpi_support) + data->context->SetDensityIndependentPixelRatio(GetDensityIndependentPixelRatio(window_handle)); + return 0; + } + break; + case WM_KEYDOWN: + { + // Override the default key event callback to add global shortcuts for the samples. + Rml::Context* context = data->context; + KeyDownCallback key_down_callback = data->key_down_callback; + + const Rml::Input::KeyIdentifier rml_key = RmlWin32::ConvertKey((int)w_param); + const int rml_modifier = RmlWin32::GetKeyModifierState(); + const float native_dp_ratio = GetDensityIndependentPixelRatio(window_handle); + + // See if we have any global shortcuts that take priority over the context. + if (key_down_callback && !key_down_callback(context, rml_key, rml_modifier, native_dp_ratio, true)) + return 0; + // Otherwise, hand the event over to the context by calling the input handler as normal. + if (!RmlWin32::WindowProcedure(context, data->text_input_method_editor, window_handle, message, w_param, l_param)) + return 0; + // The key was not consumed by the context either, try keyboard shortcuts of lower priority. + if (key_down_callback && !key_down_callback(context, rml_key, rml_modifier, native_dp_ratio, false)) + return 0; + return 0; + } + break; + default: + { + // Submit it to the platform handler for default input handling. + if (!RmlWin32::WindowProcedure(data->context, data->text_input_method_editor, window_handle, message, w_param, l_param)) + return 0; + } + break; + } + + // All unhandled messages go to DefWindowProc. + return DefWindowProc(window_handle, message, w_param, l_param); } static HWND InitializeWindow(HINSTANCE instance_handle, const std::wstring& name, int& inout_width, int& inout_height, bool allow_resize) { - // Fill out the window class struct. - WNDCLASSW window_class; + // Fill out the window class struct. + WNDCLASSW window_class; ZeroMemory(&window_class, sizeof(window_class)); - window_class.style = CS_HREDRAW | CS_VREDRAW | CS_OWNDC; - window_class.lpfnWndProc = &WindowProcedureHandler; // Attach our local event handler. - window_class.cbClsExtra = 0; - window_class.cbWndExtra = 0; - window_class.hInstance = instance_handle; - window_class.hIcon = LoadIcon(nullptr, IDI_WINLOGO); - window_class.hCursor = LoadCursor(nullptr, IDC_ARROW); - window_class.hbrBackground = nullptr; - window_class.lpszMenuName = nullptr; - window_class.lpszClassName = name.data(); - - if (!RegisterClassW(&window_class)) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "Failed to register window class"); - return nullptr; - } - - HWND window_handle = CreateWindowExW(WS_EX_APPWINDOW | WS_EX_WINDOWEDGE, - name.data(), // Window class name. - name.data(), WS_CLIPSIBLINGS | WS_CLIPCHILDREN | WS_OVERLAPPEDWINDOW, 0, 0, // Window position. - 0, 0, // Window size. - nullptr, nullptr, instance_handle, nullptr); - - if (!window_handle) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "Failed to create window"); - return nullptr; - } - - UINT window_dpi = GetWindowDpi(window_handle); - inout_width = (inout_width * (int)window_dpi) / USER_DEFAULT_SCREEN_DPI; - inout_height = (inout_height * (int)window_dpi) / USER_DEFAULT_SCREEN_DPI; - - DWORD style = (allow_resize ? WS_OVERLAPPEDWINDOW : (WS_OVERLAPPEDWINDOW & ~WS_SIZEBOX & ~WS_MAXIMIZEBOX)); - DWORD extended_style = WS_EX_APPWINDOW | WS_EX_WINDOWEDGE; - - // Adjust the window size to take the edges into account. - RECT window_rect; - window_rect.top = 0; - window_rect.left = 0; - window_rect.right = inout_width; - window_rect.bottom = inout_height; - if (has_dpi_support) - procAdjustWindowRectExForDpi(&window_rect, style, FALSE, extended_style, window_dpi); - else - AdjustWindowRectEx(&window_rect, style, FALSE, extended_style); - - SetWindowLong(window_handle, GWL_EXSTYLE, extended_style); - SetWindowLong(window_handle, GWL_STYLE, style); - - // Resize the window and center it on the screen. - Rml::Vector2i screen_size = {GetSystemMetrics(SM_CXSCREEN), GetSystemMetrics(SM_CYSCREEN)}; - Rml::Vector2i window_size = {int(window_rect.right - window_rect.left), int(window_rect.bottom - window_rect.top)}; - Rml::Vector2i window_pos = Rml::Math::Max((screen_size - window_size) / 2, Rml::Vector2i(0)); - - SetWindowPos(window_handle, HWND_TOP, window_pos.x, window_pos.y, window_size.x, window_size.y, SWP_NOACTIVATE); - - return window_handle; + window_class.style = CS_HREDRAW | CS_VREDRAW | CS_OWNDC; + window_class.lpfnWndProc = &WindowProcedureHandler; // Attach our local event handler. + window_class.cbClsExtra = 0; + window_class.cbWndExtra = 0; + window_class.hInstance = instance_handle; + window_class.hIcon = LoadIcon(nullptr, IDI_WINLOGO); + window_class.hCursor = LoadCursor(nullptr, IDC_ARROW); + window_class.hbrBackground = nullptr; + window_class.lpszMenuName = nullptr; + window_class.lpszClassName = name.data(); + + if (!RegisterClassW(&window_class)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Failed to register window class"); + return nullptr; + } + + HWND window_handle = CreateWindowExW(WS_EX_APPWINDOW | WS_EX_WINDOWEDGE, + name.data(), // Window class name. + name.data(), WS_CLIPSIBLINGS | WS_CLIPCHILDREN | WS_OVERLAPPEDWINDOW, 0, 0, // Window position. + 0, 0, // Window size. + nullptr, nullptr, instance_handle, nullptr); + + if (!window_handle) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "Failed to create window"); + return nullptr; + } + + UINT window_dpi = GetWindowDpi(window_handle); + inout_width = (inout_width * (int)window_dpi) / USER_DEFAULT_SCREEN_DPI; + inout_height = (inout_height * (int)window_dpi) / USER_DEFAULT_SCREEN_DPI; + + DWORD style = (allow_resize ? WS_OVERLAPPEDWINDOW : (WS_OVERLAPPEDWINDOW & ~WS_SIZEBOX & ~WS_MAXIMIZEBOX)); + DWORD extended_style = WS_EX_APPWINDOW | WS_EX_WINDOWEDGE; + + // Adjust the window size to take the edges into account. + RECT window_rect; + window_rect.top = 0; + window_rect.left = 0; + window_rect.right = inout_width; + window_rect.bottom = inout_height; + if (has_dpi_support) + procAdjustWindowRectExForDpi(&window_rect, style, FALSE, extended_style, window_dpi); + else + AdjustWindowRectEx(&window_rect, style, FALSE, extended_style); + + SetWindowLong(window_handle, GWL_EXSTYLE, extended_style); + SetWindowLong(window_handle, GWL_STYLE, style); + + // Resize the window and center it on the screen. + Rml::Vector2i screen_size = {GetSystemMetrics(SM_CXSCREEN), GetSystemMetrics(SM_CYSCREEN)}; + Rml::Vector2i window_size = {int(window_rect.right - window_rect.left), int(window_rect.bottom - window_rect.top)}; + Rml::Vector2i window_pos = Rml::Math::Max((screen_size - window_size) / 2, Rml::Vector2i(0)); + + SetWindowPos(window_handle, HWND_TOP, window_pos.x, window_pos.y, window_size.x, window_size.y, SWP_NOACTIVATE); + + return window_handle; } static bool CreateDeviceD3D(HWND hwnd) { - // Setup swap chain + // Setup swap chain DXGI_SWAP_CHAIN_DESC sd; ZeroMemory(&sd, sizeof(sd)); sd.BufferCount = 2; @@ -432,11 +432,11 @@ static bool CreateDeviceD3D(HWND hwnd) { //createDeviceFlags |= D3D11_CREATE_DEVICE_DEBUG; D3D_FEATURE_LEVEL featureLevel; const D3D_FEATURE_LEVEL featureLevelArray[2] = { D3D_FEATURE_LEVEL_11_0, D3D_FEATURE_LEVEL_10_0, }; - HRESULT res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_HARDWARE, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, - &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); + HRESULT res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_HARDWARE, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, + &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); if (res == DXGI_ERROR_UNSUPPORTED) // Try high-performance WARP software driver if hardware is not available. - res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_WARP, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, - &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); + res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_WARP, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, + &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); if (res != S_OK) return false; @@ -447,36 +447,36 @@ static bool CreateDeviceD3D(HWND hwnd) { static void CleanupDeviceD3D() { CleanupRenderTarget(); - if (data->d3d_resources.pSwapChain) - { - data->d3d_resources.pSwapChain->Release(); - data->d3d_resources.pSwapChain = nullptr; - } - if (data->d3d_resources.pd3dDeviceContext) - { - data->d3d_resources.pd3dDeviceContext->Release(); - data->d3d_resources.pd3dDeviceContext = nullptr; - } - if (data->d3d_resources.pd3dDevice) - { - data->d3d_resources.pd3dDevice->Release(); - data->d3d_resources.pd3dDevice = nullptr; - } + if (data->d3d_resources.pSwapChain) + { + data->d3d_resources.pSwapChain->Release(); + data->d3d_resources.pSwapChain = nullptr; + } + if (data->d3d_resources.pd3dDeviceContext) + { + data->d3d_resources.pd3dDeviceContext->Release(); + data->d3d_resources.pd3dDeviceContext = nullptr; + } + if (data->d3d_resources.pd3dDevice) + { + data->d3d_resources.pd3dDevice->Release(); + data->d3d_resources.pd3dDevice = nullptr; + } } static void CreateRenderTarget() { ID3D11Texture2D* pBackBuffer = nullptr; - data->d3d_resources.pSwapChain->GetBuffer(0, IID_PPV_ARGS(&pBackBuffer)); - data->d3d_resources.pd3dDevice->CreateRenderTargetView(pBackBuffer, nullptr, &data->d3d_resources.pMainRenderTargetView); + data->d3d_resources.pSwapChain->GetBuffer(0, IID_PPV_ARGS(&pBackBuffer)); + data->d3d_resources.pd3dDevice->CreateRenderTargetView(pBackBuffer, nullptr, &data->d3d_resources.pMainRenderTargetView); pBackBuffer->Release(); } static void CleanupRenderTarget() { - if (data->d3d_resources.pMainRenderTargetView) - { - data->d3d_resources.pMainRenderTargetView->Release(); - data->d3d_resources.pMainRenderTargetView = nullptr; - } + if (data->d3d_resources.pMainRenderTargetView) + { + data->d3d_resources.pMainRenderTargetView->Release(); + data->d3d_resources.pMainRenderTargetView = nullptr; + } } From 52d1ad98279475b620ba955d914d70cdbab0d33b Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:24:35 +0200 Subject: [PATCH 011/125] (chore): Snake case --- Backends/RmlUi_Renderer_DX11.cpp | 51 ++++++++++++++++---------------- Backends/RmlUi_Renderer_DX11.h | 16 +++++----- 2 files changed, 34 insertions(+), 33 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 93aa27b37..4b0d54416 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -40,19 +40,18 @@ RenderInterface_DX11::RenderInterface_DX11() {} { - RMLUI_ASSERTMSG(pd3dDevice, "pd3dDevice cannot be nullptr!"); - RMLUI_ASSERTMSG(pSwapChain, "pSwapChain cannot be nullptr!"); - RMLUI_ASSERTMSG(pd3dDeviceContext, "pd3dDeviceContext cannot be nullptr!"); void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context) +{ + RMLUI_ASSERTMSG(p_d3d_device, "p_d3d_device cannot be nullptr!"); + RMLUI_ASSERTMSG(p_d3d_device_context, "p_d3d_device_context cannot be nullptr!"); // Assign D3D resources - m_d3dDevice = pd3dDevice; - m_d3dContext = pd3dDeviceContext; - m_swapChain = pSwapChain; + m_d3d_device = p_d3d_device; + m_d3d_context = p_d3d_device_context; // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). - if (!m_blendState) + if (!m_blend_state) { D3D11_BLEND_DESC blendDesc; ZeroMemory(&blendDesc, sizeof(blendDesc)); @@ -66,7 +65,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* blendDesc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA; blendDesc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; blendDesc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; - HRESULT result = m_d3dDevice->CreateBlendState(&blendDesc, &m_blendState); + HRESULT result = m_d3d_device->CreateBlendState(&blendDesc, &m_blend_state); if (FAILED(result)) { Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); @@ -88,7 +87,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* rasterizerDesc.MultisampleEnable = MSAA_SAMPLES > 1; rasterizerDesc.AntialiasedLineEnable = MSAA_SAMPLES > 1; - HRESULT result = m_d3dDevice->CreateRasterizerState(&rasterizerDesc, &m_rasterizerState_scissorEnabled); + HRESULT result = m_d3d_device->CreateRasterizerState(&rasterizerDesc, &m_rasterizer_state_scissor_enabled); if (FAILED(result)) { Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRasterizerState (scissor: enabled) (%d)", result); @@ -97,7 +96,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* rasterizerDesc.ScissorEnable = false; - result = m_d3dDevice->CreateRasterizerState(&rasterizerDesc, &m_rasterizerState_scissorDisabled); + result = m_d3d_device->CreateRasterizerState(&rasterizerDesc, &m_rasterizer_state_scissor_disabled); if (FAILED(result)) { Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRasterizerState (scissor: disabled) (%d)", result); @@ -122,13 +121,15 @@ void RenderInterface_DX11::Cleanup() { m_blendState->Release(); } -void RenderInterface_DX11::BeginFrame(ID3D11RenderTargetView* renderTargetView) { - RMLUI_ASSERTMSG(renderTargetView, "renderTargetView cannot be nullptr!"); - RMLUI_ASSERTMSG(m_d3dContext, "d3dContext cannot be nullptr!"); - RMLUI_ASSERTMSG(m_d3dDevice, "d3dDevice cannot be nullptr!"); void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) +{ + RMLUI_ASSERTMSG(p_swapchain, "p_swapchain cannot be nullptr!"); + RMLUI_ASSERTMSG(p_render_target_view, "p_render_target_view cannot be nullptr!"); + RMLUI_ASSERTMSG(m_d3d_context, "d3d_context cannot be nullptr!"); + RMLUI_ASSERTMSG(m_d3d_device, "d3d_device cannot be nullptr!"); - m_boundRenderTarget = renderTargetView; + m_bound_render_target = p_render_target_view; + m_bound_swapchain = p_swapchain; D3D11_VIEWPORT d3dviewport; d3dviewport.TopLeftX = 0; @@ -137,15 +138,15 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT d3dviewport.Height = m_height; d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; - m_d3dContext->RSSetViewports(1, &d3dviewport); + m_d3d_context->RSSetViewports(1, &d3dviewport); Clear(); - SetBlendState(m_blendState); + SetBlendState(m_blend_state); } void RenderInterface_DX11::EndFrame() { // @TODO: Compositing // @TODO: Layer stack - m_boundRenderTarget = nullptr; + m_bound_render_target = nullptr; } @@ -299,11 +300,11 @@ void RenderInterface_DX11::EnableScissorRegion(bool enable) { { if (enable) { - m_d3dContext->RSSetState(m_rasterizerState_scissorEnabled); + m_d3d_context->RSSetState(m_rasterizer_state_scissor_enabled); } else { - m_d3dContext->RSSetState(m_rasterizerState_scissorDisabled); + m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); } m_scissor_region_enabled = enable; } @@ -318,7 +319,7 @@ void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) if (m_scissor_region_enabled) { - m_d3dContext->RSSetScissorRects(1, &m_rect_scissor); + m_d3d_context->RSSetScissorRects(1, &m_rect_scissor); } } @@ -331,14 +332,14 @@ void RenderInterface_DX11::SetViewport(const int width, const int height) void RenderInterface_DX11::Clear() { float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; - m_d3dContext->ClearRenderTargetView(m_boundRenderTarget, clearColor); + m_d3d_context->ClearRenderTargetView(m_bound_render_target, clearColor); } void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) { - if (blendState != m_currentBlendState) + if (blendState != m_current_blend_state) { - m_d3dContext->OMSetBlendState(blendState, 0, 0xFFFFFFFF); - m_currentBlendState = blendState; + m_d3d_context->OMSetBlendState(blendState, 0, 0xFFFFFFFF); + m_current_blend_state = blendState; } } diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 889026f6d..56cfeaab4 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -98,23 +98,23 @@ class RenderInterface_DX11 : public Rml::RenderInterface { }; // D3D11 core resources - ID3D11Device* m_d3dDevice = nullptr; - ID3D11DeviceContext* m_d3dContext = nullptr; - IDXGISwapChain* m_swapChain = nullptr; - ID3D11RenderTargetView* m_boundRenderTarget = nullptr; - ID3D11RasterizerState* m_rasterizerState_scissorEnabled = nullptr; - ID3D11RasterizerState* m_rasterizerState_scissorDisabled = nullptr; + ID3D11Device* m_d3d_device = nullptr; + ID3D11DeviceContext* m_d3d_context = nullptr; + IDXGISwapChain* m_bound_swapchain = nullptr; + ID3D11RenderTargetView* m_bound_render_target = nullptr; + ID3D11RasterizerState* m_rasterizer_state_scissor_enabled = nullptr; + ID3D11RasterizerState* m_rasterizer_state_scissor_disabled = nullptr; // Viewport dimensions int m_width = 0; int m_height = 0; - ID3D11BlendState* m_blendState = nullptr; + ID3D11BlendState* m_blend_state = nullptr; D3D11_RECT m_rect_scissor = {}; bool m_scissor_region_enabled = false; // D3D11 state - ID3D11BlendState* m_currentBlendState = nullptr; + ID3D11BlendState* m_current_blend_state = nullptr; struct D3D11State { From d830fec11f0abe9369bd2d68204e2985cab83588 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:25:06 +0200 Subject: [PATCH 012/125] (feat): Create shaders and input layout --- Backends/RmlUi_Renderer_DX11.cpp | 240 ++++++++++++++++++++++++++++++- Backends/RmlUi_Renderer_DX11.h | 25 ++++ 2 files changed, 261 insertions(+), 4 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 4b0d54416..e309779fe 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -37,9 +37,96 @@ #include "RmlUi_Include_Windows.h" -RenderInterface_DX11::RenderInterface_DX11() {} + #include "d3dcompiler.h" + #ifdef RMLUI_DX_DEBUG + #include + #endif + +// Shader source code +constexpr const char pShaderSourceText_Color[] = R"( +struct sInputData +{ + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +float4 main(const sInputData inputArgs) : SV_TARGET +{ + return inputArgs.inputColor; +} +)"; +constexpr const char pShaderSourceText_Vertex[] = R"( +struct sInputData +{ + float2 inPosition : POSITION; + float4 inColor : COLOR; + float2 inTexCoord : TEXCOORD; +}; + +struct sOutputData +{ + float4 outPosition : SV_Position; + float4 outColor : COLOR; + float2 outUV : TEXCOORD; +}; + +cbuffer ConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_translate; + float2 m_padding; + float4 m_padding1[11]; +}; + +sOutputData main(const sInputData inArgs) +{ + sOutputData result; + + float2 translatedPos = inArgs.inPosition + m_translate; + float4 resPos = mul(m_transform, float4(translatedPos.x, translatedPos.y, 0.0, 1.0)); + + result.outPosition = resPos; + result.outColor = inArgs.inColor; + result.outUV = inArgs.inTexCoord; + +#if defined(RMLUI_PREMULTIPLIED_ALPHA) + // Pre-multiply vertex colors with their alpha. + result.outColor.rgb = result.outColor.rgb * result.outColor.a; +#endif + + return result; +}; +)"; + +constexpr const char pShaderSourceText_Texture[] = R"( +struct sInputData { + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +Texture2D g_InputTexture : register(t0); + +SamplerState g_SamplerLinear : register(s0); + + +float4 main(const sInputData inputArgs) : SV_TARGET +{ + return inputArgs.inputColor * g_InputTexture.Sample(g_SamplerLinear, inputArgs.inputUV); +} +)"; + +RenderInterface_DX11::RenderInterface_DX11() { + #ifdef RMLUI_DX_DEBUG + m_default_shader_flags = D3DCOMPILE_DEBUG | D3DCOMPILE_SKIP_OPTIMIZATION; + #else + m_default_shader_flags = 0; + #endif +} + void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context) { RMLUI_ASSERTMSG(p_d3d_device, "p_d3d_device cannot be nullptr!"); @@ -66,11 +153,14 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* blendDesc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; blendDesc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; HRESULT result = m_d3d_device->CreateBlendState(&blendDesc, &m_blend_state); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); + #ifdef RMLUI_DX_DEBUG if (FAILED(result)) { Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); return; } + #endif } // Scissor regions require a rasterizer state. Cache one for scissor on and off @@ -88,37 +178,179 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* rasterizerDesc.AntialiasedLineEnable = MSAA_SAMPLES > 1; HRESULT result = m_d3d_device->CreateRasterizerState(&rasterizerDesc, &m_rasterizer_state_scissor_enabled); + RMLUI_DX_ASSERTMSG(result, "failed to CreateRasterizerState"); + #ifdef RMLUI_DX_DEBUG if (FAILED(result)) { Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRasterizerState (scissor: enabled) (%d)", result); return; } + #endif rasterizerDesc.ScissorEnable = false; result = m_d3d_device->CreateRasterizerState(&rasterizerDesc, &m_rasterizer_state_scissor_disabled); + RMLUI_DX_ASSERTMSG(result, "failed to CreateRasterizerState"); + #ifdef RMLUI_DX_DEBUG if (FAILED(result)) { Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRasterizerState (scissor: disabled) (%d)", result); return; } + #endif } - // Load shaders + // Compile and load shaders + + // Buffer shall be cleared up later, as it's required to create the input layout + ID3DBlob* p_shader_vertex_common{}; - // @TODO: + { + ID3DBlob* p_error_buff{}; + + // Common vertex shader + HRESULT result = D3DCompile(pShaderSourceText_Vertex, sizeof(pShaderSourceText_Vertex), nullptr, nullptr, nullptr, "main", "vs_5_0", + this->m_default_shader_flags, 0, &p_shader_vertex_common, &p_error_buff); + RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); + } + #endif + + DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); + + // Color fragment shader + ID3DBlob* p_shader_color_pixel{}; + + result = D3DCompile(pShaderSourceText_Color, sizeof(pShaderSourceText_Color), nullptr, nullptr, nullptr, "main", "ps_5_0", + this->m_default_shader_flags, 0, &p_shader_color_pixel, &p_error_buff); + RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); + } + #endif + + DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); + + // Texture fragment shader + ID3DBlob* p_shader_color_texture{}; + + result = D3DCompile(pShaderSourceText_Texture, sizeof(pShaderSourceText_Texture), nullptr, nullptr, nullptr, "main", "ps_5_0", + this->m_default_shader_flags, 0, &p_shader_color_texture, &p_error_buff); + RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); + } + #endif + + DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); + + // Create the shader objects + result = m_d3d_device->CreateVertexShader(p_shader_vertex_common->GetBufferPointer(), p_shader_vertex_common->GetBufferSize(), NULL, + &m_shader_vertex_common); + RMLUI_DX_ASSERTMSG(result, "failed to CreateVertexShader"); + if (FAILED(result)) + { + #ifdef RMLUI_DX_DEBUG + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create vertex shader: %d", result); + #endif + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + goto cleanup; + return; + } + + result = m_d3d_device->CreatePixelShader(p_shader_color_pixel->GetBufferPointer(), p_shader_color_pixel->GetBufferSize(), NULL, + &m_shader_pixel_color); + RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); + if (FAILED(result)) + { + #ifdef RMLUI_DX_DEBUG + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create pixel shader: %d", result); + #endif + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + goto cleanup; + return; + } + + result = m_d3d_device->CreatePixelShader(p_shader_color_texture->GetBufferPointer(), p_shader_color_texture->GetBufferSize(), NULL, + &m_shader_pixel_texture); + RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); + if (FAILED(result)) + { + #ifdef RMLUI_DX_DEBUG + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create pixel shader: %d", result); + #endif + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + goto cleanup; + return; + } + } // Create vertex layout. This will be constant to avoid copying to an intermediate struct. // @TODO: + { + D3D11_INPUT_ELEMENT_DESC polygonLayout[3]{}; + + polygonLayout[0].SemanticName = "POSITION"; + polygonLayout[0].SemanticIndex = 0; + polygonLayout[0].Format = DXGI_FORMAT_R32G32B32_FLOAT; + polygonLayout[0].InputSlot = 0; + polygonLayout[0].AlignedByteOffset = 0; + polygonLayout[0].InputSlotClass = D3D11_INPUT_PER_VERTEX_DATA; + polygonLayout[0].InstanceDataStepRate = 0; + + polygonLayout[1].SemanticName = "COLOR"; + polygonLayout[1].SemanticIndex = 0; + polygonLayout[1].Format = DXGI_FORMAT_R8G8B8A8_UNORM_SRGB; + polygonLayout[1].InputSlot = 0; + polygonLayout[1].AlignedByteOffset = D3D11_APPEND_ALIGNED_ELEMENT; + polygonLayout[1].InputSlotClass = D3D11_INPUT_PER_VERTEX_DATA; + polygonLayout[1].InstanceDataStepRate = 0; + + polygonLayout[2].SemanticName = "TEXCOORD"; + polygonLayout[2].SemanticIndex = 0; + polygonLayout[2].Format = DXGI_FORMAT_R32G32_FLOAT; + polygonLayout[2].InputSlot = 0; + polygonLayout[2].AlignedByteOffset = D3D11_APPEND_ALIGNED_ELEMENT; + polygonLayout[2].InputSlotClass = D3D11_INPUT_PER_VERTEX_DATA; + polygonLayout[2].InstanceDataStepRate = 0; + + // Get a count of the elements in the layout. + uint32_t numElements = sizeof(polygonLayout) / sizeof(polygonLayout[0]); + + // Create the vertex input layout. + HRESULT result = m_d3d_device->CreateInputLayout(polygonLayout, numElements, p_shader_vertex_common->GetBufferPointer(), + p_shader_vertex_common->GetBufferSize(), &m_vertex_layout); + RMLUI_DX_ASSERTMSG(result, "failed to CreateInputLayout"); + if (FAILED(result)) + { + goto cleanup; + return; + } + } // Create constant buffers. This is so that we can bind uniforms such as translation and color to the shaders. // @TODO: + +cleanup: + + // Release the vertex shader buffer and pixel shader buffer since they are no longer needed. + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_vertex_common); } void RenderInterface_DX11::Cleanup() { - m_blendState->Release(); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 56cfeaab4..3ca8180f8 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -37,6 +37,23 @@ #include #include +#ifdef RMLUI_DEBUG + #define RMLUI_DX_ASSERTMSG(statement, msg) RMLUI_ASSERTMSG(SUCCEEDED(statement), msg) + + // Uncomment the following line to enable additional DirectX debugging. + #define RMLUI_DX_DEBUG +#else + #define RMLUI_DX_ASSERTMSG(statement, msg) static_cast(statement) +#endif + +// Helper for a common cleanup pattern in DX11 +#define DX_CLEANUP_RESOURCE_IF_CREATED(resource) \ + if (resource) \ + { \ + resource->Release(); \ + resource = nullptr; \ + } + // Allow the user to override the number of MSAA samples #ifndef MSAA_SAMPLES #define MSAA_SAMPLES 2 @@ -97,6 +114,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::Span indices; }; + UINT m_default_shader_flags; + // D3D11 core resources ID3D11Device* m_d3d_device = nullptr; ID3D11DeviceContext* m_d3d_context = nullptr; @@ -105,10 +124,16 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11RasterizerState* m_rasterizer_state_scissor_enabled = nullptr; ID3D11RasterizerState* m_rasterizer_state_scissor_disabled = nullptr; + // Shaders + ID3D11VertexShader* m_shader_vertex_common = nullptr; + ID3D11PixelShader* m_shader_pixel_color = nullptr; + ID3D11PixelShader* m_shader_pixel_texture = nullptr; + // Viewport dimensions int m_width = 0; int m_height = 0; + ID3D11InputLayout* m_vertex_layout = nullptr; ID3D11BlendState* m_blend_state = nullptr; D3D11_RECT m_rect_scissor = {}; bool m_scissor_region_enabled = false; From db21e66154ecb8fb9e784fc350340c2bd6576a17 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:30:19 +0200 Subject: [PATCH 013/125] (fix): Make backend free DX11 resources on shutdown --- Backends/RmlUi_Backend_Win32_DX11.cpp | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 03036e84a..3edc673c4 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -178,8 +178,12 @@ void Backend::Shutdown() // As we forcefully override the global text input handler, we must reset it before the data is destroyed to avoid any potential use-after-free. if (Rml::GetTextInputHandler() == &data->text_input_method_editor) - Rml::SetTextInputHandler(nullptr); + Rml::SetTextInputHandler(nullptr); + // Cleanup renderer resources + data->render_interface.Cleanup(); + + // Shutdown DirectX11 CleanupDeviceD3D(); ::DestroyWindow(data->window_handle); From f22dac3cd6dd18d8a7099847c5267e2b3b8acfbb Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:30:39 +0200 Subject: [PATCH 014/125] (fix): Make input layout declaration more compact --- Backends/RmlUi_Renderer_DX11.cpp | 30 +++++------------------------- 1 file changed, 5 insertions(+), 25 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e309779fe..6770fafe9 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -299,31 +299,11 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // @TODO: { - D3D11_INPUT_ELEMENT_DESC polygonLayout[3]{}; - - polygonLayout[0].SemanticName = "POSITION"; - polygonLayout[0].SemanticIndex = 0; - polygonLayout[0].Format = DXGI_FORMAT_R32G32B32_FLOAT; - polygonLayout[0].InputSlot = 0; - polygonLayout[0].AlignedByteOffset = 0; - polygonLayout[0].InputSlotClass = D3D11_INPUT_PER_VERTEX_DATA; - polygonLayout[0].InstanceDataStepRate = 0; - - polygonLayout[1].SemanticName = "COLOR"; - polygonLayout[1].SemanticIndex = 0; - polygonLayout[1].Format = DXGI_FORMAT_R8G8B8A8_UNORM_SRGB; - polygonLayout[1].InputSlot = 0; - polygonLayout[1].AlignedByteOffset = D3D11_APPEND_ALIGNED_ELEMENT; - polygonLayout[1].InputSlotClass = D3D11_INPUT_PER_VERTEX_DATA; - polygonLayout[1].InstanceDataStepRate = 0; - - polygonLayout[2].SemanticName = "TEXCOORD"; - polygonLayout[2].SemanticIndex = 0; - polygonLayout[2].Format = DXGI_FORMAT_R32G32_FLOAT; - polygonLayout[2].InputSlot = 0; - polygonLayout[2].AlignedByteOffset = D3D11_APPEND_ALIGNED_ELEMENT; - polygonLayout[2].InputSlotClass = D3D11_INPUT_PER_VERTEX_DATA; - polygonLayout[2].InstanceDataStepRate = 0; + D3D11_INPUT_ELEMENT_DESC polygonLayout[] = { + {"POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0}, + {"COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, + {"TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, + }; // Get a count of the elements in the layout. uint32_t numElements = sizeof(polygonLayout) / sizeof(polygonLayout[0]); From 53e0e668e0398d36639f6beb0904ffb13489971a Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:30:55 +0200 Subject: [PATCH 015/125] (fix): Free all allocated resources on Cleanup --- Backends/RmlUi_Renderer_DX11.cpp | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 6770fafe9..0cb122944 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -330,7 +330,13 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } void RenderInterface_DX11::Cleanup() { + // Cleans up all resources + DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); + DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); + DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_vertex_common); + DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_color); + DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_texture); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) From 299b69b59f3a7cd992dc27d0a4b13633b2d8b72c Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 10:33:42 +0200 Subject: [PATCH 016/125] (fix): Reset blend state on end frame --- Backends/RmlUi_Renderer_DX11.cpp | 3 +++ Backends/RmlUi_Renderer_DX11.h | 1 + 2 files changed, 4 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 0cb122944..5a9151e1a 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -364,7 +364,10 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT void RenderInterface_DX11::EndFrame() { // @TODO: Compositing // @TODO: Layer stack + m_bound_swapchain = nullptr; m_bound_render_target = nullptr; + // Reset blend state + m_current_blend_state = nullptr; } diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 3ca8180f8..179b0a5e2 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -105,6 +105,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { pfnFreeTextureRaw FreeTextureFromFileRaw = nullptr; private: + // Changes blend state if necessary void SetBlendState(ID3D11BlendState* blendState); private: From 4200dac9b884915381a90623873ef871c2ba3e28 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 12:05:17 +0200 Subject: [PATCH 017/125] (fix): Use DEFAULT macros for rasterizer state --- Backends/RmlUi_Renderer_DX11.cpp | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 5a9151e1a..2de8cbe3b 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -169,9 +169,9 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* rasterizerDesc.FillMode = D3D11_FILL_SOLID; rasterizerDesc.CullMode = D3D11_CULL_BACK; rasterizerDesc.FrontCounterClockwise = false; - rasterizerDesc.DepthBias = 0; - rasterizerDesc.SlopeScaledDepthBias = 0.0f; - rasterizerDesc.DepthBiasClamp = 0.0f; + rasterizerDesc.DepthBias = D3D11_DEFAULT_DEPTH_BIAS; + rasterizerDesc.SlopeScaledDepthBias = D3D11_DEFAULT_SLOPE_SCALED_DEPTH_BIAS; + rasterizerDesc.DepthBiasClamp = D3D11_DEFAULT_DEPTH_BIAS_CLAMP; rasterizerDesc.DepthClipEnable = true; rasterizerDesc.ScissorEnable = true; rasterizerDesc.MultisampleEnable = MSAA_SAMPLES > 1; @@ -297,7 +297,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Create vertex layout. This will be constant to avoid copying to an intermediate struct. - // @TODO: { D3D11_INPUT_ELEMENT_DESC polygonLayout[] = { {"POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0}, From b1b3c2144b31518760e808fc457c7a4e5083c145 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 12:06:32 +0200 Subject: [PATCH 018/125] (feat): Add SetTransform support --- Backends/RmlUi_Renderer_DX11.cpp | 15 +++++++++++---- Backends/RmlUi_Renderer_DX11.h | 16 +++++++++++----- 2 files changed, 22 insertions(+), 9 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 2de8cbe3b..c0a440c3a 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -351,8 +351,8 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT D3D11_VIEWPORT d3dviewport; d3dviewport.TopLeftX = 0; d3dviewport.TopLeftY = 0; - d3dviewport.Width = m_width; - d3dviewport.Height = m_height; + d3dviewport.Width = m_viewport_width; + d3dviewport.Height = m_viewport_height; d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); @@ -545,8 +545,9 @@ void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) void RenderInterface_DX11::SetViewport(const int width, const int height) { - m_width = width; - m_height = height; + m_viewport_width = width; + m_viewport_height = height; + m_projection = Rml::Matrix4f::ProjectOrtho(0, (float)m_viewport_width, (float)m_viewport_height, 0, -10000, 10000); } void RenderInterface_DX11::Clear() @@ -563,4 +564,10 @@ void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) { } } +void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) +{ + m_transform = (new_transform ? (m_projection * (*new_transform)) : m_projection); + m_cbuffer_dirty = true; +} + #endif // RMLUI_PLATFORM_WIN32 \ No newline at end of file diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 179b0a5e2..f0d04c739 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -76,6 +76,9 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view); void EndFrame(); + void SetViewport(const int width, const int height); + void Clear(); + // -- Inherited from Rml::RenderInterface -- Rml::CompiledGeometryHandle CompileGeometry(Rml::Span vertices, Rml::Span indices) override; @@ -88,9 +91,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void EnableScissorRegion(bool enable) override; void SetScissorRegion(Rml::Rectanglei region) override; - - void SetViewport(const int width, const int height); - void Clear(); + + void SetTransform(const Rml::Matrix4f* transform) override; public: /// Called by the renderer when it wants to load a texture from disk. @@ -126,13 +128,17 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11RasterizerState* m_rasterizer_state_scissor_disabled = nullptr; // Shaders + bool m_cbuffer_dirty = true; ID3D11VertexShader* m_shader_vertex_common = nullptr; ID3D11PixelShader* m_shader_pixel_color = nullptr; ID3D11PixelShader* m_shader_pixel_texture = nullptr; // Viewport dimensions - int m_width = 0; - int m_height = 0; + int m_viewport_width = 0; + int m_viewport_height = 0; + + Rml::Matrix4f m_transform; + Rml::Matrix4f m_projection; ID3D11InputLayout* m_vertex_layout = nullptr; ID3D11BlendState* m_blend_state = nullptr; From 24a2c38f557e3bf75d1817c3fc31011c9b970c2c Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 23 Sep 2024 12:06:54 +0200 Subject: [PATCH 019/125] (feat): Implement CBuffer --- Backends/RmlUi_Renderer_DX11.cpp | 19 ++++++++++++++++++- Backends/RmlUi_Renderer_DX11.h | 13 ++++++++++++- 2 files changed, 30 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index c0a440c3a..a38299a55 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -320,7 +320,23 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Create constant buffers. This is so that we can bind uniforms such as translation and color to the shaders. - // @TODO: + { + D3D11_BUFFER_DESC cbufferDesc{}; + cbufferDesc.Usage = D3D11_USAGE_DYNAMIC; + cbufferDesc.ByteWidth = sizeof(ShaderCbuffer); + cbufferDesc.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + cbufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + cbufferDesc.MiscFlags = 0; + cbufferDesc.StructureByteStride = 0; + + HRESULT result = m_d3d_device->CreateBuffer(&cbufferDesc, NULL, &m_shader_buffer); + RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + if (FAILED(result)) + { + goto cleanup; + return; + } + } cleanup: @@ -336,6 +352,7 @@ void RenderInterface_DX11::Cleanup() { DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_vertex_common); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_color); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index f0d04c739..b9f3cadae 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -128,6 +128,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11RasterizerState* m_rasterizer_state_scissor_disabled = nullptr; // Shaders + ID3D11Buffer* m_shader_buffer = nullptr; bool m_cbuffer_dirty = true; ID3D11VertexShader* m_shader_vertex_common = nullptr; ID3D11PixelShader* m_shader_pixel_color = nullptr; @@ -153,7 +154,17 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11BlendState* m_previousBlendState = nullptr; }; - D3D11State m_previousD3DState; + D3D11State m_previousD3DState{}; + + #pragma pack(4) + struct ShaderCbuffer { + Rml::Matrix4f transform; + Rml::Vector2f translation; + float _padding[2]; + }; + #pragma pack() + + ShaderCbuffer m_shader_cbuffer_data{}; }; #endif // RMLUI_PLATFORM_WIN32 From d9bab9d8dcd86ea21f1569b5b43ba577876e1768 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:25:12 +0200 Subject: [PATCH 020/125] (fix): Link to d3dcompiler --- Backends/CMakeLists.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/CMakeLists.txt b/Backends/CMakeLists.txt index d4edcb84b..1d05101d4 100644 --- a/Backends/CMakeLists.txt +++ b/Backends/CMakeLists.txt @@ -43,7 +43,7 @@ target_sources(rmlui_backend_Win32_DX11 INTERFACE "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Renderer_DX11.h" "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Include_Windows.h" ) -target_link_libraries(rmlui_backend_Win32_DX11 INTERFACE rmlui_backend_common_headers d3d11.lib dxgi.lib ) +target_link_libraries(rmlui_backend_Win32_DX11 INTERFACE rmlui_backend_common_headers d3d11.lib dxgi.lib d3dcompiler.lib) add_library(rmlui_backend_X11_GL2 INTERFACE) target_sources(rmlui_backend_X11_GL2 INTERFACE From 1a723c488e157669e68e683b7dc67480bcc00b07 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:25:54 +0200 Subject: [PATCH 021/125] (fix): Enable DX11 debug layer in debug mode --- Backends/RmlUi_Backend_Win32_DX11.cpp | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 3edc673c4..4693d1100 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -432,8 +432,12 @@ static bool CreateDeviceD3D(HWND hwnd) { sd.Windowed = TRUE; sd.SwapEffect = DXGI_SWAP_EFFECT_DISCARD; + UINT createDeviceFlags = 0; - //createDeviceFlags |= D3D11_CREATE_DEVICE_DEBUG; +#ifdef RMLUI_DEBUG + // Enable debug layer + createDeviceFlags |= D3D11_CREATE_DEVICE_DEBUG; +#endif D3D_FEATURE_LEVEL featureLevel; const D3D_FEATURE_LEVEL featureLevelArray[2] = { D3D_FEATURE_LEVEL_11_0, D3D_FEATURE_LEVEL_10_0, }; HRESULT res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_HARDWARE, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, From b2e4322e43ae4030c62a55638efc67b8a2796927 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:27:13 +0200 Subject: [PATCH 022/125] (fix): Remove unnecessary padding in cbuffer --- Backends/RmlUi_Renderer_DX11.cpp | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index a38299a55..7287b447b 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -76,8 +76,7 @@ cbuffer ConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; - float2 m_padding; - float4 m_padding1[11]; + float2 _padding; }; sOutputData main(const sInputData inArgs) From e0705bdbc52add0e10107e91266a112fbdbd452c Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:36:08 +0200 Subject: [PATCH 023/125] (fix): Initialise descriptors before usage --- Backends/RmlUi_Renderer_DX11.cpp | 11 ++++------- Backends/RmlUi_Renderer_DX11.h | 5 +++-- 2 files changed, 7 insertions(+), 9 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 7287b447b..59623bda8 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -108,12 +108,11 @@ struct sInputData }; Texture2D g_InputTexture : register(t0); - SamplerState g_SamplerLinear : register(s0); float4 main(const sInputData inputArgs) : SV_TARGET -{ +{ return inputArgs.inputColor * g_InputTexture.Sample(g_SamplerLinear, inputArgs.inputUV); } )"; @@ -139,7 +138,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). if (!m_blend_state) { - D3D11_BLEND_DESC blendDesc; + D3D11_BLEND_DESC blendDesc{}; ZeroMemory(&blendDesc, sizeof(blendDesc)); blendDesc.AlphaToCoverageEnable = FALSE; blendDesc.IndependentBlendEnable = FALSE; @@ -164,7 +163,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Scissor regions require a rasterizer state. Cache one for scissor on and off { - D3D11_RASTERIZER_DESC rasterizerDesc; + D3D11_RASTERIZER_DESC rasterizerDesc{}; rasterizerDesc.FillMode = D3D11_FILL_SOLID; rasterizerDesc.CullMode = D3D11_CULL_BACK; rasterizerDesc.FrontCounterClockwise = false; @@ -302,11 +301,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* {"COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, {"TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, }; - - // Get a count of the elements in the layout. uint32_t numElements = sizeof(polygonLayout) / sizeof(polygonLayout[0]); - // Create the vertex input layout. HRESULT result = m_d3d_device->CreateInputLayout(polygonLayout, numElements, p_shader_vertex_common->GetBufferPointer(), p_shader_vertex_common->GetBufferSize(), &m_vertex_layout); RMLUI_DX_ASSERTMSG(result, "failed to CreateInputLayout"); @@ -335,6 +331,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* goto cleanup; return; } + } cleanup: diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index b9f3cadae..948e2d788 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -138,8 +138,9 @@ class RenderInterface_DX11 : public Rml::RenderInterface { int m_viewport_width = 0; int m_viewport_height = 0; - Rml::Matrix4f m_transform; - Rml::Matrix4f m_projection; + Rml::Matrix4f m_transform = Rml::Matrix4f::Identity(); + Rml::Matrix4f m_projection = Rml::Matrix4f::Identity(); + Rml::Vector2f m_translation = {}; ID3D11InputLayout* m_vertex_layout = nullptr; ID3D11BlendState* m_blend_state = nullptr; From 3e691117348f3a91dbd89593fa1997170503cab0 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:37:38 +0200 Subject: [PATCH 024/125] (fix): Feed macros into shader compiler --- Backends/RmlUi_Renderer_DX11.cpp | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 59623bda8..eac6c119b 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -204,10 +204,11 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* ID3DBlob* p_shader_vertex_common{}; { + const D3D_SHADER_MACRO macros[] = {"RMLUI_PREMULTIPLIED_ALPHA", nullptr, nullptr, nullptr}; ID3DBlob* p_error_buff{}; // Common vertex shader - HRESULT result = D3DCompile(pShaderSourceText_Vertex, sizeof(pShaderSourceText_Vertex), nullptr, nullptr, nullptr, "main", "vs_5_0", + HRESULT result = D3DCompile(pShaderSourceText_Vertex, sizeof(pShaderSourceText_Vertex), nullptr, macros, nullptr, "main", "vs_5_0", this->m_default_shader_flags, 0, &p_shader_vertex_common, &p_error_buff); RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); #ifdef RMLUI_DX_DEBUG @@ -222,7 +223,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Color fragment shader ID3DBlob* p_shader_color_pixel{}; - result = D3DCompile(pShaderSourceText_Color, sizeof(pShaderSourceText_Color), nullptr, nullptr, nullptr, "main", "ps_5_0", + result = D3DCompile(pShaderSourceText_Color, sizeof(pShaderSourceText_Color), nullptr, macros, nullptr, "main", "ps_5_0", this->m_default_shader_flags, 0, &p_shader_color_pixel, &p_error_buff); RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); #ifdef RMLUI_DX_DEBUG @@ -237,7 +238,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Texture fragment shader ID3DBlob* p_shader_color_texture{}; - result = D3DCompile(pShaderSourceText_Texture, sizeof(pShaderSourceText_Texture), nullptr, nullptr, nullptr, "main", "ps_5_0", + result = D3DCompile(pShaderSourceText_Texture, sizeof(pShaderSourceText_Texture), nullptr, macros, nullptr, "main", "ps_5_0", this->m_default_shader_flags, 0, &p_shader_color_texture, &p_error_buff); RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); #ifdef RMLUI_DX_DEBUG From 4880302077617e05476ce67b2d053b5f01c74401 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:38:08 +0200 Subject: [PATCH 025/125] (chore): Use nullptr instead of NULL --- Backends/RmlUi_Renderer_DX11.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index eac6c119b..120597cce 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -251,7 +251,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); // Create the shader objects - result = m_d3d_device->CreateVertexShader(p_shader_vertex_common->GetBufferPointer(), p_shader_vertex_common->GetBufferSize(), NULL, + result = m_d3d_device->CreateVertexShader(p_shader_vertex_common->GetBufferPointer(), p_shader_vertex_common->GetBufferSize(), nullptr, &m_shader_vertex_common); RMLUI_DX_ASSERTMSG(result, "failed to CreateVertexShader"); if (FAILED(result)) @@ -265,7 +265,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* return; } - result = m_d3d_device->CreatePixelShader(p_shader_color_pixel->GetBufferPointer(), p_shader_color_pixel->GetBufferSize(), NULL, + result = m_d3d_device->CreatePixelShader(p_shader_color_pixel->GetBufferPointer(), p_shader_color_pixel->GetBufferSize(), nullptr, &m_shader_pixel_color); RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); if (FAILED(result)) @@ -279,7 +279,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* return; } - result = m_d3d_device->CreatePixelShader(p_shader_color_texture->GetBufferPointer(), p_shader_color_texture->GetBufferSize(), NULL, + result = m_d3d_device->CreatePixelShader(p_shader_color_texture->GetBufferPointer(), p_shader_color_texture->GetBufferSize(), nullptr, &m_shader_pixel_texture); RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); if (FAILED(result)) From 1572241e50c3e823374b73993b11b272222608e1 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:39:10 +0200 Subject: [PATCH 026/125] (feat): Implement UpdateConstantBuffer --- Backends/RmlUi_Renderer_DX11.cpp | 25 +++++++++++++++++++++++++ Backends/RmlUi_Renderer_DX11.h | 3 +-- 2 files changed, 26 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 120597cce..f28a4b64c 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -584,4 +584,29 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) m_cbuffer_dirty = true; } +void RenderInterface_DX11::UpdateConstantBuffer() { + + if (m_cbuffer_dirty) + { + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return; + } + + // Get a pointer to the data in the constant buffer. + ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + + // Copy the data to the GPU + dataPtr->transform = m_transform; + dataPtr->translation = m_translation; + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + m_cbuffer_dirty = false; + } +} + #endif // RMLUI_PLATFORM_WIN32 \ No newline at end of file diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 948e2d788..4f6228071 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -109,6 +109,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { private: // Changes blend state if necessary void SetBlendState(ID3D11BlendState* blendState); + void UpdateConstantBuffer(); private: // @TODO: Replace with vertex / index buffer pairs @@ -164,8 +165,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { float _padding[2]; }; #pragma pack() - - ShaderCbuffer m_shader_cbuffer_data{}; }; #endif // RMLUI_PLATFORM_WIN32 From 68acddadc5a37302aa43df06d48953a73149100b Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:39:37 +0200 Subject: [PATCH 027/125] (feat): Create texture sampler --- Backends/RmlUi_Renderer_DX11.cpp | 33 ++++++++++++++++++++++++++++++-- Backends/RmlUi_Renderer_DX11.h | 1 + 2 files changed, 32 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index f28a4b64c..a8581f672 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -325,8 +325,36 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* cbufferDesc.MiscFlags = 0; cbufferDesc.StructureByteStride = 0; - HRESULT result = m_d3d_device->CreateBuffer(&cbufferDesc, NULL, &m_shader_buffer); - RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + HRESULT result = m_d3d_device->CreateBuffer(&cbufferDesc, nullptr, &m_shader_buffer); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); + if (FAILED(result)) + { + goto cleanup; + return; + } + } + + // Create sampler state for textures + { + // Create a texture sampler state description. + D3D11_SAMPLER_DESC samplerDesc{}; + samplerDesc.Filter = D3D11_FILTER_MIN_MAG_MIP_LINEAR; + samplerDesc.AddressU = D3D11_TEXTURE_ADDRESS_WRAP; + samplerDesc.AddressV = D3D11_TEXTURE_ADDRESS_WRAP; + samplerDesc.AddressW = D3D11_TEXTURE_ADDRESS_WRAP; + samplerDesc.MipLODBias = 0.0f; + samplerDesc.MaxAnisotropy = 1; + samplerDesc.ComparisonFunc = D3D11_COMPARISON_ALWAYS; + samplerDesc.BorderColor[0] = 0; + samplerDesc.BorderColor[1] = 0; + samplerDesc.BorderColor[2] = 0; + samplerDesc.BorderColor[3] = 0; + samplerDesc.MinLOD = 0; + samplerDesc.MaxLOD = D3D11_FLOAT32_MAX; + + // Create the texture sampler state. + HRESULT result = m_d3d_device->CreateSamplerState(&samplerDesc, &m_samplerState); + RMLUI_DX_ASSERTMSG(result, "failed to CreateSamplerState"); if (FAILED(result)) { goto cleanup; @@ -350,6 +378,7 @@ void RenderInterface_DX11::Cleanup() { DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_color); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_texture); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 4f6228071..1764362b9 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -131,6 +131,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { // Shaders ID3D11Buffer* m_shader_buffer = nullptr; bool m_cbuffer_dirty = true; + ID3D11SamplerState* m_samplerState = nullptr; ID3D11VertexShader* m_shader_vertex_common = nullptr; ID3D11PixelShader* m_shader_pixel_color = nullptr; ID3D11PixelShader* m_shader_pixel_texture = nullptr; From afaa119cfce4a08b926caab1259603c4ba4dd349 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:40:12 +0200 Subject: [PATCH 028/125] (feat): Implement texture loading --- Backends/RmlUi_Renderer_DX11.cpp | 61 ++++++++++++++++++++++++++++++-- Backends/RmlUi_Renderer_DX11.h | 5 +++ 2 files changed, 63 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index a8581f672..eea549f17 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -466,8 +466,7 @@ Rml::TextureHandle RenderInterface_DX11::LoadTexture(Rml::Vector2i& texture_dime return handle; } - // Image must be invalid if the file failed to load. Return invalid handle - return Rml::TextureHandle(0); + // Image must be invalid if the file failed to load. Fallback to the default loader. } Rml::FileInterface* file_interface = Rml::GetFileInterface(); @@ -549,12 +548,68 @@ Rml::TextureHandle RenderInterface_DX11::LoadTexture(Rml::Vector2i& texture_dime Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Span source, Rml::Vector2i source_dimensions) { - return Rml::TextureHandle(0); + ID3D11Texture2D* gpu_texture = nullptr; + ID3D11ShaderResourceView* gpu_texture_view = nullptr; + + D3D11_TEXTURE2D_DESC textureDesc{}; + textureDesc.Width = source_dimensions.x; + textureDesc.Height = source_dimensions.y; + textureDesc.MipLevels = 0; + textureDesc.ArraySize = 1; + textureDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM_SRGB; + textureDesc.SampleDesc.Count = 1; + textureDesc.SampleDesc.Quality = 0; + textureDesc.Usage = D3D11_USAGE_DEFAULT; + textureDesc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_RENDER_TARGET; + textureDesc.CPUAccessFlags = 0; + textureDesc.MiscFlags = D3D11_RESOURCE_MISC_GENERATE_MIPS; + + HRESULT result = m_d3d_device->CreateTexture2D(&textureDesc, nullptr, &gpu_texture); + RMLUI_DX_ASSERTMSG(result, "failed to CreateTexture2D"); + if (FAILED(result)) + { + return Rml::TextureHandle(0); + } + + // Set the row pitch of the raw image data + uint32_t rowPitch = (source_dimensions.x * 4) * sizeof(unsigned char); + + // Copy the raw image data into the texture + m_d3d_context->UpdateSubresource(gpu_texture, 0, nullptr, source.data(), rowPitch, 0); + + // Setup the shader resource view description. + D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc{}; + srvDesc.Format = textureDesc.Format; + srvDesc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE2D; + srvDesc.Texture2D.MostDetailedMip = 0; + srvDesc.Texture2D.MipLevels = -1; + + // Create the shader resource view for the texture. + result = m_d3d_device->CreateShaderResourceView(gpu_texture, &srvDesc, &gpu_texture_view); + RMLUI_DX_ASSERTMSG(result, "failed to CreateShaderResourceView"); + if (FAILED(result)) + { + DX_CLEANUP_RESOURCE_IF_CREATED(gpu_texture); + return Rml::TextureHandle(0); + } + + DX11_TextureData* texData = new DX11_TextureData; + texData->texture = gpu_texture; + texData->texture_view = gpu_texture_view; + + // Generate mipmaps for this texture. + m_d3d_context->GenerateMips(texData->texture_view); + + return Rml::TextureHandle(texData); } void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) { + DX11_TextureData* texData = (DX11_TextureData*)texture_handle; + DX_CLEANUP_RESOURCE_IF_CREATED(texData->texture_view); + DX_CLEANUP_RESOURCE_IF_CREATED(texData->texture); + delete reinterpret_cast(texData); } void RenderInterface_DX11::EnableScissorRegion(bool enable) { diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 1764362b9..0c073e683 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -149,6 +149,11 @@ class RenderInterface_DX11 : public Rml::RenderInterface { D3D11_RECT m_rect_scissor = {}; bool m_scissor_region_enabled = false; + struct DX11_TextureData { + public: + ID3D11Texture2D* texture = nullptr; + ID3D11ShaderResourceView* texture_view = nullptr; + }; // D3D11 state ID3D11BlendState* m_current_blend_state = nullptr; From adaf51f5f8b680827fadaac7fd7af3664081c1fd Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:40:55 +0200 Subject: [PATCH 029/125] (feat): Implement geometry submission --- Backends/RmlUi_Renderer_DX11.cpp | 106 ++++++++++++++++++++++++++++++- Backends/RmlUi_Renderer_DX11.h | 8 +++ 2 files changed, 111 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index eea549f17..84c292124 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -414,15 +414,115 @@ void RenderInterface_DX11::EndFrame() { Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span vertices, Rml::Span indices) { - return Rml::CompiledGeometryHandle(0); + + ID3D11Buffer* vertex_buffer = nullptr; + ID3D11Buffer* index_buffer = nullptr; + + // Bind vertex buffer + { + D3D11_BUFFER_DESC vertexBufferDesc{}; + vertexBufferDesc.Usage = D3D11_USAGE_DEFAULT; + vertexBufferDesc.ByteWidth = sizeof(Rml::Vertex) * vertices.size(); + vertexBufferDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; + vertexBufferDesc.CPUAccessFlags = 0; + vertexBufferDesc.MiscFlags = 0; + vertexBufferDesc.StructureByteStride = 0; + + D3D11_SUBRESOURCE_DATA vertexData{}; + vertexData.pSysMem = vertices.data(); + vertexData.SysMemPitch = 0; + vertexData.SysMemSlicePitch = 0; + + HRESULT result = m_d3d_device->CreateBuffer(&vertexBufferDesc, &vertexData, &vertex_buffer); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); + if (FAILED(result)) + { + return Rml::CompiledGeometryHandle(0); + } + } + + // Bind index buffer + { + + D3D11_BUFFER_DESC indexBufferDesc{}; + indexBufferDesc.Usage = D3D11_USAGE_DEFAULT; + indexBufferDesc.ByteWidth = sizeof(int) * indices.size(); + indexBufferDesc.BindFlags = D3D11_BIND_INDEX_BUFFER; + indexBufferDesc.CPUAccessFlags = 0; + indexBufferDesc.MiscFlags = 0; + indexBufferDesc.StructureByteStride = 0; + + D3D11_SUBRESOURCE_DATA indexData{}; + indexData.pSysMem = indices.data(); + indexData.SysMemPitch = 0; + indexData.SysMemSlicePitch = 0; + + HRESULT result = m_d3d_device->CreateBuffer(&indexBufferDesc, &indexData, &index_buffer); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); + if (FAILED(result)) + { + DX_CLEANUP_RESOURCE_IF_CREATED(vertex_buffer); + return Rml::CompiledGeometryHandle(0); + } + } + + DX11_GeometryData* geometryData = new DX11_GeometryData; + + geometryData->vertex_buffer = vertex_buffer; + geometryData->index_buffer = index_buffer; + geometryData->index_count = indices.size(); + + return Rml::CompiledGeometryHandle(geometryData); } -void RenderInterface_DX11::ReleaseGeometry(Rml::CompiledGeometryHandle geometry) { +void RenderInterface_DX11::ReleaseGeometry(Rml::CompiledGeometryHandle geometry) +{ + DX11_GeometryData* geometryData = (DX11_GeometryData*)geometry; + + DX_CLEANUP_RESOURCE_IF_CREATED(geometryData->vertex_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(geometryData->index_buffer); + delete reinterpret_cast(geometry); } -void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) { +void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) +{ + DX11_GeometryData* geometryData = (DX11_GeometryData*)handle; + + if (m_translation != translation) + { + m_translation = translation; + m_cbuffer_dirty = true; + } + + if (texture) + { + // Texture available + DX11_TextureData* texData = (DX11_TextureData*)texture; + m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); + m_d3d_context->PSSetShader(this->m_shader_pixel_texture, nullptr, 0); + m_d3d_context->PSSetShaderResources(0, 1, &texData->texture_view); + m_d3d_context->PSSetSamplers(0, 1, &m_samplerState); + } + else + { + // No texture, use color + m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); + m_d3d_context->PSSetShader(this->m_shader_pixel_color, nullptr, 0); + } + + UpdateConstantBuffer(); + + m_d3d_context->IASetInputLayout(m_vertex_layout); + m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); + // Bind vertex and index buffers, issue draw call + uint32_t stride = sizeof(Rml::Vertex); + uint32_t offset = 0; + m_d3d_context->IASetVertexBuffers(0, 1, &geometryData->vertex_buffer, &stride, &offset); + m_d3d_context->IASetIndexBuffer(geometryData->index_buffer, DXGI_FORMAT_R32_UINT, 0); + m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); + m_d3d_context->DrawIndexed(geometryData->index_count, 0, 0); } // Set to byte packing, or the compiler will expand our struct, which means it won't read correctly from file diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 0c073e683..8131dddb5 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -154,6 +154,14 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11Texture2D* texture = nullptr; ID3D11ShaderResourceView* texture_view = nullptr; }; + + struct DX11_GeometryData { + public: + ID3D11Buffer* vertex_buffer = nullptr; + ID3D11Buffer* index_buffer = nullptr; + uint32_t index_count = 0; + }; + // D3D11 state ID3D11BlendState* m_current_blend_state = nullptr; From 8eb491b0cf03e7ab8fd27a09307a94525d16538b Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:41:16 +0200 Subject: [PATCH 030/125] (fix): Correct invalid input layout --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 84c292124..fffe69597 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -298,8 +298,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* { D3D11_INPUT_ELEMENT_DESC polygonLayout[] = { - {"POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0}, - {"COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM_SRGB, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, + {"POSITION", 0, DXGI_FORMAT_R32G32_FLOAT, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, + {"COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, {"TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, }; uint32_t numElements = sizeof(polygonLayout) / sizeof(polygonLayout[0]); From 05e01bf88cbf18f896f79cbbd606af8b1cbb791d Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:41:32 +0200 Subject: [PATCH 031/125] (fix): Correct incorrect rasterizer state --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index fffe69597..f1d5f67cc 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -165,12 +165,12 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* { D3D11_RASTERIZER_DESC rasterizerDesc{}; rasterizerDesc.FillMode = D3D11_FILL_SOLID; - rasterizerDesc.CullMode = D3D11_CULL_BACK; + rasterizerDesc.CullMode = D3D11_CULL_NONE; rasterizerDesc.FrontCounterClockwise = false; rasterizerDesc.DepthBias = D3D11_DEFAULT_DEPTH_BIAS; rasterizerDesc.SlopeScaledDepthBias = D3D11_DEFAULT_SLOPE_SCALED_DEPTH_BIAS; rasterizerDesc.DepthBiasClamp = D3D11_DEFAULT_DEPTH_BIAS_CLAMP; - rasterizerDesc.DepthClipEnable = true; + rasterizerDesc.DepthClipEnable = false; rasterizerDesc.ScissorEnable = true; rasterizerDesc.MultisampleEnable = MSAA_SAMPLES > 1; rasterizerDesc.AntialiasedLineEnable = MSAA_SAMPLES > 1; From e5bd71e1ac951388f0694cde5a26d5401b83853c Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 00:41:47 +0200 Subject: [PATCH 032/125] (fix): Set initial state every frame --- Backends/RmlUi_Renderer_DX11.cpp | 3 +++ 1 file changed, 3 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index f1d5f67cc..5e231e985 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -399,8 +399,11 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); + m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); // Disable scissor Clear(); SetBlendState(m_blend_state); + m_d3d_context->OMSetRenderTargets(1, &p_render_target_view, nullptr); + SetTransform(nullptr); // Set no transform on new frame } void RenderInterface_DX11::EndFrame() { From fc02417967f2f02a59fa71af38ab1ac58e5facff Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 24 Sep 2024 08:51:08 +0200 Subject: [PATCH 033/125] (fix): Disable sRGB for consistency with GL3 --- Backends/RmlUi_Renderer_DX11.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 5e231e985..ddd79e232 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -659,7 +659,7 @@ Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Span Date: Tue, 24 Sep 2024 15:02:21 +0200 Subject: [PATCH 034/125] (feat): Use a hashmap to store GPU resources --- Backends/RmlUi_Renderer_DX11.cpp | 171 +++++++++++++++++++++---------- Backends/RmlUi_Renderer_DX11.h | 5 +- 2 files changed, 123 insertions(+), 53 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index ddd79e232..a0b37617b 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -117,6 +117,18 @@ float4 main(const sInputData inputArgs) : SV_TARGET } )"; +static uintptr_t HashPointer(uintptr_t inPtr) +{ + uintptr_t Value = (uintptr_t)inPtr; + Value = ~Value + (Value << 15); + Value = Value ^ (Value >> 12); + Value = Value + (Value << 2); + Value = Value ^ (Value >> 4); + Value = Value * 2057; + Value = Value ^ (Value >> 16); + return Value; +} + RenderInterface_DX11::RenderInterface_DX11() { #ifdef RMLUI_DX_DEBUG m_default_shader_flags = D3DCOMPILE_DEBUG | D3DCOMPILE_SKIP_OPTIMIZATION; @@ -292,6 +304,9 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* goto cleanup; return; } + + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); } // Create vertex layout. This will be constant to avoid copying to an intermediate struct. @@ -370,7 +385,22 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } void RenderInterface_DX11::Cleanup() { - // Cleans up all resources + + // Loop through texture cache and free all resources + for (auto& it : m_texture_cache) + { + DX_CLEANUP_RESOURCE_IF_CREATED(it.second.texture_view); + DX_CLEANUP_RESOURCE_IF_CREATED(it.second.texture); + } + + // Loop through geometry cache and free all resources + for (auto& it : m_geometry_cache) + { + DX_CLEANUP_RESOURCE_IF_CREATED(it.second.vertex_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(it.second.index_buffer); + } + + // Cleans up all general resources DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); @@ -378,6 +408,7 @@ void RenderInterface_DX11::Cleanup() { DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_color); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_texture); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); } @@ -468,64 +499,89 @@ Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Spanvertex_buffer = vertex_buffer; - geometryData->index_buffer = index_buffer; - geometryData->index_count = indices.size(); + geometryData.vertex_buffer = vertex_buffer; + geometryData.index_buffer = index_buffer; + geometryData.index_count = indices.size(); - return Rml::CompiledGeometryHandle(geometryData); + m_geometry_cache.emplace(handleId, geometryData); + return Rml::CompiledGeometryHandle(handleId); } -void RenderInterface_DX11::ReleaseGeometry(Rml::CompiledGeometryHandle geometry) +void RenderInterface_DX11::ReleaseGeometry(Rml::CompiledGeometryHandle handle) { - DX11_GeometryData* geometryData = (DX11_GeometryData*)geometry; - - DX_CLEANUP_RESOURCE_IF_CREATED(geometryData->vertex_buffer); - DX_CLEANUP_RESOURCE_IF_CREATED(geometryData->index_buffer); + if (m_geometry_cache.find(handle) != m_geometry_cache.end()) + { + DX11_GeometryData geometryData = m_geometry_cache[handle]; - delete reinterpret_cast(geometry); + DX_CLEANUP_RESOURCE_IF_CREATED(geometryData.vertex_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(geometryData.index_buffer); + + m_geometry_cache.erase(handle); + } + else + { + Rml::Log::Message(Rml::Log::LT_WARNING, "Geometry Handle %d does not exist!", handle); + } } void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) { - DX11_GeometryData* geometryData = (DX11_GeometryData*)handle; - - if (m_translation != translation) + if (m_geometry_cache.find(handle) != m_geometry_cache.end()) { - m_translation = translation; - m_cbuffer_dirty = true; - } + DX11_GeometryData geometryData = m_geometry_cache[handle]; - if (texture) - { - // Texture available - DX11_TextureData* texData = (DX11_TextureData*)texture; - m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); - m_d3d_context->PSSetShader(this->m_shader_pixel_texture, nullptr, 0); - m_d3d_context->PSSetShaderResources(0, 1, &texData->texture_view); - m_d3d_context->PSSetSamplers(0, 1, &m_samplerState); + if (m_translation != translation) + { + m_translation = translation; + m_cbuffer_dirty = true; + } + + if (texture) + { + // Texture available + if (m_texture_cache.find(texture) != m_texture_cache.end()) + { + DX11_TextureData texData = m_texture_cache[texture]; + m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); + m_d3d_context->PSSetShader(this->m_shader_pixel_texture, nullptr, 0); + m_d3d_context->PSSetShaderResources(0, 1, &texData.texture_view); + m_d3d_context->PSSetSamplers(0, 1, &m_samplerState); + } + else + { + Rml::Log::Message(Rml::Log::LT_WARNING, "Texture Handle %d does not exist!", texture); + m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); + m_d3d_context->PSSetShader(this->m_shader_pixel_color, nullptr, 0); + } + } + else + { + // No texture, use color + m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); + m_d3d_context->PSSetShader(this->m_shader_pixel_color, nullptr, 0); + } + + UpdateConstantBuffer(); + + m_d3d_context->IASetInputLayout(m_vertex_layout); + m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); + + // Bind vertex and index buffers, issue draw call + uint32_t stride = sizeof(Rml::Vertex); + uint32_t offset = 0; + m_d3d_context->IASetVertexBuffers(0, 1, &geometryData.vertex_buffer, &stride, &offset); + m_d3d_context->IASetIndexBuffer(geometryData.index_buffer, DXGI_FORMAT_R32_UINT, 0); + m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); + m_d3d_context->DrawIndexed(geometryData.index_count, 0, 0); } else { - // No texture, use color - m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); - m_d3d_context->PSSetShader(this->m_shader_pixel_color, nullptr, 0); + Rml::Log::Message(Rml::Log::LT_WARNING, "Geometry Handle %d does not exist!", handle); } - - UpdateConstantBuffer(); - - m_d3d_context->IASetInputLayout(m_vertex_layout); - m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); - - // Bind vertex and index buffers, issue draw call - uint32_t stride = sizeof(Rml::Vertex); - uint32_t offset = 0; - m_d3d_context->IASetVertexBuffers(0, 1, &geometryData->vertex_buffer, &stride, &offset); - m_d3d_context->IASetIndexBuffer(geometryData->index_buffer, DXGI_FORMAT_R32_UINT, 0); - m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); - m_d3d_context->DrawIndexed(geometryData->index_count, 0, 0); } // Set to byte packing, or the compiler will expand our struct, which means it won't read correctly from file @@ -696,23 +752,34 @@ Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Spantexture = gpu_texture; - texData->texture_view = gpu_texture_view; - // Generate mipmaps for this texture. - m_d3d_context->GenerateMips(texData->texture_view); + m_d3d_context->GenerateMips(gpu_texture_view); + + uintptr_t handleId = HashPointer((uintptr_t) gpu_texture_view); + + DX11_TextureData texData{}; + texData.texture = gpu_texture; + texData.texture_view = gpu_texture_view; - return Rml::TextureHandle(texData); + m_texture_cache.emplace(handleId, texData); + return Rml::TextureHandle(handleId); } void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) { - DX11_TextureData* texData = (DX11_TextureData*)texture_handle; - DX_CLEANUP_RESOURCE_IF_CREATED(texData->texture_view); - DX_CLEANUP_RESOURCE_IF_CREATED(texData->texture); + if (m_texture_cache.find(texture_handle) != m_texture_cache.end()) + { + DX11_TextureData texData = m_texture_cache[texture_handle]; + + DX_CLEANUP_RESOURCE_IF_CREATED(texData.texture_view); + DX_CLEANUP_RESOURCE_IF_CREATED(texData.texture); - delete reinterpret_cast(texData); + m_texture_cache.erase(texture_handle); + } + else + { + Rml::Log::Message(Rml::Log::LT_WARNING, "Texture Handle %d does not exist!", texture_handle); + } } void RenderInterface_DX11::EnableScissorRegion(bool enable) { diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 8131dddb5..c027fffab 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -82,7 +82,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { // -- Inherited from Rml::RenderInterface -- Rml::CompiledGeometryHandle CompileGeometry(Rml::Span vertices, Rml::Span indices) override; - void ReleaseGeometry(Rml::CompiledGeometryHandle geometry) override; + void ReleaseGeometry(Rml::CompiledGeometryHandle handle) override; void RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) override; Rml::TextureHandle LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) override; @@ -162,6 +162,9 @@ class RenderInterface_DX11 : public Rml::RenderInterface { uint32_t index_count = 0; }; + std::unordered_map m_texture_cache; + std::unordered_map m_geometry_cache; + // D3D11 state ID3D11BlendState* m_current_blend_state = nullptr; From a034b5db3ec6054486be5401c40f3120702e0c27 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 02:45:51 +0200 Subject: [PATCH 035/125] (fix): Remove texture cache, release texture after generating texture view Improves memory usage and effectively eliminates leaked objects. --- Backends/RmlUi_Renderer_DX11.cpp | 49 ++++++-------------------------- Backends/RmlUi_Renderer_DX11.h | 7 ----- 2 files changed, 9 insertions(+), 47 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index a0b37617b..2a34e08b7 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -386,13 +386,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* void RenderInterface_DX11::Cleanup() { - // Loop through texture cache and free all resources - for (auto& it : m_texture_cache) - { - DX_CLEANUP_RESOURCE_IF_CREATED(it.second.texture_view); - DX_CLEANUP_RESOURCE_IF_CREATED(it.second.texture); - } - // Loop through geometry cache and free all resources for (auto& it : m_geometry_cache) { @@ -543,20 +536,11 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm if (texture) { // Texture available - if (m_texture_cache.find(texture) != m_texture_cache.end()) - { - DX11_TextureData texData = m_texture_cache[texture]; - m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); - m_d3d_context->PSSetShader(this->m_shader_pixel_texture, nullptr, 0); - m_d3d_context->PSSetShaderResources(0, 1, &texData.texture_view); - m_d3d_context->PSSetSamplers(0, 1, &m_samplerState); - } - else - { - Rml::Log::Message(Rml::Log::LT_WARNING, "Texture Handle %d does not exist!", texture); - m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); - m_d3d_context->PSSetShader(this->m_shader_pixel_color, nullptr, 0); - } + ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture; + m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); + m_d3d_context->PSSetShader(this->m_shader_pixel_texture, nullptr, 0); + m_d3d_context->PSSetShaderResources(0, 1, &texture_view); + m_d3d_context->PSSetSamplers(0, 1, &m_samplerState); } else { @@ -751,35 +735,20 @@ Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::SpanGenerateMips(gpu_texture_view); uintptr_t handleId = HashPointer((uintptr_t) gpu_texture_view); - DX11_TextureData texData{}; - texData.texture = gpu_texture; - texData.texture_view = gpu_texture_view; - - m_texture_cache.emplace(handleId, texData); - return Rml::TextureHandle(handleId); + return Rml::TextureHandle(gpu_texture_view); } void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) { - if (m_texture_cache.find(texture_handle) != m_texture_cache.end()) - { - DX11_TextureData texData = m_texture_cache[texture_handle]; - - DX_CLEANUP_RESOURCE_IF_CREATED(texData.texture_view); - DX_CLEANUP_RESOURCE_IF_CREATED(texData.texture); - - m_texture_cache.erase(texture_handle); - } - else - { - Rml::Log::Message(Rml::Log::LT_WARNING, "Texture Handle %d does not exist!", texture_handle); - } + ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture_handle; + DX_CLEANUP_RESOURCE_IF_CREATED(texture_view); } void RenderInterface_DX11::EnableScissorRegion(bool enable) { diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index c027fffab..c2dc9ba99 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -149,12 +149,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { D3D11_RECT m_rect_scissor = {}; bool m_scissor_region_enabled = false; - struct DX11_TextureData { - public: - ID3D11Texture2D* texture = nullptr; - ID3D11ShaderResourceView* texture_view = nullptr; - }; - struct DX11_GeometryData { public: ID3D11Buffer* vertex_buffer = nullptr; @@ -162,7 +156,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { uint32_t index_count = 0; }; - std::unordered_map m_texture_cache; std::unordered_map m_geometry_cache; // D3D11 state From cc4563667942b4856e7a600c04d91db557dfc0bb Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 02:46:18 +0200 Subject: [PATCH 036/125] (fix): Change swapchain swap effect to flip discard --- Backends/RmlUi_Backend_Win32_DX11.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 4693d1100..84db64ea1 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -430,7 +430,7 @@ static bool CreateDeviceD3D(HWND hwnd) { sd.SampleDesc.Count = 1; sd.SampleDesc.Quality = 0; sd.Windowed = TRUE; - sd.SwapEffect = DXGI_SWAP_EFFECT_DISCARD; + sd.SwapEffect = DXGI_SWAP_EFFECT_FLIP_DISCARD; UINT createDeviceFlags = 0; From d85c76e901ed985e3caa54d39f0782d7f9ef8ffa Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 02:47:51 +0200 Subject: [PATCH 037/125] (fix): Correct typo in swapchain_occluded --- Backends/RmlUi_Backend_Win32_DX11.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 84db64ea1..7f4473a1d 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -120,7 +120,7 @@ struct BackendData { ID3D11Device* pd3dDevice = nullptr; ID3D11DeviceContext* pd3dDeviceContext = nullptr; IDXGISwapChain* pSwapChain = nullptr; - bool swapchain_cccluded = false; + bool swapchain_occluded = false; ID3D11RenderTargetView* pMainRenderTargetView = nullptr; } d3d_resources; @@ -273,7 +273,7 @@ void Backend::PresentFrame() // Present HRESULT hr = data->d3d_resources.pSwapChain->Present(1, 0); // Present with vsync //HRESULT hr = g_pSwapChain->Present(0, 0); // Present without vsync - data->d3d_resources.swapchain_cccluded = (hr == DXGI_STATUS_OCCLUDED); + data->d3d_resources.swapchain_occluded = (hr == DXGI_STATUS_OCCLUDED); // Optional, used to mark frames during performance profiling. RMLUI_FrameMark; From b57d1d9c19cf66fd3874de7ea29d7975468d7270 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 12:38:40 +0200 Subject: [PATCH 038/125] (feat): Add depth stencil state --- Backends/RmlUi_Renderer_DX11.cpp | 21 ++++++++++++++++++--- Backends/RmlUi_Renderer_DX11.h | 1 + 2 files changed, 19 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 2a34e08b7..9ce0e3c3a 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -375,7 +375,20 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* goto cleanup; return; } + } + // Create depth stencil state + { + D3D11_DEPTH_STENCIL_DESC desc; + ZeroMemory(&desc, sizeof(desc)); + desc.DepthEnable = false; + desc.DepthWriteMask = D3D11_DEPTH_WRITE_MASK_ALL; + desc.DepthFunc = D3D11_COMPARISON_ALWAYS; + desc.StencilEnable = false; + desc.FrontFace.StencilFailOp = desc.FrontFace.StencilDepthFailOp = desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; + desc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + desc.BackFace = desc.FrontFace; + m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state); } cleanup: @@ -394,15 +407,16 @@ void RenderInterface_DX11::Cleanup() { } // Cleans up all general resources + DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); + DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); - DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); - DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_vertex_common); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_color); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_texture); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); - DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); + DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_vertex_common); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) @@ -424,6 +438,7 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); // Disable scissor + m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state, 0); Clear(); SetBlendState(m_blend_state); m_d3d_context->OMSetRenderTargets(1, &p_render_target_view, nullptr); diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index c2dc9ba99..e34d14dd9 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -127,6 +127,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11RenderTargetView* m_bound_render_target = nullptr; ID3D11RasterizerState* m_rasterizer_state_scissor_enabled = nullptr; ID3D11RasterizerState* m_rasterizer_state_scissor_disabled = nullptr; + ID3D11DepthStencilState* m_depth_stencil_state = nullptr; // Shaders ID3D11Buffer* m_shader_buffer = nullptr; From 4d2cc40fed2241c60b1b03828e0e26bb9a357206 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 12:39:03 +0200 Subject: [PATCH 039/125] (fix): Add default blend factor to blend state --- Backends/RmlUi_Renderer_DX11.cpp | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 9ce0e3c3a..cc3441501 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -811,7 +811,8 @@ void RenderInterface_DX11::Clear() void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) { if (blendState != m_current_blend_state) { - m_d3d_context->OMSetBlendState(blendState, 0, 0xFFFFFFFF); + const float blend_factor[4] = {0.f, 0.f, 0.f, 0.f}; + m_d3d_context->OMSetBlendState(blendState, blend_factor, 0xFFFFFFFF); m_current_blend_state = blendState; } } From 7591ce275fea640698302bad5a68684f699c0cd4 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 12:39:25 +0200 Subject: [PATCH 040/125] (feat): Backup and restore DX11 state on frame begin / end --- Backends/RmlUi_Renderer_DX11.cpp | 70 ++++++++++++++++++++++++++++++++ Backends/RmlUi_Renderer_DX11.h | 25 ++++++++++-- 2 files changed, 92 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index cc3441501..e04a62d1a 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -426,6 +426,28 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT RMLUI_ASSERTMSG(m_d3d_context, "d3d_context cannot be nullptr!"); RMLUI_ASSERTMSG(m_d3d_device, "d3d_device cannot be nullptr!"); + // Backup DX11 state + m_previous_d3d_state.ScissorRectsCount = m_previous_d3d_state.ViewportsCount = D3D11_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE; + m_d3d_context->RSGetScissorRects(&m_previous_d3d_state.ScissorRectsCount, m_previous_d3d_state.ScissorRects); + m_d3d_context->RSGetViewports(&m_previous_d3d_state.ViewportsCount, m_previous_d3d_state.Viewports); + m_d3d_context->RSGetState(&m_previous_d3d_state.RS); + m_d3d_context->OMGetBlendState(&m_previous_d3d_state.BlendState, m_previous_d3d_state.BlendFactor, &m_previous_d3d_state.SampleMask); + m_d3d_context->OMGetDepthStencilState(&m_previous_d3d_state.DepthStencilState, &m_previous_d3d_state.StencilRef); + m_d3d_context->PSGetShaderResources(0, 1, &m_previous_d3d_state.PSShaderResource); + m_d3d_context->PSGetSamplers(0, 1, &m_previous_d3d_state.PSSampler); + m_previous_d3d_state.PSInstancesCount = m_previous_d3d_state.VSInstancesCount = m_previous_d3d_state.GSInstancesCount = 256; + m_d3d_context->PSGetShader(&m_previous_d3d_state.PS, m_previous_d3d_state.PSInstances, &m_previous_d3d_state.PSInstancesCount); + m_d3d_context->VSGetShader(&m_previous_d3d_state.VS, m_previous_d3d_state.VSInstances, &m_previous_d3d_state.VSInstancesCount); + m_d3d_context->VSGetConstantBuffers(0, 1, &m_previous_d3d_state.VSConstantBuffer); + m_d3d_context->GSGetShader(&m_previous_d3d_state.GS, m_previous_d3d_state.GSInstances, &m_previous_d3d_state.GSInstancesCount); + + m_d3d_context->IAGetPrimitiveTopology(&m_previous_d3d_state.PrimitiveTopology); + m_d3d_context->IAGetIndexBuffer(&m_previous_d3d_state.IndexBuffer, &m_previous_d3d_state.IndexBufferFormat, + &m_previous_d3d_state.IndexBufferOffset); + m_d3d_context->IAGetVertexBuffers(0, 1, &m_previous_d3d_state.VertexBuffer, &m_previous_d3d_state.VertexBufferStride, + &m_previous_d3d_state.VertexBufferOffset); + m_d3d_context->IAGetInputLayout(&m_previous_d3d_state.InputLayout); + m_bound_render_target = p_render_target_view; m_bound_swapchain = p_swapchain; @@ -452,6 +474,54 @@ void RenderInterface_DX11::EndFrame() { m_bound_render_target = nullptr; // Reset blend state m_current_blend_state = nullptr; + + // Restore modified DX state + m_d3d_context->RSSetScissorRects(m_previous_d3d_state.ScissorRectsCount, m_previous_d3d_state.ScissorRects); + m_d3d_context->RSSetViewports(m_previous_d3d_state.ViewportsCount, m_previous_d3d_state.Viewports); + m_d3d_context->RSSetState(m_previous_d3d_state.RS); + if (m_previous_d3d_state.RS) + m_previous_d3d_state.RS->Release(); + m_d3d_context->OMSetBlendState(m_previous_d3d_state.BlendState, m_previous_d3d_state.BlendFactor, m_previous_d3d_state.SampleMask); + if (m_previous_d3d_state.BlendState) + m_previous_d3d_state.BlendState->Release(); + m_d3d_context->OMSetDepthStencilState(m_previous_d3d_state.DepthStencilState, m_previous_d3d_state.StencilRef); + if (m_previous_d3d_state.DepthStencilState) + m_previous_d3d_state.DepthStencilState->Release(); + m_d3d_context->PSSetShaderResources(0, 1, &m_previous_d3d_state.PSShaderResource); + if (m_previous_d3d_state.PSShaderResource) + m_previous_d3d_state.PSShaderResource->Release(); + m_d3d_context->PSSetSamplers(0, 1, &m_previous_d3d_state.PSSampler); + if (m_previous_d3d_state.PSSampler) + m_previous_d3d_state.PSSampler->Release(); + m_d3d_context->PSSetShader(m_previous_d3d_state.PS, m_previous_d3d_state.PSInstances, m_previous_d3d_state.PSInstancesCount); + if (m_previous_d3d_state.PS) + m_previous_d3d_state.PS->Release(); + for (UINT i = 0; i < m_previous_d3d_state.PSInstancesCount; i++) + if (m_previous_d3d_state.PSInstances[i]) + m_previous_d3d_state.PSInstances[i]->Release(); + m_d3d_context->VSSetShader(m_previous_d3d_state.VS, m_previous_d3d_state.VSInstances, m_previous_d3d_state.VSInstancesCount); + if (m_previous_d3d_state.VS) + m_previous_d3d_state.VS->Release(); + m_d3d_context->VSSetConstantBuffers(0, 1, &m_previous_d3d_state.VSConstantBuffer); + if (m_previous_d3d_state.VSConstantBuffer) + m_previous_d3d_state.VSConstantBuffer->Release(); + m_d3d_context->GSSetShader(m_previous_d3d_state.GS, m_previous_d3d_state.GSInstances, m_previous_d3d_state.GSInstancesCount); + if (m_previous_d3d_state.GS) + m_previous_d3d_state.GS->Release(); + for (UINT i = 0; i < m_previous_d3d_state.VSInstancesCount; i++) + if (m_previous_d3d_state.VSInstances[i]) + m_previous_d3d_state.VSInstances[i]->Release(); + m_d3d_context->IASetPrimitiveTopology(m_previous_d3d_state.PrimitiveTopology); + m_d3d_context->IASetIndexBuffer(m_previous_d3d_state.IndexBuffer, m_previous_d3d_state.IndexBufferFormat, m_previous_d3d_state.IndexBufferOffset); + if (m_previous_d3d_state.IndexBuffer) + m_previous_d3d_state.IndexBuffer->Release(); + m_d3d_context->IASetVertexBuffers(0, 1, &m_previous_d3d_state.VertexBuffer, &m_previous_d3d_state.VertexBufferStride, + &m_previous_d3d_state.VertexBufferOffset); + if (m_previous_d3d_state.VertexBuffer) + m_previous_d3d_state.VertexBuffer->Release(); + m_d3d_context->IASetInputLayout(m_previous_d3d_state.InputLayout); + if (m_previous_d3d_state.InputLayout) + m_previous_d3d_state.InputLayout->Release(); } diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index e34d14dd9..b24bab553 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -163,11 +163,30 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11BlendState* m_current_blend_state = nullptr; struct D3D11State { - - ID3D11BlendState* m_previousBlendState = nullptr; + UINT ScissorRectsCount, ViewportsCount; + D3D11_RECT ScissorRects[D3D11_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE]; + D3D11_VIEWPORT Viewports[D3D11_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE]; + ID3D11RasterizerState* RS; + ID3D11BlendState* BlendState; + FLOAT BlendFactor[4]; + UINT SampleMask; + UINT StencilRef; + ID3D11DepthStencilState* DepthStencilState; + ID3D11ShaderResourceView* PSShaderResource; + ID3D11SamplerState* PSSampler; + ID3D11PixelShader* PS; + ID3D11VertexShader* VS; + ID3D11GeometryShader* GS; + UINT PSInstancesCount, VSInstancesCount, GSInstancesCount; + ID3D11ClassInstance *PSInstances[256], *VSInstances[256], *GSInstances[256]; // 256 is max according to PSSetShader documentation + D3D11_PRIMITIVE_TOPOLOGY PrimitiveTopology; + ID3D11Buffer *IndexBuffer, *VertexBuffer, *VSConstantBuffer; + UINT IndexBufferOffset, VertexBufferStride, VertexBufferOffset; + DXGI_FORMAT IndexBufferFormat; + ID3D11InputLayout* InputLayout; }; - D3D11State m_previousD3DState{}; + D3D11State m_previous_d3d_state{}; #pragma pack(4) struct ShaderCbuffer { From 16b0120da2bbb6bef0f4385cafc2ad736549cd50 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 12:51:16 +0200 Subject: [PATCH 041/125] (chore): Remove redundant struct --- Backends/RmlUi_Renderer_DX11.h | 6 ------ 1 file changed, 6 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index b24bab553..5821c5e69 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -112,12 +112,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void UpdateConstantBuffer(); private: - // @TODO: Replace with vertex / index buffer pairs - struct GeometryView { - Rml::Span vertices; - Rml::Span indices; - }; - UINT m_default_shader_flags; // D3D11 core resources From d6dbabbf6c49a6f531f40e77e22328316feca29c Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 14:47:33 +0200 Subject: [PATCH 042/125] (chore): Use snake case --- Backends/RmlUi_Renderer_DX11.cpp | 174 +++++++++++++++++++------------ Backends/RmlUi_Renderer_DX11.h | 43 ++++---- 2 files changed, 129 insertions(+), 88 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e04a62d1a..ab29f2f14 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -427,30 +427,34 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT RMLUI_ASSERTMSG(m_d3d_device, "d3d_device cannot be nullptr!"); // Backup DX11 state - m_previous_d3d_state.ScissorRectsCount = m_previous_d3d_state.ViewportsCount = D3D11_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE; - m_d3d_context->RSGetScissorRects(&m_previous_d3d_state.ScissorRectsCount, m_previous_d3d_state.ScissorRects); - m_d3d_context->RSGetViewports(&m_previous_d3d_state.ViewportsCount, m_previous_d3d_state.Viewports); - m_d3d_context->RSGetState(&m_previous_d3d_state.RS); - m_d3d_context->OMGetBlendState(&m_previous_d3d_state.BlendState, m_previous_d3d_state.BlendFactor, &m_previous_d3d_state.SampleMask); - m_d3d_context->OMGetDepthStencilState(&m_previous_d3d_state.DepthStencilState, &m_previous_d3d_state.StencilRef); - m_d3d_context->PSGetShaderResources(0, 1, &m_previous_d3d_state.PSShaderResource); - m_d3d_context->PSGetSamplers(0, 1, &m_previous_d3d_state.PSSampler); - m_previous_d3d_state.PSInstancesCount = m_previous_d3d_state.VSInstancesCount = m_previous_d3d_state.GSInstancesCount = 256; - m_d3d_context->PSGetShader(&m_previous_d3d_state.PS, m_previous_d3d_state.PSInstances, &m_previous_d3d_state.PSInstancesCount); - m_d3d_context->VSGetShader(&m_previous_d3d_state.VS, m_previous_d3d_state.VSInstances, &m_previous_d3d_state.VSInstancesCount); - m_d3d_context->VSGetConstantBuffers(0, 1, &m_previous_d3d_state.VSConstantBuffer); - m_d3d_context->GSGetShader(&m_previous_d3d_state.GS, m_previous_d3d_state.GSInstances, &m_previous_d3d_state.GSInstancesCount); - - m_d3d_context->IAGetPrimitiveTopology(&m_previous_d3d_state.PrimitiveTopology); - m_d3d_context->IAGetIndexBuffer(&m_previous_d3d_state.IndexBuffer, &m_previous_d3d_state.IndexBufferFormat, - &m_previous_d3d_state.IndexBufferOffset); - m_d3d_context->IAGetVertexBuffers(0, 1, &m_previous_d3d_state.VertexBuffer, &m_previous_d3d_state.VertexBufferStride, - &m_previous_d3d_state.VertexBufferOffset); - m_d3d_context->IAGetInputLayout(&m_previous_d3d_state.InputLayout); + { + m_previous_d3d_state.scissor_rects_count = m_previous_d3d_state.viewports_count = D3D11_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE; + m_d3d_context->RSGetScissorRects(&m_previous_d3d_state.scissor_rects_count, m_previous_d3d_state.scissor_rects); + m_d3d_context->RSGetViewports(&m_previous_d3d_state.viewports_count, m_previous_d3d_state.viewports); + m_d3d_context->RSGetState(&m_previous_d3d_state.rastizer_state); + m_d3d_context->OMGetBlendState(&m_previous_d3d_state.blend_state, m_previous_d3d_state.blend_factor, &m_previous_d3d_state.sample_mask); + m_d3d_context->OMGetDepthStencilState(&m_previous_d3d_state.depth_stencil_state, &m_previous_d3d_state.stencil_ref); + m_d3d_context->PSGetShaderResources(0, 1, &m_previous_d3d_state.pixel_shader_shader_resource); + m_d3d_context->PSGetSamplers(0, 1, &m_previous_d3d_state.pixel_shader_sampler); + m_previous_d3d_state.pixel_shader_instances_count = m_previous_d3d_state.vertex_shader_instances_count = m_previous_d3d_state.geometry_shader_instances_count = 256; + m_d3d_context->PSGetShader(&m_previous_d3d_state.pixel_shader, m_previous_d3d_state.pixel_shader_instances, &m_previous_d3d_state.pixel_shader_instances_count); + m_d3d_context->VSGetShader(&m_previous_d3d_state.vertex_shader, m_previous_d3d_state.vertex_shader_instances, + &m_previous_d3d_state.vertex_shader_instances_count); + m_d3d_context->VSGetConstantBuffers(0, 1, &m_previous_d3d_state.vertex_shader_constant_buffer); + m_d3d_context->GSGetShader(&m_previous_d3d_state.geometry_shader, m_previous_d3d_state.geometry_shader_instances, &m_previous_d3d_state.geometry_shader_instances_count); + + m_d3d_context->IAGetPrimitiveTopology(&m_previous_d3d_state.primitive_topology); + m_d3d_context->IAGetIndexBuffer(&m_previous_d3d_state.index_buffer, &m_previous_d3d_state.index_buffer_format, + &m_previous_d3d_state.index_buffer_offset); + m_d3d_context->IAGetVertexBuffers(0, 1, &m_previous_d3d_state.vertex_buffer, &m_previous_d3d_state.vertex_buffer_stride, + &m_previous_d3d_state.vertex_buffer_offset); + m_d3d_context->IAGetInputLayout(&m_previous_d3d_state.input_layout); + } m_bound_render_target = p_render_target_view; m_bound_swapchain = p_swapchain; + // Initialise DX11 state for RmlUi D3D11_VIEWPORT d3dviewport; d3dviewport.TopLeftX = 0; d3dviewport.TopLeftY = 0; @@ -476,52 +480,89 @@ void RenderInterface_DX11::EndFrame() { m_current_blend_state = nullptr; // Restore modified DX state - m_d3d_context->RSSetScissorRects(m_previous_d3d_state.ScissorRectsCount, m_previous_d3d_state.ScissorRects); - m_d3d_context->RSSetViewports(m_previous_d3d_state.ViewportsCount, m_previous_d3d_state.Viewports); - m_d3d_context->RSSetState(m_previous_d3d_state.RS); - if (m_previous_d3d_state.RS) - m_previous_d3d_state.RS->Release(); - m_d3d_context->OMSetBlendState(m_previous_d3d_state.BlendState, m_previous_d3d_state.BlendFactor, m_previous_d3d_state.SampleMask); - if (m_previous_d3d_state.BlendState) - m_previous_d3d_state.BlendState->Release(); - m_d3d_context->OMSetDepthStencilState(m_previous_d3d_state.DepthStencilState, m_previous_d3d_state.StencilRef); - if (m_previous_d3d_state.DepthStencilState) - m_previous_d3d_state.DepthStencilState->Release(); - m_d3d_context->PSSetShaderResources(0, 1, &m_previous_d3d_state.PSShaderResource); - if (m_previous_d3d_state.PSShaderResource) - m_previous_d3d_state.PSShaderResource->Release(); - m_d3d_context->PSSetSamplers(0, 1, &m_previous_d3d_state.PSSampler); - if (m_previous_d3d_state.PSSampler) - m_previous_d3d_state.PSSampler->Release(); - m_d3d_context->PSSetShader(m_previous_d3d_state.PS, m_previous_d3d_state.PSInstances, m_previous_d3d_state.PSInstancesCount); - if (m_previous_d3d_state.PS) - m_previous_d3d_state.PS->Release(); - for (UINT i = 0; i < m_previous_d3d_state.PSInstancesCount; i++) - if (m_previous_d3d_state.PSInstances[i]) - m_previous_d3d_state.PSInstances[i]->Release(); - m_d3d_context->VSSetShader(m_previous_d3d_state.VS, m_previous_d3d_state.VSInstances, m_previous_d3d_state.VSInstancesCount); - if (m_previous_d3d_state.VS) - m_previous_d3d_state.VS->Release(); - m_d3d_context->VSSetConstantBuffers(0, 1, &m_previous_d3d_state.VSConstantBuffer); - if (m_previous_d3d_state.VSConstantBuffer) - m_previous_d3d_state.VSConstantBuffer->Release(); - m_d3d_context->GSSetShader(m_previous_d3d_state.GS, m_previous_d3d_state.GSInstances, m_previous_d3d_state.GSInstancesCount); - if (m_previous_d3d_state.GS) - m_previous_d3d_state.GS->Release(); - for (UINT i = 0; i < m_previous_d3d_state.VSInstancesCount; i++) - if (m_previous_d3d_state.VSInstances[i]) - m_previous_d3d_state.VSInstances[i]->Release(); - m_d3d_context->IASetPrimitiveTopology(m_previous_d3d_state.PrimitiveTopology); - m_d3d_context->IASetIndexBuffer(m_previous_d3d_state.IndexBuffer, m_previous_d3d_state.IndexBufferFormat, m_previous_d3d_state.IndexBufferOffset); - if (m_previous_d3d_state.IndexBuffer) - m_previous_d3d_state.IndexBuffer->Release(); - m_d3d_context->IASetVertexBuffers(0, 1, &m_previous_d3d_state.VertexBuffer, &m_previous_d3d_state.VertexBufferStride, - &m_previous_d3d_state.VertexBufferOffset); - if (m_previous_d3d_state.VertexBuffer) - m_previous_d3d_state.VertexBuffer->Release(); - m_d3d_context->IASetInputLayout(m_previous_d3d_state.InputLayout); - if (m_previous_d3d_state.InputLayout) - m_previous_d3d_state.InputLayout->Release(); + { + m_d3d_context->RSSetScissorRects(m_previous_d3d_state.scissor_rects_count, m_previous_d3d_state.scissor_rects); + m_d3d_context->RSSetViewports(m_previous_d3d_state.viewports_count, m_previous_d3d_state.viewports); + m_d3d_context->RSSetState(m_previous_d3d_state.rastizer_state); + if (m_previous_d3d_state.rastizer_state) + { + m_previous_d3d_state.rastizer_state->Release(); + } + m_d3d_context->OMSetBlendState(m_previous_d3d_state.blend_state, m_previous_d3d_state.blend_factor, m_previous_d3d_state.sample_mask); + if (m_previous_d3d_state.blend_state) + { + m_previous_d3d_state.blend_state->Release(); + } + m_d3d_context->OMSetDepthStencilState(m_previous_d3d_state.depth_stencil_state, m_previous_d3d_state.stencil_ref); + if (m_previous_d3d_state.depth_stencil_state) + { + m_previous_d3d_state.depth_stencil_state->Release(); + } + m_d3d_context->PSSetShaderResources(0, 1, &m_previous_d3d_state.pixel_shader_shader_resource); + if (m_previous_d3d_state.pixel_shader_shader_resource) + { + m_previous_d3d_state.pixel_shader_shader_resource->Release(); + } + m_d3d_context->PSSetSamplers(0, 1, &m_previous_d3d_state.pixel_shader_sampler); + if (m_previous_d3d_state.pixel_shader_sampler) + { + m_previous_d3d_state.pixel_shader_sampler->Release(); + } + m_d3d_context->PSSetShader(m_previous_d3d_state.pixel_shader, m_previous_d3d_state.pixel_shader_instances, m_previous_d3d_state.pixel_shader_instances_count); + if (m_previous_d3d_state.pixel_shader) + { + m_previous_d3d_state.pixel_shader->Release(); + } + for (UINT i = 0; i < m_previous_d3d_state.pixel_shader_instances_count; i++) + { + if (m_previous_d3d_state.pixel_shader_instances[i]) + { + m_previous_d3d_state.pixel_shader_instances[i]->Release(); + } + } + m_d3d_context->VSSetShader(m_previous_d3d_state.vertex_shader, m_previous_d3d_state.vertex_shader_instances, + m_previous_d3d_state.vertex_shader_instances_count); + if (m_previous_d3d_state.vertex_shader) + { + m_previous_d3d_state.vertex_shader->Release(); + } + m_d3d_context->VSSetConstantBuffers(0, 1, &m_previous_d3d_state.vertex_shader_constant_buffer); + if (m_previous_d3d_state.vertex_shader_constant_buffer) + { + m_previous_d3d_state.vertex_shader_constant_buffer->Release(); + } + m_d3d_context->GSSetShader(m_previous_d3d_state.geometry_shader, m_previous_d3d_state.geometry_shader_instances, + m_previous_d3d_state.geometry_shader_instances_count); + if (m_previous_d3d_state.geometry_shader) + { + m_previous_d3d_state.geometry_shader->Release(); + } + for (UINT i = 0; i < m_previous_d3d_state.vertex_shader_instances_count; i++) + { + if (m_previous_d3d_state.vertex_shader_instances[i]) + { + m_previous_d3d_state.vertex_shader_instances[i]->Release(); + } + } + m_d3d_context->IASetPrimitiveTopology(m_previous_d3d_state.primitive_topology); + m_d3d_context->IASetIndexBuffer(m_previous_d3d_state.index_buffer, m_previous_d3d_state.index_buffer_format, + m_previous_d3d_state.index_buffer_offset); + if (m_previous_d3d_state.index_buffer) + { + m_previous_d3d_state.index_buffer->Release(); + } + m_d3d_context->IASetVertexBuffers(0, 1, &m_previous_d3d_state.vertex_buffer, &m_previous_d3d_state.vertex_buffer_stride, + &m_previous_d3d_state.vertex_buffer_offset); + if (m_previous_d3d_state.vertex_buffer) + { + m_previous_d3d_state.vertex_buffer->Release(); + } + m_d3d_context->IASetInputLayout(m_previous_d3d_state.input_layout); + if (m_previous_d3d_state.input_layout) + { + m_previous_d3d_state.input_layout->Release(); + } + } } @@ -555,7 +596,6 @@ Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span Date: Wed, 25 Sep 2024 15:39:53 +0200 Subject: [PATCH 043/125] (feat): Add support for window resizing --- Backends/RmlUi_Backend_Win32_DX11.cpp | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 7f4473a1d..a83312f28 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -126,6 +126,7 @@ struct BackendData { bool context_dimensions_dirty = true; Rml::Vector2i window_dimensions; + Rml::Vector2i resize_dimensions; bool running = true; // Arguments set during event processing and nulled otherwise. @@ -261,7 +262,16 @@ void Backend::RequestExit() void Backend::BeginFrame() { RMLUI_ASSERT(data); - // @TODO: Check if should resize, if so, re-create swapchain + + // Handle window resize (we don't resize directly in the WM_SIZE handler) + if (data->resize_dimensions.x != 0 && data->resize_dimensions.y != 0) + { + CleanupRenderTarget(); + data->d3d_resources.pSwapChain->ResizeBuffers(0, data->resize_dimensions.x, data->resize_dimensions.y, DXGI_FORMAT_UNKNOWN, 0); + data->resize_dimensions.x = data->resize_dimensions.y = 0; + CreateRenderTarget(); + } + data->render_interface.BeginFrame(data->d3d_resources.pSwapChain, data->d3d_resources.pMainRenderTargetView); } @@ -298,6 +308,9 @@ static LRESULT CALLBACK WindowProcedureHandler(HWND window_handle, UINT message, const int height = HIWORD(l_param); data->window_dimensions.x = width; data->window_dimensions.y = height; + data->resize_dimensions.x = width; + data->resize_dimensions.y = height; + data->context_dimensions_dirty = true; data->render_interface.SetViewport(width, height); if (data->context) data->context->SetDimensions(data->window_dimensions); @@ -310,7 +323,7 @@ static LRESULT CALLBACK WindowProcedureHandler(HWND window_handle, UINT message, SetWindowPos(window_handle, NULL, new_pos->left, new_pos->top, new_pos->right - new_pos->left, new_pos->bottom - new_pos->top, SWP_NOZORDER | SWP_NOACTIVATE); if (data->context && has_dpi_support) - data->context->SetDensityIndependentPixelRatio(GetDensityIndependentPixelRatio(window_handle)); + data->context->SetDensityIndependentPixelRatio(GetDensityIndependentPixelRatio(window_handle)); return 0; } break; From 0160747bb8fccf5fc362cbcb260fdd89df6439ea Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 15:40:03 +0200 Subject: [PATCH 044/125] (feat): Handle Aero Snap --- Backends/RmlUi_Backend_Win32_DX11.cpp | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index a83312f28..d271d7fdb 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -296,6 +296,10 @@ static LRESULT CALLBACK WindowProcedureHandler(HWND window_handle, UINT message, switch (message) { + case WM_NCACTIVATE: + case WM_NCPAINT: + // Handle Windows Aero Snap + return DefWindowProcW(window_handle, message, w_param, l_param); case WM_CLOSE: { data->running = false; From f6dbb560b8628a0caeda4f100d6f31eff02adbf5 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 15:40:18 +0200 Subject: [PATCH 045/125] (chore): Whitelist DX11 backend for effects sample --- Samples/basic/effects/CMakeLists.txt | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Samples/basic/effects/CMakeLists.txt b/Samples/basic/effects/CMakeLists.txt index 520d18e8f..d58d06cca 100644 --- a/Samples/basic/effects/CMakeLists.txt +++ b/Samples/basic/effects/CMakeLists.txt @@ -1,5 +1,5 @@ -if(NOT RMLUI_BACKEND MATCHES "GL3$") - message(STATUS "Effects sample disabled - only available on GL3 backends") +if(NOT RMLUI_BACKEND MATCHES "GL3$|DX11$") + message(STATUS "Effects sample disabled - only available on GL3 and DX11 backends") return() endif() From aa6bbe09a7947e5137f13afd008d35e200deedf3 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 22:37:28 +0200 Subject: [PATCH 046/125] (feat): Add stb_image loader #define RMLUI_USE_STB_IMAGE_LOADER to use it, and add stb_image.h to the include dir --- Backends/RmlUi_Backend_Win32_DX11.cpp | 66 +++++++++++++++++++++++++++ 1 file changed, 66 insertions(+) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index d271d7fdb..3f9452885 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -36,6 +36,16 @@ #include #include +#ifdef RMLUI_USE_STB_IMAGE_LOADER +#define STB_IMAGE_IMPLEMENTATION +#include "stb_image.h" +#include + +// stb_image based loaders +void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize); +void FreeTexture(uint8_t* pData); +#endif + /** High DPI support using Windows Per Monitor V2 DPI awareness. @@ -157,6 +167,17 @@ bool Backend::Initialize(const char* window_name, int width, int height, bool al return false; } +#ifdef RMLUI_USE_STB_IMAGE_LOADER + // Assign stb_image loader to render interface + data->render_interface.LoadTextureFromFileRaw = &LoadTexture; + data->render_interface.FreeTextureFromFileRaw = &FreeTexture; + // Disable pre-multiplication because loading images through stb_image will inconsistently return pre-multiplied images (only iPhone PNGs are pre-multipled) + // We handle pre-multiplication manually on decode + stbi_set_unpremultiply_on_load(true); + // iPhone PNGs are encoded as BGRA, this tells stb_image to unpack as RGBA + stbi_convert_iphone_png_to_rgb(true); +#endif + data->render_interface.Init(data->d3d_resources.pd3dDevice, data->d3d_resources.pd3dDeviceContext); data->window_handle = window_handle; @@ -505,3 +526,48 @@ static void CleanupRenderTarget() data->d3d_resources.pMainRenderTargetView = nullptr; } } + +#ifdef RMLUI_USE_STB_IMAGE_LOADER +void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize) +{ + // Find where on disk the file is + Rml::FileInterface* file_interface = Rml::GetFileInterface(); + Rml::FileHandle file_handle = file_interface->Open(filename); + if (!file_handle) + { + // Tell the backend that the data is invalid (*pData = 0), and return + *pData = nullptr; + return; + } + + // Load the file through stb_image + int texture_width, texture_height, num_channels; + *pData = stbi_load_from_file((FILE*)file_handle, &texture_width, &texture_height, &num_channels, 0); + + // If the file data is correct + if (*pData != nullptr) + { + // Assign image parameters + *pWidth = texture_width; + *pHeight = texture_height; + + // Compute number of elements in texture + *pDataSize = texture_width * texture_height * num_channels; + + // Pre-multiply the data + uint8_t* dataView = *pData; + for (int i = 0; i < *pDataSize; i += 4) + { + dataView[i + 0] = (uint8_t)((dataView[i + 0] * dataView[i + 3]) / 255); + dataView[i + 1] = (uint8_t)((dataView[i + 1] * dataView[i + 3]) / 255); + dataView[i + 2] = (uint8_t)((dataView[i + 2] * dataView[i + 3]) / 255); + } + } +} + +void FreeTexture(uint8_t* pData) +{ + // Free the allocated memory once the texture has been uploaded to the GPU + stbi_image_free(pData); +} +#endif \ No newline at end of file From b24e51ae5357deaec6a2416070d75d0c924f3fac Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 25 Sep 2024 23:47:43 +0200 Subject: [PATCH 047/125] (chore): Correct code style --- Backends/RmlUi_Renderer_DX11.cpp | 57 ++++++++++++++++---------------- Backends/RmlUi_Renderer_DX11.h | 7 ++-- 2 files changed, 33 insertions(+), 31 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index ab29f2f14..1523df1d9 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -129,7 +129,8 @@ static uintptr_t HashPointer(uintptr_t inPtr) return Value; } -RenderInterface_DX11::RenderInterface_DX11() { +RenderInterface_DX11::RenderInterface_DX11() +{ #ifdef RMLUI_DX_DEBUG m_default_shader_flags = D3DCOMPILE_DEBUG | D3DCOMPILE_SKIP_OPTIMIZATION; #else @@ -184,8 +185,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* rasterizerDesc.DepthBiasClamp = D3D11_DEFAULT_DEPTH_BIAS_CLAMP; rasterizerDesc.DepthClipEnable = false; rasterizerDesc.ScissorEnable = true; - rasterizerDesc.MultisampleEnable = MSAA_SAMPLES > 1; - rasterizerDesc.AntialiasedLineEnable = MSAA_SAMPLES > 1; + rasterizerDesc.MultisampleEnable = NUM_MSAA_SAMPLES > 1; + rasterizerDesc.AntialiasedLineEnable = NUM_MSAA_SAMPLES > 1; HRESULT result = m_d3d_device->CreateRasterizerState(&rasterizerDesc, &m_rasterizer_state_scissor_enabled); RMLUI_DX_ASSERTMSG(result, "failed to CreateRasterizerState"); @@ -211,10 +212,9 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } // Compile and load shaders - + // Buffer shall be cleared up later, as it's required to create the input layout ID3DBlob* p_shader_vertex_common{}; - { const D3D_SHADER_MACRO macros[] = {"RMLUI_PREMULTIPLIED_ALPHA", nullptr, nullptr, nullptr}; ID3DBlob* p_error_buff{}; @@ -310,7 +310,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } // Create vertex layout. This will be constant to avoid copying to an intermediate struct. - { D3D11_INPUT_ELEMENT_DESC polygonLayout[] = { {"POSITION", 0, DXGI_FORMAT_R32G32_FLOAT, 0, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_VERTEX_DATA, 0}, @@ -330,7 +329,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } // Create constant buffers. This is so that we can bind uniforms such as translation and color to the shaders. - { D3D11_BUFFER_DESC cbufferDesc{}; cbufferDesc.Usage = D3D11_USAGE_DYNAMIC; @@ -351,7 +349,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Create sampler state for textures { - // Create a texture sampler state description. D3D11_SAMPLER_DESC samplerDesc{}; samplerDesc.Filter = D3D11_FILTER_MIN_MAG_MIP_LINEAR; samplerDesc.AddressU = D3D11_TEXTURE_ADDRESS_WRAP; @@ -367,7 +364,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* samplerDesc.MinLOD = 0; samplerDesc.MaxLOD = D3D11_FLOAT32_MAX; - // Create the texture sampler state. HRESULT result = m_d3d_device->CreateSamplerState(&samplerDesc, &m_samplerState); RMLUI_DX_ASSERTMSG(result, "failed to CreateSamplerState"); if (FAILED(result)) @@ -397,8 +393,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_vertex_common); } -void RenderInterface_DX11::Cleanup() { - +void RenderInterface_DX11::Cleanup() +{ // Loop through geometry cache and free all resources for (auto& it : m_geometry_cache) { @@ -436,12 +432,15 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT m_d3d_context->OMGetDepthStencilState(&m_previous_d3d_state.depth_stencil_state, &m_previous_d3d_state.stencil_ref); m_d3d_context->PSGetShaderResources(0, 1, &m_previous_d3d_state.pixel_shader_shader_resource); m_d3d_context->PSGetSamplers(0, 1, &m_previous_d3d_state.pixel_shader_sampler); - m_previous_d3d_state.pixel_shader_instances_count = m_previous_d3d_state.vertex_shader_instances_count = m_previous_d3d_state.geometry_shader_instances_count = 256; - m_d3d_context->PSGetShader(&m_previous_d3d_state.pixel_shader, m_previous_d3d_state.pixel_shader_instances, &m_previous_d3d_state.pixel_shader_instances_count); + m_previous_d3d_state.pixel_shader_instances_count = m_previous_d3d_state.vertex_shader_instances_count = + m_previous_d3d_state.geometry_shader_instances_count = 256; + m_d3d_context->PSGetShader(&m_previous_d3d_state.pixel_shader, m_previous_d3d_state.pixel_shader_instances, + &m_previous_d3d_state.pixel_shader_instances_count); m_d3d_context->VSGetShader(&m_previous_d3d_state.vertex_shader, m_previous_d3d_state.vertex_shader_instances, &m_previous_d3d_state.vertex_shader_instances_count); m_d3d_context->VSGetConstantBuffers(0, 1, &m_previous_d3d_state.vertex_shader_constant_buffer); - m_d3d_context->GSGetShader(&m_previous_d3d_state.geometry_shader, m_previous_d3d_state.geometry_shader_instances, &m_previous_d3d_state.geometry_shader_instances_count); + m_d3d_context->GSGetShader(&m_previous_d3d_state.geometry_shader, m_previous_d3d_state.geometry_shader_instances, + &m_previous_d3d_state.geometry_shader_instances_count); m_d3d_context->IAGetPrimitiveTopology(&m_previous_d3d_state.primitive_topology); m_d3d_context->IAGetIndexBuffer(&m_previous_d3d_state.index_buffer, &m_previous_d3d_state.index_buffer_format, @@ -467,11 +466,12 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state, 0); Clear(); SetBlendState(m_blend_state); - m_d3d_context->OMSetRenderTargets(1, &p_render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &m_bound_render_target, nullptr); SetTransform(nullptr); // Set no transform on new frame } -void RenderInterface_DX11::EndFrame() { +void RenderInterface_DX11::EndFrame() +{ // @TODO: Compositing // @TODO: Layer stack m_bound_swapchain = nullptr; @@ -508,7 +508,8 @@ void RenderInterface_DX11::EndFrame() { { m_previous_d3d_state.pixel_shader_sampler->Release(); } - m_d3d_context->PSSetShader(m_previous_d3d_state.pixel_shader, m_previous_d3d_state.pixel_shader_instances, m_previous_d3d_state.pixel_shader_instances_count); + m_d3d_context->PSSetShader(m_previous_d3d_state.pixel_shader, m_previous_d3d_state.pixel_shader_instances, + m_previous_d3d_state.pixel_shader_instances_count); if (m_previous_d3d_state.pixel_shader) { m_previous_d3d_state.pixel_shader->Release(); @@ -565,9 +566,8 @@ void RenderInterface_DX11::EndFrame() { } } - -Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span vertices, Rml::Span indices) { - +Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span vertices, Rml::Span indices) +{ ID3D11Buffer* vertex_buffer = nullptr; ID3D11Buffer* index_buffer = nullptr; @@ -617,8 +617,8 @@ Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::SpanGenerateMips(gpu_texture_view); - uintptr_t handleId = HashPointer((uintptr_t) gpu_texture_view); + uintptr_t handleId = HashPointer((uintptr_t)gpu_texture_view); return Rml::TextureHandle(gpu_texture_view); } @@ -876,8 +876,8 @@ void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) DX_CLEANUP_RESOURCE_IF_CREATED(texture_view); } -void RenderInterface_DX11::EnableScissorRegion(bool enable) { - +void RenderInterface_DX11::EnableScissorRegion(bool enable) +{ if (enable != m_scissor_region_enabled) { if (enable) @@ -918,7 +918,8 @@ void RenderInterface_DX11::Clear() m_d3d_context->ClearRenderTargetView(m_bound_render_target, clearColor); } -void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) { +void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) +{ if (blendState != m_current_blend_state) { const float blend_factor[4] = {0.f, 0.f, 0.f, 0.f}; @@ -933,8 +934,8 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) m_cbuffer_dirty = true; } -void RenderInterface_DX11::UpdateConstantBuffer() { - +void RenderInterface_DX11::UpdateConstantBuffer() +{ if (m_cbuffer_dirty) { D3D11_MAPPED_SUBRESOURCE mappedResource{}; diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 6368f4ac9..86e913127 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -55,8 +55,8 @@ } // Allow the user to override the number of MSAA samples -#ifndef MSAA_SAMPLES -#define MSAA_SAMPLES 2 +#ifndef NUM_MSAA_SAMPLES + #define NUM_MSAA_SAMPLES 2 #endif // Use these typedefs to overload the default implementation with another image loader so that you can handle more than uncompressed TGA if you wish. @@ -153,9 +153,10 @@ class RenderInterface_DX11 : public Rml::RenderInterface { std::unordered_map m_geometry_cache; - // D3D11 state + // D3D11 state for RmlUi rendering ID3D11BlendState* m_current_blend_state = nullptr; + // Backup of prior D3D11 state struct D3D11State { UINT scissor_rects_count, viewports_count; D3D11_RECT scissor_rects[D3D11_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE]; From 2d7b7a0b91aa20a759af1ac4ed731202223eedc6 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 01:06:26 +0200 Subject: [PATCH 048/125] (feat): Implement layer system Note: Not finished, we still need shaders and filters and compositing --- Backends/RmlUi_Renderer_DX11.cpp | 575 +++++++++++++++++++++++++++++-- Backends/RmlUi_Renderer_DX11.h | 85 ++++- 2 files changed, 619 insertions(+), 41 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 1523df1d9..3fc0b4353 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -32,6 +32,8 @@ #include "RmlUi_Renderer_DX11.h" #include + #include + #include #include #include @@ -44,7 +46,7 @@ #endif // Shader source code -constexpr const char pShaderSourceText_Color[] = R"( +constexpr const char shader_source_text_color[] = R"( struct sInputData { float4 inputPos : SV_Position; @@ -57,7 +59,8 @@ float4 main(const sInputData inputArgs) : SV_TARGET return inputArgs.inputColor; } )"; -constexpr const char pShaderSourceText_Vertex[] = R"( + +constexpr const char shader_source_text_vertex[] = R"( struct sInputData { float2 inPosition : POSITION; @@ -99,7 +102,7 @@ sOutputData main(const sInputData inArgs) }; )"; -constexpr const char pShaderSourceText_Texture[] = R"( +constexpr const char shader_source_text_texture[] = R"( struct sInputData { float4 inputPos : SV_Position; @@ -117,6 +120,209 @@ float4 main(const sInputData inputArgs) : SV_TARGET } )"; +constexpr const char shader_passthrough_source_vertex[] = R"( +struct sInputData +{ + float2 inPosition : POSITION; + float4 inColor : COLOR; + float2 inTexCoord : TEXCOORD; +}; + +struct sOutputData +{ + float4 outPosition : SV_Position; + float4 outColor : COLOR; + float2 outUV : TEXCOORD; +}; + +cbuffer ConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_translate; + float2 _padding; +}; + +sOutputData main(const sInputData inArgs) +{ + sOutputData result; + + result.outPosition = float4(inArgs.inPosition, 0.0, 1.0); + result.outColor = inArgs.inColor; + result.outUV = inArgs.inTexCoord; + + return result; +}; +)"; + +constexpr const char shader_passthrough_source_fragment[] = R"( +struct sInputData +{ + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +Texture2D g_InputTexture : register(t0); +SamplerState g_SamplerLinear : register(s0); + + +float4 main(const sInputData inputArgs) : SV_TARGET +{ + return g_InputTexture.Sample(g_SamplerLinear, inputArgs.inputUV); +} +)"; + +namespace Gfx { +struct RenderTargetData { + int width = 0, height = 0; + ID3D11RenderTargetView* render_target_view = nullptr; // To write to colour attachment buffer + ID3D11DepthStencilView* depth_stencil_view = nullptr; // To write to stencil buffer + ID3D11Texture2D* render_target_texture = nullptr; // For MSAA resolve + ID3D11ShaderResourceView* render_target_shader_resource_view = nullptr; // To blit + bool owns_depth_stencil_buffer = false; +}; + +enum class RenderTargetAttachment { None, Depth, DepthStencil }; + +static bool CreateRenderTarget(ID3D11Device* p_device, RenderTargetData& out_rt, int width, int height, int samples, + RenderTargetAttachment attachment, ID3D11DepthStencilView* shared_depth_stencil_buffer) +{ + // Generate render target + D3D11_TEXTURE2D_DESC texture_desc = {}; + ZeroMemory(&texture_desc, sizeof(D3D11_TEXTURE2D_DESC)); + texture_desc.Width = width; + texture_desc.Height = height; + texture_desc.MipLevels = 1; + texture_desc.ArraySize = 1; + texture_desc.Format = DXGI_FORMAT_R8G8B8A8_UNORM; + texture_desc.SampleDesc.Count = (samples > 0) ? samples : 1; // MSAA + texture_desc.SampleDesc.Quality = 0; + texture_desc.Usage = D3D11_USAGE_DEFAULT; + texture_desc.BindFlags = D3D11_BIND_RENDER_TARGET | D3D11_BIND_SHADER_RESOURCE; + texture_desc.CPUAccessFlags = 0; + texture_desc.MiscFlags = 0; + + ID3D11Texture2D* rt_texture = nullptr; + HRESULT result = p_device->CreateTexture2D(&texture_desc, nullptr, &rt_texture); + RMLUI_DX_ASSERTMSG(result, "failed to CreateTexture2D"); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateTexture2D (%d)", result); + return false; + } + + D3D11_RENDER_TARGET_VIEW_DESC render_target_view_desc = {}; + render_target_view_desc.Format = texture_desc.Format; + render_target_view_desc.ViewDimension = samples > 0 ? D3D11_RTV_DIMENSION_TEXTURE2DMS : D3D11_RTV_DIMENSION_TEXTURE2D; + render_target_view_desc.Texture2D.MipSlice = 0; + + ID3D11RenderTargetView* render_target_view = nullptr; + result = p_device->CreateRenderTargetView(rt_texture, &render_target_view_desc, &render_target_view); + RMLUI_DX_ASSERTMSG(result, "failed to CreateRenderTargetView"); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateRenderTargetView (%d)", result); + return false; + } + + D3D11_SHADER_RESOURCE_VIEW_DESC render_target_shader_resource_view_desc = {}; + render_target_shader_resource_view_desc.Format = texture_desc.Format; + render_target_shader_resource_view_desc.ViewDimension = samples > 0 ? D3D11_SRV_DIMENSION_TEXTURE2DMS : D3D11_SRV_DIMENSION_TEXTURE2D; + render_target_shader_resource_view_desc.Texture2D.MostDetailedMip = 0; + render_target_shader_resource_view_desc.Texture2D.MipLevels = 1; + + ID3D11ShaderResourceView* render_target_shader_resource_view = nullptr; + result = p_device->CreateShaderResourceView(rt_texture, &render_target_shader_resource_view_desc, &render_target_shader_resource_view); + RMLUI_DX_ASSERTMSG(result, "failed to CreateShaderResourceView"); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateShaderResourceView (%d)", result); + return false; + } + + // Generate stencil buffer if necessary + ID3D11DepthStencilView* depth_stencil_view = nullptr; + + if (attachment != RenderTargetAttachment::None) + { + if (shared_depth_stencil_buffer) + { + // Share the depth/stencil buffer + depth_stencil_view = shared_depth_stencil_buffer; + depth_stencil_view->AddRef(); // Increase reference count since we're sharing it + } + else + { + // Create a new depth/stencil buffer + D3D11_TEXTURE2D_DESC depthDesc = {}; + depthDesc.Width = width; + depthDesc.Height = height; + depthDesc.MipLevels = 1; + depthDesc.ArraySize = 1; + depthDesc.Format = + (attachment == RenderTargetAttachment::DepthStencil) ? DXGI_FORMAT_D32_FLOAT_S8X24_UINT : DXGI_FORMAT_D24_UNORM_S8_UINT; + depthDesc.SampleDesc.Count = (samples > 0) ? samples : 1; + depthDesc.SampleDesc.Quality = 0; + depthDesc.Usage = D3D11_USAGE_DEFAULT; + depthDesc.BindFlags = D3D11_BIND_DEPTH_STENCIL; + depthDesc.CPUAccessFlags = 0; + depthDesc.MiscFlags = 0; + + ID3D11Texture2D* depth_stencil_texture = nullptr; + result = p_device->CreateTexture2D(&depthDesc, nullptr, &depth_stencil_texture); + RMLUI_DX_ASSERTMSG(result, "failed to CreateTexture2D"); + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateTexture2D (%d)", result); + return false; + } + + D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc = {}; + dsvDesc.Format = depthDesc.Format; + dsvDesc.ViewDimension = (samples > 0) ? D3D11_DSV_DIMENSION_TEXTURE2DMS : D3D11_DSV_DIMENSION_TEXTURE2D; + dsvDesc.Texture2D.MipSlice = 0; + + result = p_device->CreateDepthStencilView(depth_stencil_texture, &dsvDesc, &depth_stencil_view); + RMLUI_DX_ASSERTMSG(result, "failed to CreateDepthStencilView"); + if (FAILED(result)) + { + depth_stencil_texture->Release(); + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateDepthStencilView (%d)", result); + return false; + } + + depth_stencil_texture->Release(); + } + } + + out_rt = {}; + out_rt.width = width; + out_rt.height = height; + out_rt.render_target_view = render_target_view; + out_rt.render_target_texture = rt_texture; + out_rt.render_target_shader_resource_view = render_target_shader_resource_view; + out_rt.depth_stencil_view = depth_stencil_view; + out_rt.owns_depth_stencil_buffer = shared_depth_stencil_buffer != nullptr; + + return true; +} + +static void DestroyRenderTarget(RenderTargetData& rt) +{ + DX_CLEANUP_RESOURCE_IF_CREATED(rt.render_target_view); + DX_CLEANUP_RESOURCE_IF_CREATED(rt.render_target_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(rt.render_target_shader_resource_view); + DX_CLEANUP_RESOURCE_IF_CREATED(rt.depth_stencil_view); + rt = {}; +} + +static void BindTexture(ID3D11DeviceContext* context, const RenderTargetData& rt, uint32_t slot = 1) +{ + context->PSSetShaderResources(0, slot, &rt.render_target_shader_resource_view); +} + +} // namespace Gfx + static uintptr_t HashPointer(uintptr_t inPtr) { uintptr_t Value = (uintptr_t)inPtr; @@ -146,6 +352,12 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Assign D3D resources m_d3d_device = p_d3d_device; m_d3d_context = p_d3d_device_context; + m_render_layers.SetD3DResources(m_d3d_device); + + // Pre-cache quad for blitting + Rml::Mesh mesh; + Rml::MeshUtilities::GenerateQuad(mesh, Rml::Vector2f(-1), Rml::Vector2f(2), {}); + m_fullscreen_quad_geometry = RenderInterface_DX11::CompileGeometry(mesh.vertices, mesh.indices); // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). @@ -220,7 +432,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* ID3DBlob* p_error_buff{}; // Common vertex shader - HRESULT result = D3DCompile(pShaderSourceText_Vertex, sizeof(pShaderSourceText_Vertex), nullptr, macros, nullptr, "main", "vs_5_0", + HRESULT result = D3DCompile(shader_source_text_vertex, sizeof(shader_source_text_vertex), nullptr, macros, nullptr, "main", "vs_5_0", this->m_default_shader_flags, 0, &p_shader_vertex_common, &p_error_buff); RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); #ifdef RMLUI_DX_DEBUG @@ -235,7 +447,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Color fragment shader ID3DBlob* p_shader_color_pixel{}; - result = D3DCompile(pShaderSourceText_Color, sizeof(pShaderSourceText_Color), nullptr, macros, nullptr, "main", "ps_5_0", + result = D3DCompile(shader_source_text_color, sizeof(shader_source_text_color), nullptr, macros, nullptr, "main", "ps_5_0", this->m_default_shader_flags, 0, &p_shader_color_pixel, &p_error_buff); RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); #ifdef RMLUI_DX_DEBUG @@ -250,7 +462,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Texture fragment shader ID3DBlob* p_shader_color_texture{}; - result = D3DCompile(pShaderSourceText_Texture, sizeof(pShaderSourceText_Texture), nullptr, macros, nullptr, "main", "ps_5_0", + result = D3DCompile(shader_source_text_texture, sizeof(shader_source_text_texture), nullptr, macros, nullptr, "main", "ps_5_0", this->m_default_shader_flags, 0, &p_shader_color_texture, &p_error_buff); RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); #ifdef RMLUI_DX_DEBUG @@ -262,6 +474,36 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); + // Passthrough vertex shader + ID3DBlob* p_shader_passthrough_vertex{}; + + result = D3DCompile(shader_passthrough_source_vertex, sizeof(shader_passthrough_source_vertex), nullptr, macros, nullptr, "main", + "vs_5_0", this->m_default_shader_flags, 0, &p_shader_passthrough_vertex, &p_error_buff); + RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); + } + #endif + + DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); + + // Passthrough fragment shader + ID3DBlob* p_shader_passthrough_pixel{}; + + result = D3DCompile(shader_passthrough_source_fragment, sizeof(shader_passthrough_source_fragment), nullptr, macros, nullptr, "main", + "ps_5_0", this->m_default_shader_flags, 0, &p_shader_passthrough_pixel, &p_error_buff); + RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); + } + #endif + + DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); + // Create the shader objects result = m_d3d_device->CreateVertexShader(p_shader_vertex_common->GetBufferPointer(), p_shader_vertex_common->GetBufferSize(), nullptr, &m_shader_vertex_common); @@ -273,6 +515,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* #endif DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); goto cleanup; return; } @@ -287,6 +531,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* #endif DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); goto cleanup; return; } @@ -301,12 +547,49 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* #endif DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); + goto cleanup; + return; + } + + // Passthrough + result = m_d3d_device->CreateVertexShader(p_shader_passthrough_vertex->GetBufferPointer(), p_shader_passthrough_vertex->GetBufferSize(), + nullptr, &m_shader_passthrough_vertex); + RMLUI_DX_ASSERTMSG(result, "failed to CreateVertexShader"); + if (FAILED(result)) + { + #ifdef RMLUI_DX_DEBUG + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create vertex shader: %d", result); + #endif + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); + goto cleanup; + return; + } + + result = m_d3d_device->CreatePixelShader(p_shader_passthrough_pixel->GetBufferPointer(), p_shader_passthrough_pixel->GetBufferSize(), nullptr, + &m_shader_passthrough_fragment); + RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); + if (FAILED(result)) + { + #ifdef RMLUI_DX_DEBUG + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create pixel shader: %d", result); + #endif + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); goto cleanup; return; } DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); + DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); } // Create vertex layout. This will be constant to avoid copying to an intermediate struct. @@ -373,18 +656,38 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } } - // Create depth stencil state + // Create depth stencil states { D3D11_DEPTH_STENCIL_DESC desc; ZeroMemory(&desc, sizeof(desc)); desc.DepthEnable = false; desc.DepthWriteMask = D3D11_DEPTH_WRITE_MASK_ALL; desc.DepthFunc = D3D11_COMPARISON_ALWAYS; - desc.StencilEnable = false; desc.FrontFace.StencilFailOp = desc.FrontFace.StencilDepthFailOp = desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; desc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; desc.BackFace = desc.FrontFace; - m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state); + desc.StencilEnable = false; + // Disabled + m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_disable); + + desc.StencilEnable = true; + desc.StencilReadMask = 0xFF; + desc.StencilWriteMask = 0xFF; + desc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_REPLACE; + desc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + desc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; + desc.BackFace = desc.FrontFace; + // Set and SetInverse + m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_stencil_set); + + desc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_INCR; + desc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + desc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; + desc.BackFace = desc.FrontFace; + // Intersect + m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_stencil_intersect); } cleanup: @@ -405,9 +708,11 @@ void RenderInterface_DX11::Cleanup() // Cleans up all general resources DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); - DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state); + DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); + DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_passthrough_fragment); + DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_passthrough_vertex); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_color); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_texture); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); @@ -417,6 +722,8 @@ void RenderInterface_DX11::Cleanup() void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) { + RMLUI_ASSERT(m_viewport_width >= 1 && m_viewport_height >= 1); + RMLUI_ASSERTMSG(p_swapchain, "p_swapchain cannot be nullptr!"); RMLUI_ASSERTMSG(p_render_target_view, "p_render_target_view cannot be nullptr!"); RMLUI_ASSERTMSG(m_d3d_context, "d3d_context cannot be nullptr!"); @@ -462,21 +769,54 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); - m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); // Disable scissor - m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state, 0); - Clear(); SetBlendState(m_blend_state); + m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); // Disable scissor + m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); m_d3d_context->OMSetRenderTargets(1, &m_bound_render_target, nullptr); - SetTransform(nullptr); // Set no transform on new frame + Clear(); + + SetTransform(nullptr); + + m_render_layers.BeginFrame(m_viewport_width, m_viewport_height); + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, nullptr); + float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; + m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, clearColor); + + // UseProgram(ProgramId::None); + // program_transform_dirty.set(); + m_scissor_state = Rml::Rectanglei::MakeInvalid(); } void RenderInterface_DX11::EndFrame() { - // @TODO: Compositing - // @TODO: Layer stack + RMLUI_ASSERTMSG(m_bound_render_target, "m_bound_render_target cannot be nullptr!"); + + const Gfx::RenderTargetData& rt_active = m_render_layers.GetTopLayer(); + const Gfx::RenderTargetData& rt_postprocess = m_render_layers.GetPostprocessPrimary(); + + // Resolve MSAA to the post-process framebuffer. + m_d3d_context->ResolveSubresource(rt_postprocess.render_target_texture, 0, rt_active.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); + + // Draw to bound_render_target (usually the swapchain RTV) + m_d3d_context->OMSetRenderTargets(1, &m_bound_render_target, nullptr); + + // Assuming we have an opaque background, we can just write to it with the premultiplied alpha blend mode and we'll get the correct result. + // Instead, if we had a transparent destination that didn't use premultiplied alpha, we would need to perform a manual un-premultiplication step. + Gfx::BindTexture(m_d3d_context, rt_postprocess); + + // @TODO: UseProgram + // @TODO: Find a pattern for flipped textures + // UseProgram(ProgramId::Passthrough); + m_d3d_context->VSSetShader(this->m_shader_passthrough_vertex, nullptr, 0); + m_d3d_context->PSSetShader(this->m_shader_passthrough_fragment, nullptr, 0); + + DrawFullscreenQuad(); + + m_render_layers.EndFrame(); + + // Reset internal state m_bound_swapchain = nullptr; m_bound_render_target = nullptr; - // Reset blend state m_current_blend_state = nullptr; // Restore modified DX state @@ -658,13 +998,20 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm m_cbuffer_dirty = true; } - if (texture) + if (texture == TexturePostprocess) + { + + } + else if (texture) { // Texture available - ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture; m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); m_d3d_context->PSSetShader(this->m_shader_pixel_texture, nullptr, 0); - m_d3d_context->PSSetShaderResources(0, 1, &texture_view); + if (texture != TextureEnableWithoutBinding) + { + ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture; + m_d3d_context->PSSetShaderResources(0, 1, &texture_view); + } m_d3d_context->PSSetSamplers(0, 1, &m_samplerState); } else @@ -876,33 +1223,79 @@ void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) DX_CLEANUP_RESOURCE_IF_CREATED(texture_view); } -void RenderInterface_DX11::EnableScissorRegion(bool enable) +void RenderInterface_DX11::DrawFullscreenQuad() +{ + RenderGeometry(m_fullscreen_quad_geometry, {}, RenderInterface_DX11::TexturePostprocess); +} + +void RenderInterface_DX11::DrawFullscreenQuad(Rml::Vector2f uv_offset, Rml::Vector2f uv_scaling) { - if (enable != m_scissor_region_enabled) + Rml::Mesh mesh; + Rml::MeshUtilities::GenerateQuad(mesh, Rml::Vector2f(-1), Rml::Vector2f(2), {}); + if (uv_offset != Rml::Vector2f() || uv_scaling != Rml::Vector2f(1.f)) { - if (enable) - { + for (Rml::Vertex& vertex : mesh.vertices) + vertex.tex_coord = (vertex.tex_coord * uv_scaling) + uv_offset; + } + const Rml::CompiledGeometryHandle geometry = CompileGeometry(mesh.vertices, mesh.indices); + RenderGeometry(geometry, {}, RenderInterface_DX11::TexturePostprocess); + ReleaseGeometry(geometry); +} + +/// Flip vertical axis of the rectangle, and move its origin to the vertically opposite side of the viewport. +/// @note Changes coordinate system from RmlUi to OpenGL, or equivalently in reverse. +/// @note The Rectangle::Top and Rectangle::Bottom members will have reverse meaning in the returned rectangle. +static Rml::Rectanglei VerticallyFlipped(Rml::Rectanglei rect, int viewport_height) +{ + RMLUI_ASSERT(rect.Valid()); + Rml::Rectanglei flipped_rect = rect; + flipped_rect.p0.y = viewport_height - rect.p1.y; + flipped_rect.p1.y = viewport_height - rect.p0.y; + return flipped_rect; +} + + +void RenderInterface_DX11::SetScissor(Rml::Rectanglei region, bool vertically_flip) +{ + if (region.Valid() != m_scissor_state.Valid()) + { + if (region.Valid()) m_d3d_context->RSSetState(m_rasterizer_state_scissor_enabled); - } else - { m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); - } - m_scissor_region_enabled = enable; } -} -void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) -{ - m_rect_scissor.left = region.Left(); - m_rect_scissor.top = region.Top(); - m_rect_scissor.bottom = region.Bottom(); - m_rect_scissor.right = region.Right(); + if (region.Valid() && vertically_flip) + region = VerticallyFlipped(region, m_viewport_height); - if (m_scissor_region_enabled) + if (region.Valid() && region != m_scissor_state) { - m_d3d_context->RSSetScissorRects(1, &m_rect_scissor); + // Some render APIs don't like offscreen positions (WebGL in particular), so clamp them to the viewport. + const int x = Rml::Math::Clamp(region.Left(), 0, m_viewport_width); + const int y = Rml::Math::Clamp(m_viewport_height - region.Bottom(), 0, m_viewport_height); + + D3D11_RECT rect_scissor = {}; + rect_scissor.left = x; + rect_scissor.top = y; + rect_scissor.right = x + region.Width(); + rect_scissor.bottom = y + region.Height(); + + m_d3d_context->RSSetScissorRects(1, &rect_scissor); } + + m_scissor_state = region; +} + +void RenderInterface_DX11::EnableScissorRegion(bool enable) +{ + // Assume enable is immediately followed by a SetScissorRegion() call, and ignore it here. + if (!enable) + SetScissor(Rml::Rectanglei::MakeInvalid(), false); +} + +void RenderInterface_DX11::SetScissorRegion(Rml::Rectanglei region) +{ + SetScissor(region); } void RenderInterface_DX11::SetViewport(const int width, const int height) @@ -959,4 +1352,110 @@ void RenderInterface_DX11::UpdateConstantBuffer() } } -#endif // RMLUI_PLATFORM_WIN32 \ No newline at end of file +RenderInterface_DX11::RenderLayerStack::RenderLayerStack() +{ + rt_postprocess.resize(4); +} + +RenderInterface_DX11::RenderLayerStack::~RenderLayerStack() +{ + DestroyRenderTargets(); +} + +void RenderInterface_DX11::RenderLayerStack::SetD3DResources(ID3D11Device* device) +{ + RMLUI_ASSERTMSG(!m_d3d_device, "D3D11Device has already been set!"); + m_d3d_device = device; +} + +Rml::LayerHandle RenderInterface_DX11::RenderLayerStack::PushLayer() +{ + RMLUI_ASSERT(layers_size <= (int)rt_layers.size()); + + if (layers_size == (int)rt_layers.size()) + { + // All framebuffers should share a single stencil buffer. + ID3D11DepthStencilView* shared_depth_stencil = (rt_layers.empty() ? nullptr : rt_layers.front().depth_stencil_view); + + rt_layers.push_back(Gfx::RenderTargetData{}); + Gfx::CreateRenderTarget(m_d3d_device, rt_layers.back(), width, height, NUM_MSAA_SAMPLES, Gfx::RenderTargetAttachment::DepthStencil, shared_depth_stencil); + } + + layers_size += 1; + return GetTopLayerHandle(); +} + + +void RenderInterface_DX11::RenderLayerStack::PopLayer() +{ + RMLUI_ASSERT(layers_size > 0); + layers_size -= 1; +} + +const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::GetLayer(Rml::LayerHandle layer) const +{ + RMLUI_ASSERT((size_t)layer < (size_t)layers_size); + return rt_layers[layer]; +} + +const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::GetTopLayer() const +{ + return GetLayer(GetTopLayerHandle()); +} + +Rml::LayerHandle RenderInterface_DX11::RenderLayerStack::GetTopLayerHandle() const +{ + RMLUI_ASSERT(layers_size > 0); + return static_cast(layers_size - 1); +} + +void RenderInterface_DX11::RenderLayerStack::SwapPostprocessPrimarySecondary() +{ + std::swap(rt_postprocess[0], rt_postprocess[1]); +} + +void RenderInterface_DX11::RenderLayerStack::BeginFrame(int new_width, int new_height) +{ + RMLUI_ASSERT(layers_size == 0); + + if (new_width != width || new_height != height) + { + width = new_width; + height = new_height; + + DestroyRenderTargets(); + } + + PushLayer(); +} + +void RenderInterface_DX11::RenderLayerStack::EndFrame() +{ + RMLUI_ASSERT(layers_size == 1); + PopLayer(); +} + +void RenderInterface_DX11::RenderLayerStack::DestroyRenderTargets() +{ + RMLUI_ASSERTMSG(layers_size == 0, "Do not call this during frame rendering, that is, between BeginFrame() and EndFrame()."); + + for (Gfx::RenderTargetData& fb : rt_layers) + Gfx::DestroyRenderTarget(fb); + + rt_layers.clear(); + + for (Gfx::RenderTargetData& fb : rt_postprocess) + Gfx::DestroyRenderTarget(fb); +} + +const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::EnsureRenderTargetPostprocess(int index) +{ + RMLUI_ASSERT(index < (int)rt_postprocess.size()) + Gfx::RenderTargetData& rt = rt_postprocess[index]; + if (!rt.render_target_view) + Gfx::CreateRenderTarget(m_d3d_device, rt, width, height, 0, Gfx::RenderTargetAttachment::None, 0); + return rt; +} + +#endif // RMLUI_PLATFORM_WIN32 + diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 86e913127..a08963acd 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -64,6 +64,12 @@ typedef void (*pfnLoadTextureRaw)(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize); typedef void (*pfnFreeTextureRaw)(uint8_t* pData); +class RenderLayerStack; +namespace Gfx { +struct ProgramData; +struct RenderTargetData; +} // namespace Gfx + class RenderInterface_DX11 : public Rml::RenderInterface { public: RenderInterface_DX11(); @@ -94,6 +100,11 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void SetTransform(const Rml::Matrix4f* transform) override; + // Can be passed to RenderGeometry() to enable texture rendering without changing the bound texture. + static constexpr Rml::TextureHandle TextureEnableWithoutBinding = Rml::TextureHandle(-1); + // Can be passed to RenderGeometry() to leave the bound texture and used program unchanged. + static constexpr Rml::TextureHandle TexturePostprocess = Rml::TextureHandle(-2); + public: /// Called by the renderer when it wants to load a texture from disk. /// @param[in] pFilename A Rml string containing the file name. @@ -111,6 +122,11 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void SetBlendState(ID3D11BlendState* blendState); void UpdateConstantBuffer(); + void SetScissor(Rml::Rectanglei region, bool vertically_flip = false); + + void DrawFullscreenQuad(); + void DrawFullscreenQuad(Rml::Vector2f uv_offset, Rml::Vector2f uv_scaling = Rml::Vector2f(1.f)); + private: UINT m_default_shader_flags; @@ -121,7 +137,14 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11RenderTargetView* m_bound_render_target = nullptr; ID3D11RasterizerState* m_rasterizer_state_scissor_enabled = nullptr; ID3D11RasterizerState* m_rasterizer_state_scissor_disabled = nullptr; - ID3D11DepthStencilState* m_depth_stencil_state = nullptr; + + // Depth stencil states + bool m_is_stencil_enabled = false; + bool m_stencil_dirty = false; + uint32_t m_stencil_ref_value = 0; + ID3D11DepthStencilState* m_depth_stencil_state_disable = nullptr; // Stencil Off + ID3D11DepthStencilState* m_depth_stencil_state_stencil_set = nullptr; // Clipmask Set/Inverse + ID3D11DepthStencilState* m_depth_stencil_state_stencil_intersect = nullptr; // Clipmask Intersect. Can have several test values // Shaders ID3D11Buffer* m_shader_buffer = nullptr; @@ -130,6 +153,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11VertexShader* m_shader_vertex_common = nullptr; ID3D11PixelShader* m_shader_pixel_color = nullptr; ID3D11PixelShader* m_shader_pixel_texture = nullptr; + ID3D11VertexShader* m_shader_passthrough_vertex = nullptr; + ID3D11PixelShader* m_shader_passthrough_fragment = nullptr; // Viewport dimensions int m_viewport_width = 0; @@ -141,8 +166,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11InputLayout* m_vertex_layout = nullptr; ID3D11BlendState* m_blend_state = nullptr; - D3D11_RECT m_rect_scissor = {}; - bool m_scissor_region_enabled = false; + Rml::Rectanglei m_scissor_state = Rml::Rectanglei::MakeInvalid(); struct DX11_GeometryData { public: @@ -191,6 +215,61 @@ class RenderInterface_DX11 : public Rml::RenderInterface { float _padding[2]; }; #pragma pack() + + Rml::CompiledGeometryHandle m_fullscreen_quad_geometry = 0; + + /* + Manages render targets, including the layer stack and postprocessing render targets. + + Layers can be pushed and popped, creating new render targets as needed. Typically, geometry is rendered to the top + layer. The layer render targets may have MSAA enabled. + + Postprocessing render targets are separate from the layers, and are commonly used to apply texture-wide effects + such as filters. They are used both as input and output during rendering, and do not use MSAA. + */ + class RenderLayerStack { + public: + RenderLayerStack(); + ~RenderLayerStack(); + + void SetD3DResources(ID3D11Device* device); + + // Push a new layer. All references to previously retrieved layers are invalidated. + Rml::LayerHandle PushLayer(); + + // Pop the top layer. All references to previously retrieved layers are invalidated. + void PopLayer(); + + const Gfx::RenderTargetData& GetLayer(Rml::LayerHandle layer) const; + const Gfx::RenderTargetData& GetTopLayer() const; + Rml::LayerHandle GetTopLayerHandle() const; + + const Gfx::RenderTargetData& GetPostprocessPrimary() { return EnsureRenderTargetPostprocess(0); } + const Gfx::RenderTargetData& GetPostprocessSecondary() { return EnsureRenderTargetPostprocess(1); } + const Gfx::RenderTargetData& GetPostprocessTertiary() { return EnsureRenderTargetPostprocess(2); } + const Gfx::RenderTargetData& GetBlendMask() { return EnsureRenderTargetPostprocess(3); } + + void SwapPostprocessPrimarySecondary(); + + void BeginFrame(int new_width, int new_height); + void EndFrame(); + + private: + void DestroyRenderTargets(); + const Gfx::RenderTargetData& EnsureRenderTargetPostprocess(int index); + + int width = 0, height = 0; + + // The number of active layers is manually tracked since we re-use the render targets stored in the fb_layers stack. + int layers_size = 0; + + Rml::Vector rt_layers; + Rml::Vector rt_postprocess; + + ID3D11Device* m_d3d_device = nullptr; + }; + + RenderLayerStack m_render_layers; }; #endif // RMLUI_PLATFORM_WIN32 From a86f001b1bda28a124f31cb0cbe22dc4214abf82 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 01:17:02 +0200 Subject: [PATCH 049/125] (fix): Flip UVs for passthrough Passthrough only appears to be used for blitting render textures on GL3, so I'll flip UVs in shader (alternatively we could flip UVs in the quad, but I'll investigate this later). --- Backends/RmlUi_Renderer_DX11.cpp | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 3fc0b4353..d675c5b89 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -146,9 +146,8 @@ sOutputData main(const sInputData inArgs) { sOutputData result; - result.outPosition = float4(inArgs.inPosition, 0.0, 1.0); - result.outColor = inArgs.inColor; - result.outUV = inArgs.inTexCoord; + result.outPosition = float4(inArgs.inPosition.xy, 0.0f, 1.0f); + result.outUV = float2(inArgs.inTexCoord.x, 1.0f - inArgs.inTexCoord.y); return result; }; From a5b16d3a01987590b10d401900616857a20a3fe3 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 09:38:17 +0200 Subject: [PATCH 050/125] (fix): Release depth stencil variants --- Backends/RmlUi_Renderer_DX11.cpp | 2 ++ 1 file changed, 2 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index d675c5b89..96198e3b0 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -708,6 +708,8 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_disable); + DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_intersect); + DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_set); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_passthrough_fragment); From 29cd4cae512badcd6f20c0f127e007998d288283 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 09:38:48 +0200 Subject: [PATCH 051/125] (feat): Implement clip masks Currently broken --- Backends/RmlUi_Backend_Win32_DX11.cpp | 1 - Backends/RmlUi_Renderer_DX11.cpp | 61 +++++++++++++++++++++++++++ Backends/RmlUi_Renderer_DX11.h | 5 ++- 3 files changed, 65 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 3f9452885..92353d95f 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -470,7 +470,6 @@ static bool CreateDeviceD3D(HWND hwnd) { sd.Windowed = TRUE; sd.SwapEffect = DXGI_SWAP_EFFECT_FLIP_DISCARD; - UINT createDeviceFlags = 0; #ifdef RMLUI_DEBUG // Enable debug layer diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 96198e3b0..e094f059f 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1353,6 +1353,67 @@ void RenderInterface_DX11::UpdateConstantBuffer() } } +void RenderInterface_DX11::EnableClipMask(bool enable) +{ + if (enable != m_is_stencil_enabled) + { + m_is_stencil_enabled = enable; + if (!enable) + { + m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); + } + } +} + +void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rml::CompiledGeometryHandle geometry, Rml::Vector2f translation) +{ + RMLUI_ASSERT(m_is_stencil_enabled); + using Rml::ClipMaskOperation; + + ID3D11DepthStencilState* stencil_state = m_depth_stencil_state_disable; + + UINT stencil_test_value = 0; + switch (operation) + { + case ClipMaskOperation::Set: + { + stencil_test_value = 1; + stencil_state = m_depth_stencil_state_stencil_set; + } + break; + case ClipMaskOperation::SetInverse: + { + stencil_test_value = 0; + stencil_state = m_depth_stencil_state_stencil_set; + } + break; + case ClipMaskOperation::Intersect: + { + stencil_test_value += 1; + stencil_state = m_depth_stencil_state_stencil_intersect; + } + break; + } + + m_d3d_context->OMSetDepthStencilState(stencil_state, stencil_test_value); + + bool clear_stencil = (operation == ClipMaskOperation::Set || operation == ClipMaskOperation::SetInverse); + if (clear_stencil) + { + // Clear stencil buffer + Rml::LayerHandle layer_handle = m_render_layers.GetTopLayerHandle(); + const Gfx::RenderTargetData& rtData = m_render_layers.GetLayer(layer_handle); + m_d3d_context->ClearDepthStencilView(rtData.depth_stencil_view, D3D11_CLEAR_STENCIL, 1.0f, 0); + } + + // @TODO: Stencil buffer is inverted? + RenderGeometry(geometry, translation, {}); + + // Restore state + // @performance Is this even necessary? + m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); +} + RenderInterface_DX11::RenderLayerStack::RenderLayerStack() { rt_postprocess.resize(4); diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index a08963acd..c77a10592 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -97,7 +97,10 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void EnableScissorRegion(bool enable) override; void SetScissorRegion(Rml::Rectanglei region) override; - + + void EnableClipMask(bool enable) override; + void RenderToClipMask(Rml::ClipMaskOperation mask_operation, Rml::CompiledGeometryHandle geometry, Rml::Vector2f translation) override; + void SetTransform(const Rml::Matrix4f* transform) override; // Can be passed to RenderGeometry() to enable texture rendering without changing the bound texture. From 2def0dfd254489f9767c6f1481c78064bd582152 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 09:39:43 +0200 Subject: [PATCH 052/125] (feat): Implement layer handling --- Backends/RmlUi_Renderer_DX11.cpp | 50 ++++++++++++++++++++++++++++++++ Backends/RmlUi_Renderer_DX11.h | 5 ++++ 2 files changed, 55 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e094f059f..85c22e240 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1328,6 +1328,56 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) m_cbuffer_dirty = true; } +Rml::LayerHandle RenderInterface_DX11::PushLayer() +{ + const Rml::LayerHandle layer_handle = m_render_layers.PushLayer(); + + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + float colors[4] = {0, 0, 0, 0}; + m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, colors); + + return layer_handle; +} + +void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source, Rml::LayerHandle destination, Rml::BlendMode blend_mode, + Rml::Span filters) +{ + using Rml::BlendMode; + + #if 0 + // Blit source layer to postprocessing buffer. Do this regardless of whether we actually have any filters to be + // applied, because we need to resolve the multi-sampled framebuffer in any case. + // @performance If we have BlendMode::Replace and no filters or mask then we can just blit directly to the destination. + BlitLayerToPostprocessPrimary(source_handle); + + // Render the filters, the PostprocessPrimary framebuffer is used for both input and output. + RenderFilters(filters); + + // Render to the destination layer. + glBindFramebuffer(GL_FRAMEBUFFER, render_layers.GetLayer(destination_handle).framebuffer); + Gfx::BindTexture(render_layers.GetPostprocessPrimary()); + + UseProgram(ProgramId::Passthrough); + + if (blend_mode == BlendMode::Replace) + glDisable(GL_BLEND); + + DrawFullscreenQuad(); + + if (blend_mode == BlendMode::Replace) + glEnable(GL_BLEND); + + if (destination_handle != render_layers.GetTopLayerHandle()) + glBindFramebuffer(GL_FRAMEBUFFER, render_layers.GetTopLayer().framebuffer); + #endif +} + +void RenderInterface_DX11::PopLayer() +{ + m_render_layers.PopLayer(); + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); +} + void RenderInterface_DX11::UpdateConstantBuffer() { if (m_cbuffer_dirty) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index c77a10592..dd97053fe 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -103,6 +103,11 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void SetTransform(const Rml::Matrix4f* transform) override; + Rml::LayerHandle PushLayer() override; + void CompositeLayers(Rml::LayerHandle source, Rml::LayerHandle destination, Rml::BlendMode blend_mode, + Rml::Span filters) override; + void PopLayer() override; + // Can be passed to RenderGeometry() to enable texture rendering without changing the bound texture. static constexpr Rml::TextureHandle TextureEnableWithoutBinding = Rml::TextureHandle(-1); // Can be passed to RenderGeometry() to leave the bound texture and used program unchanged. From 0d631a7edb1e113ffcaf191fadc8ae2c579cd8e0 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 23:42:17 +0200 Subject: [PATCH 053/125] (feat): Add all shader programs --- Backends/RmlUi_Renderer_DX11.cpp | 815 ++++++++++++++++++++++--------- Backends/RmlUi_Renderer_DX11.h | 13 +- 2 files changed, 589 insertions(+), 239 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 85c22e240..adb9e5b35 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -46,52 +46,50 @@ #endif // Shader source code -constexpr const char shader_source_text_color[] = R"( -struct sInputData -{ - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; -}; -float4 main(const sInputData inputArgs) : SV_TARGET -{ - return inputArgs.inputColor; -} -)"; +#define MAX_NUM_STOPS 16 +#define BLUR_SIZE 7 +#define BLUR_NUM_WEIGHTS ((BLUR_SIZE + 1) / 2) + +#define RMLUI_STRINGIFY_IMPL(x) #x +#define RMLUI_STRINGIFY(x) RMLUI_STRINGIFY_IMPL(x) -constexpr const char shader_source_text_vertex[] = R"( -struct sInputData +#define RMLUI_SHADER_HEADER \ + "#define MAX_NUM_STOPS " RMLUI_STRINGIFY(MAX_NUM_STOPS) "\n#line " RMLUI_STRINGIFY(__LINE__) "\n" + +constexpr const char shader_vert_main[] = RMLUI_SHADER_HEADER R"( +struct VS_Input { float2 inPosition : POSITION; float4 inColor : COLOR; float2 inTexCoord : TEXCOORD; }; -struct sOutputData +struct PS_INPUT { float4 outPosition : SV_Position; float4 outColor : COLOR; float2 outUV : TEXCOORD; }; -cbuffer ConstantBuffer : register(b0) +cbuffer SharedConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; float2 _padding; + float4 _padding2[21]; // Padding so that cbuffer aligns with the largest one (gradient) }; -sOutputData main(const sInputData inArgs) +PS_INPUT VSMain(const VS_Input IN) { - sOutputData result; + PS_INPUT result = (PS_INPUT)0; - float2 translatedPos = inArgs.inPosition + m_translate; + float2 translatedPos = IN.inPosition + m_translate; float4 resPos = mul(m_transform, float4(translatedPos.x, translatedPos.y, 0.0, 1.0)); result.outPosition = resPos; - result.outColor = inArgs.inColor; - result.outUV = inArgs.inTexCoord; + result.outColor = IN.inColor; + result.outUV = IN.inTexCoord; #if defined(RMLUI_PREMULTIPLIED_ALPHA) // Pre-multiply vertex colors with their alpha. @@ -102,8 +100,8 @@ sOutputData main(const sInputData inArgs) }; )"; -constexpr const char shader_source_text_texture[] = R"( -struct sInputData +constexpr const char shader_frag_texture[] = RMLUI_SHADER_HEADER R"( +struct PS_Input { float4 inputPos : SV_Position; float4 inputColor : COLOR; @@ -113,22 +111,159 @@ struct sInputData Texture2D g_InputTexture : register(t0); SamplerState g_SamplerLinear : register(s0); +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + return IN.inputColor * g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); +}; +)"; -float4 main(const sInputData inputArgs) : SV_TARGET +constexpr const char shader_frag_color[] = RMLUI_SHADER_HEADER R"( +struct PS_Input { - return inputArgs.inputColor * g_InputTexture.Sample(g_SamplerLinear, inputArgs.inputUV); -} + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + return IN.inputColor; +}; )"; -constexpr const char shader_passthrough_source_vertex[] = R"( -struct sInputData +enum class ShaderGradientFunction { Linear, Radial, Conic, RepeatingLinear, RepeatingRadial, RepeatingConic }; // Must match shader definitions below. + +// We need to round up at compile-time so that we can embed the +#define CEILING(x, y) (((x) + (y)-1) / (y)) +static const char shader_frag_gradient[] = RMLUI_SHADER_HEADER "#define MAX_NUM_STOPS_PACKED (uint)" RMLUI_STRINGIFY(CEILING(MAX_NUM_STOPS, 4)) R"( +#define LINEAR 0 +#define RADIAL 1 +#define CONIC 2 +#define REPEATING_LINEAR 3 +#define REPEATING_RADIAL 4 +#define REPEATING_CONIC 5 +#define PI 3.14159265 + +cbuffer SharedConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_translate; + + // One to one translation of the OpenGL uniforms results in a LOT of wasted space due to CBuffer alignment rules. + // Changes from GL3: + // - Moved m_num_stops below m_func (saved 4 bytes of padding). + // - Packed m_stop_positions into a float4[MAX_NUM_STOPS / 4] array, as each array element starts a new 16-byte row. + // The below layout has 0 bytes of padding. + + int m_func; // one of the above definitions + int m_num_stops; + float2 m_p; // linear: starting point, radial: center, conic: center + float2 m_v; // linear: vector to ending point, radial: 2d curvature (inverse radius), conic: angled unit vector + float4 m_stop_colors[MAX_NUM_STOPS]; + float4 m_stop_positions[MAX_NUM_STOPS_PACKED]; // normalized, 0 -> starting point, 1 -> ending point +}; + +// Hide the way the data is packed in the cbuffer through a macro +// @NOTE: Hardcoded for MAX_NUM_STOPS 16. +// i >> 2 => i >> sqrt(MAX_NUM_STOPS) +#define GET_STOP_POS(i) (m_stop_positions[i >> 2][i & 3]) + +struct PS_Input +{ + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +#define glsl_mod(x,y) (((x)-(y)*floor((x)/(y)))) + +float4 lerp_stop_colors(float t) { + float4 color = m_stop_colors[0]; + + for (int i = 1; i < m_num_stops; i++) + color = lerp(color, m_stop_colors[i], smoothstep(GET_STOP_POS(i-1), GET_STOP_POS(i), t)); + + return color; +}; + +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + float t = 0.0; + + if (m_func == LINEAR || m_func == REPEATING_LINEAR) { + float dist_square = dot(m_v, m_v); + float2 V = IN.inputPos.xy - m_p; + t = dot(m_v, V) / dist_square; + } + else if (m_func == RADIAL || m_func == REPEATING_RADIAL) { + float2 V = IN.inputPos.xy - m_p; + t = length(m_v * V); + } + else if (m_func == CONIC || m_func == REPEATING_CONIC) { + float2x2 R = float2x2(m_v.x, -m_v.y, m_v.y, m_v.x); + float2 V = mul(R, (IN.inputPos.xy - m_p)); + t = 0.5 + atan2(V.y, -V.x) / (2.0 * PI); + } + + if (m_func == REPEATING_LINEAR || m_func == REPEATING_RADIAL || m_func == REPEATING_CONIC) { + float t0 = GET_STOP_POS(0); + float t1 = GET_STOP_POS(m_num_stops - 1); + t = t0 + glsl_mod(t - t0, t1 - t0); + } + + return IN.inputColor * lerp_stop_colors(t); +}; +)"; + +// "Creation" by Danilo Guanabara, based on: https://www.shadertoy.com/view/XsXXDn +static const char shader_frag_creation[] = RMLUI_SHADER_HEADER R"( +struct PS_Input +{ + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +cbuffer SharedConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_translate; + float2 m_dimensions; + float m_value; + float3 _padding; + float4 _padding2[20]; // Padding so that cbuffer aligns with the largest one (gradient) +}; + +#define glsl_mod(x,y) (((x)-(y)*floor((x)/(y)))) + +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + float t = m_value; + float3 c; + float l; + for (int i = 0; i < 3; i++) { + float2 p = IN.inputUV; + float2 uv = p; + p -= .5; + p.x *= m_dimensions.x / m_dimensions.y; + float z = t + ((float)i) * .07; + l = length(p); + uv += p / l * (sin(z) + 1.) * abs(sin(l * 9. - z - z)); + c[i] = .01 / length(glsl_mod(uv, 1.) - .5); + } + return float4(c / l, IN.inputColor.a); +}; +)"; + +constexpr const char shader_vert_passthrough[] = RMLUI_SHADER_HEADER R"( +struct VS_Input { float2 inPosition : POSITION; float4 inColor : COLOR; float2 inTexCoord : TEXCOORD; }; -struct sOutputData +struct PS_Input { float4 outPosition : SV_Position; float4 outColor : COLOR; @@ -140,21 +275,22 @@ cbuffer ConstantBuffer : register(b0) float4x4 m_transform; float2 m_translate; float2 _padding; + float4 _padding2[21]; // Padding so that cbuffer aligns with the largest one (gradient) }; -sOutputData main(const sInputData inArgs) +PS_Input VSMain(const VS_Input IN) { - sOutputData result; + PS_Input result = (PS_Input)0; - result.outPosition = float4(inArgs.inPosition.xy, 0.0f, 1.0f); - result.outUV = float2(inArgs.inTexCoord.x, 1.0f - inArgs.inTexCoord.y); + result.outPosition = float4(IN.inPosition.xy, 0.0f, 1.0f); + result.outUV = float2(IN.inTexCoord.x, 1.0f - IN.inTexCoord.y); return result; }; )"; -constexpr const char shader_passthrough_source_fragment[] = R"( -struct sInputData +constexpr const char shader_frag_passthrough[] = RMLUI_SHADER_HEADER R"( +struct PS_Input { float4 inputPos : SV_Position; float4 inputColor : COLOR; @@ -164,14 +300,355 @@ struct sInputData Texture2D g_InputTexture : register(t0); SamplerState g_SamplerLinear : register(s0); +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + return g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); +}; +)"; + +static const char shader_frag_color_matrix[] = RMLUI_SHADER_HEADER R"( + +Texture2D g_InputTexture : register(t0); +SamplerState g_SamplerLinear : register(s0); + +cbuffer ConstantBuffer : register(b0) +{ + float4x4 m_color_matrix; + float4 _padding[22]; // Padding so that cbuffer aligns with the largest one (gradient) +}; -float4 main(const sInputData inputArgs) : SV_TARGET +struct PS_Input { - return g_InputTexture.Sample(g_SamplerLinear, inputArgs.inputUV); -} + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + // The general case uses a 4x5 color matrix for full rgba transformation, plus a constant term with the last column. + // However, we only consider the case of rgb transformations. Thus, we could in principle use a 3x4 matrix, but we + // keep the alpha row for simplicity. + // In the general case we should do the matrix transformation in non-premultiplied space. However, without alpha + // transformations, we can do it directly in premultiplied space to avoid the extra division and multiplication + // steps. In this space, the constant term needs to be multiplied by the alpha value, instead of unity. + float4 texColor = g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); + float3 transformedColor = mul(m_color_matrix, texColor).rgb; + return float4(transformedColor, texColor.a); +}; )"; +static const char shader_frag_blend_mask[] = RMLUI_SHADER_HEADER R"( +Texture2D g_InputTexture : register(t0); +SamplerState g_SamplerLinear : register(s0); +Texture2D g_MaskTexture : register(t1); + +struct PS_Input +{ + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + float4 texColor = g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); + float maskAlpha = g_MaskTexture.Sample(g_SamplerLinear, IN.inputUV).a; + return texColor * maskAlpha; +}; +)"; + +#define RMLUI_SHADER_BLUR_HEADER \ + RMLUI_SHADER_HEADER "\n#define BLUR_SIZE " RMLUI_STRINGIFY(BLUR_SIZE) "\n#define BLUR_NUM_WEIGHTS " RMLUI_STRINGIFY(BLUR_NUM_WEIGHTS) + +static const char shader_vert_blur[] = RMLUI_SHADER_BLUR_HEADER R"( +struct VS_Input +{ + float2 inPosition : POSITION; + float4 inColor : COLOR; + float2 inTexCoord : TEXCOORD; +}; + +struct PS_INPUT +{ + float4 outPosition : SV_Position; + float4 outColor : COLOR; + float2 outUV : TEXCOORD; +}; + +cbuffer SharedConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_translate; + float2 _padding; + float4 _padding2[21]; // Padding so that cbuffer aligns with the largest one (gradient) +}; + +PS_INPUT VSMain(const VS_Input IN) +{ + PS_INPUT result = (PS_INPUT)0; + + float2 translatedPos = IN.inPosition + m_translate; + float4 resPos = mul(m_transform, float4(translatedPos.x, translatedPos.y, 0.0, 1.0)); + + result.outPosition = resPos; + result.outColor = IN.inColor; + result.outUV = IN.inTexCoord; + +#if defined(RMLUI_PREMULTIPLIED_ALPHA) + // Pre-multiply vertex colors with their alpha. + result.outColor.rgb = result.outColor.rgb * result.outColor.a; +#endif + + return result; +}; +)"; + +static const char shader_frag_blur[] = RMLUI_SHADER_BLUR_HEADER R"( +Texture2D g_InputTexture : register(t0); +SamplerState g_SamplerLinear : register(s0); + +cbuffer ConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_translate; + float4 m_weights; + float2 m_texelOffset; + float2 m_texCoordMin; + float2 m_texCoordMax; + float4 _padding[19]; // Padding so that cbuffer aligns with the largest one (gradient) +}; + +struct PS_Input +{ + float4 inputPos : SV_Position; + float2 inputUV[BLUR_SIZE] : TEXCOORD; +}; + +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + float4 color = float4(0.0, 0.0, 0.0, 0.0); + for(int i = 0; i < BLUR_SIZE; i++) + { + float2 in_region = step(m_texCoordMin, IN.inputUV[i]) * step(IN.inputUV[i], m_texCoordMax); + color += g_InputTexture.Sample(g_SamplerLinear, IN.inputUV[i]) * in_region.x * in_region.y * m_weights[abs(i - BLUR_NUM_WEIGHTS + 1)]; + } + return color; +}; +)"; + +static const char shader_frag_drop_shadow[] = RMLUI_SHADER_HEADER R"( +Texture2D g_InputTexture : register(t0); +SamplerState g_SamplerLinear : register(s0); + +cbuffer ConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_translate; + float2 m_texCoordMin; + float2 m_texCoordMax; + float4 m_color; + float2 _padding; + float4 _padding2[19]; // Padding so that cbuffer aligns with the largest one (gradient) +}; + +struct PS_Input +{ + float4 inputPos : SV_Position; + float4 inputColor : COLOR; + float2 inputUV : TEXCOORD; +}; + +float4 PSMain(const PS_Input IN) : SV_TARGET +{ + float2 in_region = step(m_texCoordMin, IN.inputUV) * step(IN.inputUV, m_texCoordMax); + return g_InputTexture.Sample(g_SamplerLinear, IN.inputUV).a * in_region.x * in_region.y * m_color; +}; +)"; + +enum class ProgramId { + None, + Color, + Texture, + Gradient, + Creation, + Passthrough, + ColorMatrix, + BlendMask, + Blur, + DropShadow, + Count, +}; +enum class VertShaderId { + Main, + Passthrough, + Blur, + Count, +}; +enum class FragShaderId { + Color, + Texture, + Gradient, + Creation, + Passthrough, + ColorMatrix, + BlendMask, + Blur, + DropShadow, + Count, +}; +enum class ShaderType { + Vertex, + Fragment, +}; + namespace Gfx { + +struct VertShaderDefinition { + VertShaderId id; + const char* name_str; + const char* code_str; + const size_t code_size; +}; +struct FragShaderDefinition { + FragShaderId id; + const char* name_str; + const char* code_str; + const size_t code_size; +}; +struct ProgramDefinition { + ProgramId id; + const char* name_str; + VertShaderId vert_shader; + FragShaderId frag_shader; +}; + +// clang-format off +static const VertShaderDefinition vert_shader_definitions[] = { + {VertShaderId::Main, "main", shader_vert_main, sizeof(shader_vert_main)}, + {VertShaderId::Passthrough, "passthrough", shader_vert_passthrough, sizeof(shader_vert_passthrough)}, + {VertShaderId::Blur, "blur", shader_vert_blur, sizeof(shader_vert_blur)}, +}; +static const FragShaderDefinition frag_shader_definitions[] = { + {FragShaderId::Color, "color", shader_frag_color, sizeof(shader_frag_color)}, + {FragShaderId::Texture, "texture", shader_frag_texture, sizeof(shader_frag_texture)}, + {FragShaderId::Gradient, "gradient", shader_frag_gradient, sizeof(shader_frag_gradient)}, + {FragShaderId::Creation, "creation", shader_frag_creation, sizeof(shader_frag_creation)}, + {FragShaderId::Passthrough, "passthrough", shader_frag_passthrough, sizeof(shader_frag_passthrough)}, + {FragShaderId::ColorMatrix, "color_matrix", shader_frag_color_matrix, sizeof(shader_frag_color_matrix)}, + {FragShaderId::BlendMask, "blend_mask", shader_frag_blend_mask, sizeof(shader_frag_blend_mask)}, + {FragShaderId::Blur, "blur", shader_frag_blur, sizeof(shader_frag_blur)}, + {FragShaderId::DropShadow, "drop_shadow", shader_frag_drop_shadow, sizeof(shader_frag_drop_shadow)}, +}; +static const ProgramDefinition program_definitions[] = { + {ProgramId::Color, "color", VertShaderId::Main, FragShaderId::Color}, + {ProgramId::Texture, "texture", VertShaderId::Main, FragShaderId::Texture}, + {ProgramId::Gradient, "gradient", VertShaderId::Main, FragShaderId::Gradient}, + {ProgramId::Creation, "creation", VertShaderId::Main, FragShaderId::Creation}, + {ProgramId::Passthrough, "passthrough", VertShaderId::Passthrough, FragShaderId::Passthrough}, + {ProgramId::ColorMatrix, "color_matrix", VertShaderId::Passthrough, FragShaderId::ColorMatrix}, + {ProgramId::BlendMask, "blend_mask", VertShaderId::Passthrough, FragShaderId::BlendMask}, + {ProgramId::Blur, "blur", VertShaderId::Blur, FragShaderId::Blur}, + {ProgramId::DropShadow, "drop_shadow", VertShaderId::Passthrough, FragShaderId::DropShadow}, +}; +// clang-format on + +template +class EnumArray { +public: + const T& operator[](Enum id) const + { + RMLUI_ASSERT((size_t)id < (size_t)Enum::Count); + return ids[size_t(id)]; + } + T& operator[](Enum id) + { + RMLUI_ASSERT((size_t)id < (size_t)Enum::Count); + return ids[size_t(id)]; + } + auto begin() const { return ids.begin(); } + auto end() const { return ids.end(); } + +private: + Rml::Array ids = {}; +}; + +struct ShaderProgram { +public: + ID3D11VertexShader* vertex_shader; + ID3D11PixelShader* pixel_shader; +}; + +using Programs = EnumArray; +using VertShaders = EnumArray; +using FragShaders = EnumArray; + +struct ProgramData { + Programs programs; + VertShaders vert_shaders; + FragShaders frag_shaders; +}; + +// Create the shader, 'shader_type' is either GL_VERTEX_SHADER or GL_FRAGMENT_SHADER. +static bool CreateShader(ID3D11Device* p_device, ID3DBlob*& out_shader_dxil, ShaderType shader_type, const char* code_string, const size_t code_length) +{ + RMLUI_ASSERT(shader_type == ShaderType::Vertex || shader_type == ShaderType::Fragment); + + #ifdef RMLUI_DX_DEBUG + static const UINT default_shader_flags = D3DCOMPILE_DEBUG | D3DCOMPILE_SKIP_OPTIMIZATION; + #else + static const UINT default_shader_flags = 0; + #endif + + // Compile shader + const D3D_SHADER_MACRO macros[] = {"RMLUI_PREMULTIPLIED_ALPHA", nullptr, nullptr, nullptr}; + ID3DBlob* p_error_buff = nullptr; + + HRESULT result = D3DCompile(code_string, code_length, nullptr, macros, nullptr, shader_type == ShaderType::Vertex ? "VSMain" : "PSMain", + shader_type == ShaderType::Vertex ? "vs_5_0" : "ps_5_0", default_shader_flags, 0, &out_shader_dxil, &p_error_buff); + RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); + return false; + } + #endif + + DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); + + return true; +} + +static bool CreateProgram(ID3D11Device* p_device, ShaderProgram& out_program, ProgramId program_id, ID3DBlob* vertex_shader, ID3DBlob* fragment_shader) +{ + RMLUI_ASSERT(vertex_shader); + RMLUI_ASSERT(fragment_shader); + + HRESULT result = + p_device->CreateVertexShader(vertex_shader->GetBufferPointer(), vertex_shader->GetBufferSize(), nullptr, &out_program.vertex_shader); + RMLUI_DX_ASSERTMSG(result, "failed to CreateVertexShader"); + if (FAILED(result)) + { + #ifdef RMLUI_DX_DEBUG + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create vertex shader: %d", result); + #endif + return false; + } + + result = p_device->CreatePixelShader(fragment_shader->GetBufferPointer(), fragment_shader->GetBufferSize(), nullptr, &out_program.pixel_shader); + RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); + if (FAILED(result)) + { + #ifdef RMLUI_DX_DEBUG + Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create pixel shader: %d", result); + #endif + return false; + } + + return true; +} + struct RenderTargetData { int width = 0, height = 0; ID3D11RenderTargetView* render_target_view = nullptr; // To write to colour attachment buffer @@ -320,6 +797,56 @@ static void BindTexture(ID3D11DeviceContext* context, const RenderTargetData& rt context->PSSetShaderResources(0, slot, &rt.render_target_shader_resource_view); } +static bool CreateShaders(ID3D11Device* p_device, ProgramData& data) +{ + RMLUI_ASSERT(std::all_of(data.vert_shaders.begin(), data.vert_shaders.end(), [](auto&& value) { return value == nullptr; })); + RMLUI_ASSERT(std::all_of(data.frag_shaders.begin(), data.frag_shaders.end(), [](auto&& value) { return value == nullptr; })); + RMLUI_ASSERT(std::all_of(data.programs.begin(), data.programs.end(), [](const ShaderProgram& value) { return value.vertex_shader == nullptr || value.pixel_shader == nullptr; })); + auto ReportError = [](const char* type, const char* name) { + Rml::Log::Message(Rml::Log::LT_ERROR, "Could not create shader %s: '%s'.", type, name); + return false; + }; + + for (const VertShaderDefinition& def : vert_shader_definitions) + { + if (!CreateShader(p_device, data.vert_shaders[def.id], ShaderType::Vertex, def.code_str, def.code_size)) + return ReportError("vertex shader", def.name_str); + } + + for (const FragShaderDefinition& def : frag_shader_definitions) + { + if (!CreateShader(p_device, data.frag_shaders[def.id], ShaderType::Fragment, def.code_str, def.code_size)) + return ReportError("fragment shader", def.name_str); + } + + for (const ProgramDefinition& def : program_definitions) + { + if (!CreateProgram(p_device, data.programs[def.id], def.id, data.vert_shaders[def.vert_shader], data.frag_shaders[def.frag_shader])) + return ReportError("program", def.name_str); + } + + return true; +} + +static void DestroyShaders(const ProgramData& data) +{ + for (ShaderProgram programs : data.programs) + { + DX_CLEANUP_RESOURCE_IF_CREATED(programs.vertex_shader); + DX_CLEANUP_RESOURCE_IF_CREATED(programs.pixel_shader); + } + + for (ID3DBlob* shader_blob : data.vert_shaders) + { + DX_CLEANUP_RESOURCE_IF_CREATED(shader_blob); + } + + for (ID3DBlob* shader_blob : data.frag_shaders) + { + DX_CLEANUP_RESOURCE_IF_CREATED(shader_blob); + } +} + } // namespace Gfx static uintptr_t HashPointer(uintptr_t inPtr) @@ -334,14 +861,8 @@ static uintptr_t HashPointer(uintptr_t inPtr) return Value; } -RenderInterface_DX11::RenderInterface_DX11() -{ - #ifdef RMLUI_DX_DEBUG - m_default_shader_flags = D3DCOMPILE_DEBUG | D3DCOMPILE_SKIP_OPTIMIZATION; - #else - m_default_shader_flags = 0; - #endif -} +RenderInterface_DX11::RenderInterface_DX11() {} +RenderInterface_DX11::~RenderInterface_DX11() {} void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context) { @@ -423,172 +944,10 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } // Compile and load shaders - - // Buffer shall be cleared up later, as it's required to create the input layout - ID3DBlob* p_shader_vertex_common{}; + auto mut_program_data = Rml::MakeUnique(); + if (Gfx::CreateShaders(m_d3d_device, *mut_program_data)) { - const D3D_SHADER_MACRO macros[] = {"RMLUI_PREMULTIPLIED_ALPHA", nullptr, nullptr, nullptr}; - ID3DBlob* p_error_buff{}; - - // Common vertex shader - HRESULT result = D3DCompile(shader_source_text_vertex, sizeof(shader_source_text_vertex), nullptr, macros, nullptr, "main", "vs_5_0", - this->m_default_shader_flags, 0, &p_shader_vertex_common, &p_error_buff); - RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); - #ifdef RMLUI_DX_DEBUG - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); - } - #endif - - DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); - - // Color fragment shader - ID3DBlob* p_shader_color_pixel{}; - - result = D3DCompile(shader_source_text_color, sizeof(shader_source_text_color), nullptr, macros, nullptr, "main", "ps_5_0", - this->m_default_shader_flags, 0, &p_shader_color_pixel, &p_error_buff); - RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); - #ifdef RMLUI_DX_DEBUG - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); - } - #endif - - DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); - - // Texture fragment shader - ID3DBlob* p_shader_color_texture{}; - - result = D3DCompile(shader_source_text_texture, sizeof(shader_source_text_texture), nullptr, macros, nullptr, "main", "ps_5_0", - this->m_default_shader_flags, 0, &p_shader_color_texture, &p_error_buff); - RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); - #ifdef RMLUI_DX_DEBUG - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); - } - #endif - - DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); - - // Passthrough vertex shader - ID3DBlob* p_shader_passthrough_vertex{}; - - result = D3DCompile(shader_passthrough_source_vertex, sizeof(shader_passthrough_source_vertex), nullptr, macros, nullptr, "main", - "vs_5_0", this->m_default_shader_flags, 0, &p_shader_passthrough_vertex, &p_error_buff); - RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); - #ifdef RMLUI_DX_DEBUG - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); - } - #endif - - DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); - - // Passthrough fragment shader - ID3DBlob* p_shader_passthrough_pixel{}; - - result = D3DCompile(shader_passthrough_source_fragment, sizeof(shader_passthrough_source_fragment), nullptr, macros, nullptr, "main", - "ps_5_0", this->m_default_shader_flags, 0, &p_shader_passthrough_pixel, &p_error_buff); - RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); - #ifdef RMLUI_DX_DEBUG - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to compile shader: %s", (char*)p_error_buff->GetBufferPointer()); - } - #endif - - DX_CLEANUP_RESOURCE_IF_CREATED(p_error_buff); - - // Create the shader objects - result = m_d3d_device->CreateVertexShader(p_shader_vertex_common->GetBufferPointer(), p_shader_vertex_common->GetBufferSize(), nullptr, - &m_shader_vertex_common); - RMLUI_DX_ASSERTMSG(result, "failed to CreateVertexShader"); - if (FAILED(result)) - { - #ifdef RMLUI_DX_DEBUG - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create vertex shader: %d", result); - #endif - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); - goto cleanup; - return; - } - - result = m_d3d_device->CreatePixelShader(p_shader_color_pixel->GetBufferPointer(), p_shader_color_pixel->GetBufferSize(), nullptr, - &m_shader_pixel_color); - RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); - if (FAILED(result)) - { - #ifdef RMLUI_DX_DEBUG - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create pixel shader: %d", result); - #endif - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); - goto cleanup; - return; - } - - result = m_d3d_device->CreatePixelShader(p_shader_color_texture->GetBufferPointer(), p_shader_color_texture->GetBufferSize(), nullptr, - &m_shader_pixel_texture); - RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); - if (FAILED(result)) - { - #ifdef RMLUI_DX_DEBUG - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create pixel shader: %d", result); - #endif - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); - goto cleanup; - return; - } - - // Passthrough - result = m_d3d_device->CreateVertexShader(p_shader_passthrough_vertex->GetBufferPointer(), p_shader_passthrough_vertex->GetBufferSize(), - nullptr, &m_shader_passthrough_vertex); - RMLUI_DX_ASSERTMSG(result, "failed to CreateVertexShader"); - if (FAILED(result)) - { - #ifdef RMLUI_DX_DEBUG - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create vertex shader: %d", result); - #endif - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); - goto cleanup; - return; - } - - result = m_d3d_device->CreatePixelShader(p_shader_passthrough_pixel->GetBufferPointer(), p_shader_passthrough_pixel->GetBufferSize(), nullptr, - &m_shader_passthrough_fragment); - RMLUI_DX_ASSERTMSG(result, "failed to CreatePixelShader"); - if (FAILED(result)) - { - #ifdef RMLUI_DX_DEBUG - Rml::Log::Message(Rml::Log::Type::LT_ERROR, "failed to create pixel shader: %d", result); - #endif - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); - goto cleanup; - return; - } - - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_pixel); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_color_texture); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_vertex); - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_passthrough_pixel); + program_data = std::move(mut_program_data); } // Create vertex layout. This will be constant to avoid copying to an intermediate struct. @@ -600,8 +959,9 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* }; uint32_t numElements = sizeof(polygonLayout) / sizeof(polygonLayout[0]); - HRESULT result = m_d3d_device->CreateInputLayout(polygonLayout, numElements, p_shader_vertex_common->GetBufferPointer(), - p_shader_vertex_common->GetBufferSize(), &m_vertex_layout); + HRESULT result = + m_d3d_device->CreateInputLayout(polygonLayout, numElements, program_data->vert_shaders[VertShaderId::Main]->GetBufferPointer(), + program_data->vert_shaders[VertShaderId::Main]->GetBufferSize(), &m_vertex_layout); RMLUI_DX_ASSERTMSG(result, "failed to CreateInputLayout"); if (FAILED(result)) { @@ -688,15 +1048,16 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Intersect m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_stencil_intersect); } - -cleanup: - - // Release the vertex shader buffer and pixel shader buffer since they are no longer needed. - DX_CLEANUP_RESOURCE_IF_CREATED(p_shader_vertex_common); } void RenderInterface_DX11::Cleanup() { + if (program_data) + { + Gfx::DestroyShaders(*program_data); + program_data.reset(); + } + // Loop through geometry cache and free all resources for (auto& it : m_geometry_cache) { @@ -712,13 +1073,8 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_set); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); - DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_passthrough_fragment); - DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_passthrough_vertex); - DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_color); - DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_pixel_texture); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); - DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_vertex_common); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) @@ -783,7 +1139,7 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, clearColor); - // UseProgram(ProgramId::None); + UseProgram(ProgramId::None); // program_transform_dirty.set(); m_scissor_state = Rml::Rectanglei::MakeInvalid(); } @@ -805,11 +1161,8 @@ void RenderInterface_DX11::EndFrame() // Instead, if we had a transparent destination that didn't use premultiplied alpha, we would need to perform a manual un-premultiplication step. Gfx::BindTexture(m_d3d_context, rt_postprocess); - // @TODO: UseProgram // @TODO: Find a pattern for flipped textures - // UseProgram(ProgramId::Passthrough); - m_d3d_context->VSSetShader(this->m_shader_passthrough_vertex, nullptr, 0); - m_d3d_context->PSSetShader(this->m_shader_passthrough_fragment, nullptr, 0); + UseProgram(ProgramId::Passthrough); DrawFullscreenQuad(); @@ -1006,8 +1359,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm else if (texture) { // Texture available - m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); - m_d3d_context->PSSetShader(this->m_shader_pixel_texture, nullptr, 0); + UseProgram(ProgramId::Texture); if (texture != TextureEnableWithoutBinding) { ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture; @@ -1018,8 +1370,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm else { // No texture, use color - m_d3d_context->VSSetShader(this->m_shader_vertex_common, nullptr, 0); - m_d3d_context->PSSetShader(this->m_shader_pixel_color, nullptr, 0); + UseProgram(ProgramId::Color); } UpdateConstantBuffer(); diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index dd97053fe..52b70919e 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -64,6 +64,7 @@ typedef void (*pfnLoadTextureRaw)(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize); typedef void (*pfnFreeTextureRaw)(uint8_t* pData); +enum class ProgramId; class RenderLayerStack; namespace Gfx { struct ProgramData; @@ -73,6 +74,7 @@ struct RenderTargetData; class RenderInterface_DX11 : public Rml::RenderInterface { public: RenderInterface_DX11(); + ~RenderInterface_DX11(); // Resource initialisation and cleanup void Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context); @@ -129,6 +131,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { // Changes blend state if necessary void SetBlendState(ID3D11BlendState* blendState); void UpdateConstantBuffer(); + void UseProgram(ProgramId program_id); void SetScissor(Rml::Rectanglei region, bool vertically_flip = false); @@ -136,8 +139,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void DrawFullscreenQuad(Rml::Vector2f uv_offset, Rml::Vector2f uv_scaling = Rml::Vector2f(1.f)); private: - UINT m_default_shader_flags; - // D3D11 core resources ID3D11Device* m_d3d_device = nullptr; ID3D11DeviceContext* m_d3d_context = nullptr; @@ -158,11 +159,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11Buffer* m_shader_buffer = nullptr; bool m_cbuffer_dirty = true; ID3D11SamplerState* m_samplerState = nullptr; - ID3D11VertexShader* m_shader_vertex_common = nullptr; - ID3D11PixelShader* m_shader_pixel_color = nullptr; - ID3D11PixelShader* m_shader_pixel_texture = nullptr; - ID3D11VertexShader* m_shader_passthrough_vertex = nullptr; - ID3D11PixelShader* m_shader_passthrough_fragment = nullptr; // Viewport dimensions int m_viewport_width = 0; @@ -174,7 +170,10 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11InputLayout* m_vertex_layout = nullptr; ID3D11BlendState* m_blend_state = nullptr; + + ProgramId active_program = {}; Rml::Rectanglei m_scissor_state = Rml::Rectanglei::MakeInvalid(); + Rml::UniquePtr program_data; struct DX11_GeometryData { public: From be31214fcdc190359de93d5b651b8115781faafd Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 23:42:55 +0200 Subject: [PATCH 054/125] (feat): Add all properties to cbuffer struct Most are unused currently, will be used when UseShader is implemented. --- Backends/RmlUi_Renderer_DX11.h | 39 ++++++++++++++++++++++++++++++---- 1 file changed, 35 insertions(+), 4 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 52b70919e..7b71c34f2 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -216,10 +216,41 @@ class RenderInterface_DX11 : public Rml::RenderInterface { D3D11State m_previous_d3d_state{}; #pragma pack(4) - struct ShaderCbuffer { - Rml::Matrix4f transform; - Rml::Vector2f translation; - float _padding[2]; + union ShaderCbuffer { + struct { + Rml::Matrix4f transform; + Rml::Vector2f translation; + }; + struct Gradient { + int _padding[18]; + int func; + int num_stops; + Rml::Vector2f p; + Rml::Vector2f v; + Rml::Vector4f stop_colors[16]; + float stop_positions[16]; + } gradient; + struct ColorMatrix { + Rml::Matrix4f color_matrix; + } color_matrix; + struct Blur { + int _padding[18]; + Rml::Vector4f weights; + Rml::Vector2f texelOffset; + Rml::Vector2f texCoordMin; + Rml::Vector2f texCoordMax; + } blur; + struct DropShadow { + int _padding[18]; + Rml::Vector2f texCoordMin; + Rml::Vector2f texCoordMax; + Rml::Vector4f color; + } dropShadow; + struct Creation { + int _padding[18]; + Rml::Vector2f dimensions; + float value; + } creation; }; #pragma pack() From e7ad804702749f05cbbb7bcc9f0fcced11d20b6f Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 23:44:02 +0200 Subject: [PATCH 055/125] (fix): Correct clip masks --- Backends/RmlUi_Renderer_DX11.cpp | 96 +++++++++++++++++++++++--------- Backends/RmlUi_Renderer_DX11.h | 4 +- 2 files changed, 72 insertions(+), 28 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index adb9e5b35..d8762ed0d 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -881,21 +881,45 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). - if (!m_blend_state) + if (!m_blend_state_enable) { - D3D11_BLEND_DESC blendDesc{}; - ZeroMemory(&blendDesc, sizeof(blendDesc)); - blendDesc.AlphaToCoverageEnable = FALSE; - blendDesc.IndependentBlendEnable = FALSE; - blendDesc.RenderTarget[0].BlendEnable = TRUE; - blendDesc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE; - blendDesc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA; - blendDesc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; - blendDesc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ONE; - blendDesc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA; - blendDesc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; - blendDesc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; - HRESULT result = m_d3d_device->CreateBlendState(&blendDesc, &m_blend_state); + D3D11_BLEND_DESC blend_desc{}; + ZeroMemory(&blend_desc, sizeof(blend_desc)); + blend_desc.AlphaToCoverageEnable = false; + blend_desc.IndependentBlendEnable = false; + blend_desc.RenderTarget[0].BlendEnable = false; + HRESULT result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_disable); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); + return; + } + #endif + + blend_desc.RenderTarget[0].BlendEnable = true; + blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE; + blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA; + blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ONE; + blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA; + blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; + result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_enable); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); + return; + } + #endif + + ZeroMemory(&blend_desc, sizeof(blend_desc)); + blend_desc.RenderTarget[0].BlendEnable = false; + blend_desc.RenderTarget[0].RenderTargetWriteMask = 0; + result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_disable_color); RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); #ifdef RMLUI_DX_DEBUG if (FAILED(result)) @@ -965,7 +989,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* RMLUI_DX_ASSERTMSG(result, "failed to CreateInputLayout"); if (FAILED(result)) { - goto cleanup; return; } } @@ -984,7 +1007,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); if (FAILED(result)) { - goto cleanup; return; } } @@ -1010,7 +1032,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* RMLUI_DX_ASSERTMSG(result, "failed to CreateSamplerState"); if (FAILED(result)) { - goto cleanup; return; } } @@ -1020,18 +1041,18 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* D3D11_DEPTH_STENCIL_DESC desc; ZeroMemory(&desc, sizeof(desc)); desc.DepthEnable = false; - desc.DepthWriteMask = D3D11_DEPTH_WRITE_MASK_ALL; - desc.DepthFunc = D3D11_COMPARISON_ALWAYS; - desc.FrontFace.StencilFailOp = desc.FrontFace.StencilDepthFailOp = desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; - desc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; + desc.FrontFace.StencilFunc = D3D11_COMPARISON_EQUAL; + desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; + desc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + desc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; desc.BackFace = desc.FrontFace; desc.StencilEnable = false; // Disabled m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_disable); desc.StencilEnable = true; - desc.StencilReadMask = 0xFF; - desc.StencilWriteMask = 0xFF; + desc.StencilReadMask = D3D11_DEFAULT_STENCIL_READ_MASK; + desc.StencilWriteMask = D3D11_DEFAULT_STENCIL_WRITE_MASK; desc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_REPLACE; desc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; @@ -1067,7 +1088,9 @@ void RenderInterface_DX11::Cleanup() // Cleans up all general resources DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); - DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_enable); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable_color); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_intersect); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_set); @@ -1126,16 +1149,15 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); - SetBlendState(m_blend_state); + SetBlendState(m_blend_state_enable); m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); // Disable scissor m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); - m_d3d_context->OMSetRenderTargets(1, &m_bound_render_target, nullptr); Clear(); SetTransform(nullptr); m_render_layers.BeginFrame(m_viewport_width, m_viewport_height); - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, clearColor); @@ -1754,6 +1776,19 @@ void RenderInterface_DX11::UpdateConstantBuffer() } } +void RenderInterface_DX11::UseProgram(ProgramId program_id) { + RMLUI_ASSERT(program_data); + if (active_program != program_id) + { + if (program_id != ProgramId::None) + { + m_d3d_context->VSSetShader(program_data->programs[program_id].vertex_shader, nullptr, 0); + m_d3d_context->PSSetShader(program_data->programs[program_id].pixel_shader, nullptr, 0); + } + active_program = program_id; + } +} + void RenderInterface_DX11::EnableClipMask(bool enable) { if (enable != m_is_stencil_enabled) @@ -1795,7 +1830,12 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm } break; } + + // Disable writing to the color of the render target + float blendFactor[4] = {0.f, 0.f, 0.f, 0.f}; + m_d3d_context->OMSetBlendState(m_blend_state_disable_color, blendFactor, 0xffffffff); + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); m_d3d_context->OMSetDepthStencilState(stencil_state, stencil_test_value); bool clear_stencil = (operation == ClipMaskOperation::Set || operation == ClipMaskOperation::SetInverse); @@ -1813,6 +1853,8 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm // Restore state // @performance Is this even necessary? m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); + // Restore blend state + m_d3d_context->OMSetBlendState(m_current_blend_state, blendFactor, 0xffffffff); } RenderInterface_DX11::RenderLayerStack::RenderLayerStack() diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 7b71c34f2..53baaecf7 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -169,7 +169,9 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::Vector2f m_translation = {}; ID3D11InputLayout* m_vertex_layout = nullptr; - ID3D11BlendState* m_blend_state = nullptr; + ID3D11BlendState* m_blend_state_enable = nullptr; + ID3D11BlendState* m_blend_state_disable = nullptr; + ID3D11BlendState* m_blend_state_disable_color = nullptr; ProgramId active_program = {}; Rml::Rectanglei m_scissor_state = Rml::Rectanglei::MakeInvalid(); From e337ca686c6627cf2b4b9375a01f1668ee253e39 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 26 Sep 2024 23:44:29 +0200 Subject: [PATCH 056/125] (feat): Half implement layer compositing Filters are ignored currently --- Backends/RmlUi_Renderer_DX11.cpp | 118 +++++++++++++++++++++++++++---- Backends/RmlUi_Renderer_DX11.h | 6 ++ 2 files changed, 112 insertions(+), 12 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index d8762ed0d..ad4c1c922 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1564,7 +1564,10 @@ Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::SpanUpdateSubresource(gpu_texture, 0, nullptr, source.data(), rowPitch, 0); + if (source.size() > 0) + { + m_d3d_context->UpdateSubresource(gpu_texture, 0, nullptr, source.data(), rowPitch, 0); + } // Setup the shader resource view description. D3D11_SHADER_RESOURCE_VIEW_DESC srvDesc{}; @@ -1584,7 +1587,10 @@ Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::SpanGenerateMips(gpu_texture_view); + if (source.size() > 0) + { + m_d3d_context->GenerateMips(gpu_texture_view); + } uintptr_t handleId = HashPointer((uintptr_t)gpu_texture_view); @@ -1701,6 +1707,15 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) m_cbuffer_dirty = true; } +void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle) +{ + const Gfx::RenderTargetData& source = m_render_layers.GetLayer(layer_handle); + const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessPrimary(); + + // Blit and resolve MSAA. Any active scissor state will restrict the size of the blit region. + m_d3d_context->ResolveSubresource(destination.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); +} + Rml::LayerHandle RenderInterface_DX11::PushLayer() { const Rml::LayerHandle layer_handle = m_render_layers.PushLayer(); @@ -1712,37 +1727,37 @@ Rml::LayerHandle RenderInterface_DX11::PushLayer() return layer_handle; } -void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source, Rml::LayerHandle destination, Rml::BlendMode blend_mode, +void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml::LayerHandle destination_handle, Rml::BlendMode blend_mode, Rml::Span filters) { using Rml::BlendMode; - #if 0 // Blit source layer to postprocessing buffer. Do this regardless of whether we actually have any filters to be // applied, because we need to resolve the multi-sampled framebuffer in any case. // @performance If we have BlendMode::Replace and no filters or mask then we can just blit directly to the destination. BlitLayerToPostprocessPrimary(source_handle); // Render the filters, the PostprocessPrimary framebuffer is used for both input and output. - RenderFilters(filters); + // @TODO: RenderFilters implementation + // RenderFilters(filters); // Render to the destination layer. - glBindFramebuffer(GL_FRAMEBUFFER, render_layers.GetLayer(destination_handle).framebuffer); - Gfx::BindTexture(render_layers.GetPostprocessPrimary()); + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetLayer(destination_handle).render_target_view, + m_render_layers.GetLayer(destination_handle).depth_stencil_view); + Gfx::BindTexture(m_d3d_context, m_render_layers.GetPostprocessPrimary()); UseProgram(ProgramId::Passthrough); if (blend_mode == BlendMode::Replace) - glDisable(GL_BLEND); + SetBlendState(m_blend_state_disable); DrawFullscreenQuad(); if (blend_mode == BlendMode::Replace) - glEnable(GL_BLEND); + SetBlendState(m_blend_state_enable); - if (destination_handle != render_layers.GetTopLayerHandle()) - glBindFramebuffer(GL_FRAMEBUFFER, render_layers.GetTopLayer().framebuffer); - #endif + if (destination_handle != m_render_layers.GetTopLayerHandle()) + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); } void RenderInterface_DX11::PopLayer() @@ -1751,6 +1766,85 @@ void RenderInterface_DX11::PopLayer() m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); } +Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() +{ + RMLUI_ASSERT(m_scissor_state.Valid()); + const Rml::Rectanglei bounds = m_scissor_state; + + Rml::TextureHandle render_texture = GenerateTexture({}, bounds.Size()); + if (!render_texture) + return {}; + + BlitLayerToPostprocessPrimary(m_render_layers.GetTopLayerHandle()); + + EnableScissorRegion(false); + + const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); + + // Flip the image vertically, as that convention is used for textures, and move to origin. + D3D11_BOX source_box{}; + source_box.left = bounds.Left(); + source_box.right = bounds.Right(); + source_box.top = source.height - bounds.Bottom(); + source_box.bottom = source.height - bounds.Top(); + source_box.front = 0; + source_box.back = 1; + + // Perform a copy from the source render target to the destination render target + m_d3d_context->CopySubresourceRegion(destination.render_target_texture, 0, 0, 0, 0, source.render_target_texture, 0, &source_box); + + // Now we need to copy the destination texture to the final texture for rendering + // Bind the destination texture as the source for copying to the final texture (render_texture) + // after extracting the associated resource with it + ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)render_texture; + ID3D11Resource* texture_resource = nullptr; + texture_view->GetResource(&texture_resource); + + // Copy the destination texture to the final texture resource + D3D11_BOX copy_box{}; + copy_box.left = 0; + copy_box.right = bounds.Width(); + copy_box.top = 0; + copy_box.bottom = bounds.Height(); + copy_box.front = 0; + copy_box.back = 1; + + // Copy the blitted content from the destination to the final texture + m_d3d_context->CopySubresourceRegion(texture_resource, 0, 0, 0, 0, destination.render_target_texture, 0, ©_box); + + // Restore state and free memory + SetScissor(bounds); + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + texture_resource->Release(); + + return render_texture; +} + +/* +Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() +{ + BlitLayerToPostprocessPrimary(m_render_layers.GetTopLayerHandle()); + + const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& destination = m_render_layers.GetBlendMask(); + + glBindFramebuffer(GL_FRAMEBUFFER, destination.framebuffer); + BindTexture(source); + UseProgram(ProgramId::Passthrough); + glDisable(GL_BLEND); + + DrawFullscreenQuad(); + + glEnable(GL_BLEND); + glBindFramebuffer(GL_FRAMEBUFFER, m_render_layers.GetTopLayer().framebuffer); + + CompiledFilter filter = {}; + filter.type = FilterType::MaskImage; + return reinterpret_cast(new CompiledFilter(std::move(filter))); +} +*/ + void RenderInterface_DX11::UpdateConstantBuffer() { if (m_cbuffer_dirty) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 53baaecf7..014031cc0 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -110,6 +110,10 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::Span filters) override; void PopLayer() override; + Rml::TextureHandle SaveLayerAsTexture() override; + + // Rml::CompiledFilterHandle SaveLayerAsMaskImage() override; + // Can be passed to RenderGeometry() to enable texture rendering without changing the bound texture. static constexpr Rml::TextureHandle TextureEnableWithoutBinding = Rml::TextureHandle(-1); // Can be passed to RenderGeometry() to leave the bound texture and used program unchanged. @@ -133,6 +137,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void UpdateConstantBuffer(); void UseProgram(ProgramId program_id); + void BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle); + void SetScissor(Rml::Rectanglei region, bool vertically_flip = false); void DrawFullscreenQuad(); From 06f09465d42144b8e083c6cc496cdfb028dc7368 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 00:42:14 +0200 Subject: [PATCH 057/125] (feat): Implement shaders --- Backends/RmlUi_Renderer_DX11.cpp | 194 +++++++++++++++++++++++++++++++ Backends/RmlUi_Renderer_DX11.h | 4 + 2 files changed, 198 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index ad4c1c922..e878fd35c 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1845,6 +1845,200 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() } */ +enum class CompiledShaderType { Invalid = 0, Gradient, Creation }; +struct CompiledShader { + CompiledShaderType type; + + // Gradient + ShaderGradientFunction gradient_function; + Rml::Vector2f p; + Rml::Vector2f v; + Rml::Vector stop_positions; + Rml::Vector stop_colors; + + // Shader + Rml::Vector2f dimensions; +}; + +Rml::CompiledShaderHandle RenderInterface_DX11::CompileShader(const Rml::String& name, const Rml::Dictionary& parameters) +{ + auto ApplyColorStopList = [](CompiledShader& shader, const Rml::Dictionary& shader_parameters) { + auto it = shader_parameters.find("color_stop_list"); + RMLUI_ASSERT(it != shader_parameters.end() && it->second.GetType() == Rml::Variant::COLORSTOPLIST); + const Rml::ColorStopList& color_stop_list = it->second.GetReference(); + const int num_stops = Rml::Math::Min((int)color_stop_list.size(), MAX_NUM_STOPS); + + shader.stop_positions.resize(num_stops); + shader.stop_colors.resize(num_stops); + for (int i = 0; i < num_stops; i++) + { + const Rml::ColorStop& stop = color_stop_list[i]; + RMLUI_ASSERT(stop.position.unit == Rml::Unit::NUMBER); + shader.stop_positions[i] = stop.position.number; + shader.stop_colors[i] = ConvertToColorf(stop.color); + } + }; + + CompiledShader shader = {}; + + if (name == "linear-gradient") + { + shader.type = CompiledShaderType::Gradient; + const bool repeating = Rml::Get(parameters, "repeating", false); + shader.gradient_function = (repeating ? ShaderGradientFunction::RepeatingLinear : ShaderGradientFunction::Linear); + shader.p = Rml::Get(parameters, "p0", Rml::Vector2f(0.f)); + shader.v = Rml::Get(parameters, "p1", Rml::Vector2f(0.f)) - shader.p; + ApplyColorStopList(shader, parameters); + } + else if (name == "radial-gradient") + { + shader.type = CompiledShaderType::Gradient; + const bool repeating = Rml::Get(parameters, "repeating", false); + shader.gradient_function = (repeating ? ShaderGradientFunction::RepeatingRadial : ShaderGradientFunction::Radial); + shader.p = Rml::Get(parameters, "center", Rml::Vector2f(0.f)); + shader.v = Rml::Vector2f(1.f) / Rml::Get(parameters, "radius", Rml::Vector2f(1.f)); + ApplyColorStopList(shader, parameters); + } + else if (name == "conic-gradient") + { + shader.type = CompiledShaderType::Gradient; + const bool repeating = Rml::Get(parameters, "repeating", false); + shader.gradient_function = (repeating ? ShaderGradientFunction::RepeatingConic : ShaderGradientFunction::Conic); + shader.p = Rml::Get(parameters, "center", Rml::Vector2f(0.f)); + const float angle = Rml::Get(parameters, "angle", 0.f); + shader.v = {Rml::Math::Cos(angle), Rml::Math::Sin(angle)}; + ApplyColorStopList(shader, parameters); + } + else if (name == "shader") + { + const Rml::String value = Rml::Get(parameters, "value", Rml::String()); + if (value == "creation") + { + shader.type = CompiledShaderType::Creation; + shader.dimensions = Rml::Get(parameters, "dimensions", Rml::Vector2f(0.f)); + } + } + + if (shader.type != CompiledShaderType::Invalid) + return reinterpret_cast(new CompiledShader(std::move(shader))); + + Rml::Log::Message(Rml::Log::LT_WARNING, "Unsupported shader type '%s'.", name.c_str()); + return {}; +} + +void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, Rml::CompiledGeometryHandle geometry_handle, + Rml::Vector2f translation, Rml::TextureHandle /*texture*/) +{ + RMLUI_ASSERT(shader_handle && geometry_handle); + const CompiledShader& shader = *reinterpret_cast(shader_handle); + const CompiledShaderType type = shader.type; + + DX11_GeometryData geometry{}; + if (m_geometry_cache.find(geometry_handle) != m_geometry_cache.end()) + { + geometry = m_geometry_cache[geometry_handle]; + } + + switch (type) + { + case CompiledShaderType::Gradient: + { + RMLUI_ASSERT(shader.stop_positions.size() == shader.stop_colors.size()); + const int num_stops = (int)shader.stop_positions.size(); + + UseProgram(ProgramId::Gradient); + + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return; + } + + // Get a pointer to the data in the constant buffer. + ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + + // Copy the data to the GPU + dataPtr->transform = m_transform; + dataPtr->translation = translation; + dataPtr->gradient.func = static_cast(shader.gradient_function); + dataPtr->gradient.p = shader.p; + dataPtr->gradient.v = shader.v; + dataPtr->gradient.num_stops = num_stops; + // Reset stop positions and colours to 0 + memset(dataPtr->gradient.stop_positions, 0, sizeof(dataPtr->gradient.stop_positions)); + memset(dataPtr->gradient.stop_colors, 0, sizeof(dataPtr->gradient.stop_colors)); + // Copy to stop position and colours + memcpy_s(&dataPtr->gradient.stop_positions, num_stops * sizeof(float), shader.stop_positions.data(), num_stops * sizeof(float)); + memcpy_s(&dataPtr->gradient.stop_colors, num_stops * sizeof(Rml::Vector4f), shader.stop_colors.data(), num_stops * sizeof(Rml::Vector4f)); + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + + // Issue draw call + m_d3d_context->IASetInputLayout(m_vertex_layout); + m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); + m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); + uint32_t stride = sizeof(Rml::Vertex); + uint32_t offset = 0; + m_d3d_context->IASetVertexBuffers(0, 1, &geometry.vertex_buffer, &stride, &offset); + m_d3d_context->IASetIndexBuffer(geometry.index_buffer, DXGI_FORMAT_R32_UINT, 0); + m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); + m_d3d_context->DrawIndexed(geometry.index_count, 0, 0); + } + break; + case CompiledShaderType::Creation: + { + const double time = Rml::GetSystemInterface()->GetElapsedTime(); + + UseProgram(ProgramId::Creation); + + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return; + } + + // Get a pointer to the data in the constant buffer. + ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + + // Copy the data to the GPU + dataPtr->transform = m_transform; + dataPtr->translation = translation; + dataPtr->creation.value = (float)time; + dataPtr->creation.dimensions = shader.dimensions; + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + + // Issue draw call + m_d3d_context->IASetInputLayout(m_vertex_layout); + m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); + m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); + uint32_t stride = sizeof(Rml::Vertex); + uint32_t offset = 0; + m_d3d_context->IASetVertexBuffers(0, 1, &geometry.vertex_buffer, &stride, &offset); + m_d3d_context->IASetIndexBuffer(geometry.index_buffer, DXGI_FORMAT_R32_UINT, 0); + m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); + m_d3d_context->DrawIndexed(geometry.index_count, 0, 0); + } + break; + case CompiledShaderType::Invalid: + { + Rml::Log::Message(Rml::Log::LT_WARNING, "Unhandled render shader %d.", (int)type); + } + break; + } +} + +void RenderInterface_DX11::ReleaseShader(Rml::CompiledShaderHandle shader_handle) +{ + delete reinterpret_cast(shader_handle); +} + void RenderInterface_DX11::UpdateConstantBuffer() { if (m_cbuffer_dirty) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 014031cc0..f77a09db1 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -113,6 +113,10 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::TextureHandle SaveLayerAsTexture() override; // Rml::CompiledFilterHandle SaveLayerAsMaskImage() override; + Rml::CompiledShaderHandle CompileShader(const Rml::String& name, const Rml::Dictionary& parameters) override; + void RenderShader(Rml::CompiledShaderHandle shader_handle, Rml::CompiledGeometryHandle geometry_handle, Rml::Vector2f translation, + Rml::TextureHandle texture) override; + void ReleaseShader(Rml::CompiledShaderHandle effect_handle) override; // Can be passed to RenderGeometry() to enable texture rendering without changing the bound texture. static constexpr Rml::TextureHandle TextureEnableWithoutBinding = Rml::TextureHandle(-1); From 2b3fa50538f31e0477f492c08f9d161405f84f88 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 00:42:33 +0200 Subject: [PATCH 058/125] (chore): Snake case --- Backends/RmlUi_Renderer_DX11.cpp | 8 ++++---- Backends/RmlUi_Renderer_DX11.h | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e878fd35c..5f89589a1 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1691,13 +1691,13 @@ void RenderInterface_DX11::Clear() m_d3d_context->ClearRenderTargetView(m_bound_render_target, clearColor); } -void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blendState) +void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blend_state) { - if (blendState != m_current_blend_state) + if (blend_state != m_current_blend_state) { const float blend_factor[4] = {0.f, 0.f, 0.f, 0.f}; - m_d3d_context->OMSetBlendState(blendState, blend_factor, 0xFFFFFFFF); - m_current_blend_state = blendState; + m_d3d_context->OMSetBlendState(blend_state, blend_factor, 0xFFFFFFFF); + m_current_blend_state = blend_state; } } diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index f77a09db1..3b7107a3c 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -137,7 +137,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { private: // Changes blend state if necessary - void SetBlendState(ID3D11BlendState* blendState); + void SetBlendState(ID3D11BlendState* blend_state); void UpdateConstantBuffer(); void UseProgram(ProgramId program_id); From 0966a9730038ebe5d74337aa6411f356cfa65743 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 00:43:31 +0200 Subject: [PATCH 059/125] (feat): Implement filters (no rendering) --- Backends/RmlUi_Renderer_DX11.cpp | 151 ++++++++++++++++++++++++++++++- Backends/RmlUi_Renderer_DX11.h | 6 +- 2 files changed, 152 insertions(+), 5 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 5f89589a1..1e6309579 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -32,10 +32,12 @@ #include "RmlUi_Renderer_DX11.h" #include - #include - #include + #include #include + #include #include + #include + #include #include "RmlUi_Include_Windows.h" @@ -1603,6 +1605,14 @@ void RenderInterface_DX11::ReleaseTexture(Rml::TextureHandle texture_handle) DX_CLEANUP_RESOURCE_IF_CREATED(texture_view); } +static Rml::Colourf ConvertToColorf(Rml::ColourbPremultiplied c0) +{ + Rml::Colourf result; + for (int i = 0; i < 4; i++) + result[i] = (1.f / 255.f) * float(c0[i]); + return result; +} + void RenderInterface_DX11::DrawFullscreenQuad() { RenderGeometry(m_fullscreen_quad_geometry, {}, RenderInterface_DX11::TexturePostprocess); @@ -1821,7 +1831,24 @@ Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() return render_texture; } -/* +enum class FilterType { Invalid = 0, Passthrough, Blur, DropShadow, ColorMatrix, MaskImage }; +struct CompiledFilter { + FilterType type; + + // Passthrough + float blend_factor; + + // Blur + float sigma; + + // Drop shadow + Rml::Vector2f offset; + Rml::ColourbPremultiplied color; + + // ColorMatrix + Rml::Matrix4f color_matrix; +}; + Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() { BlitLayerToPostprocessPrimary(m_render_layers.GetTopLayerHandle()); @@ -1843,7 +1870,123 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() filter.type = FilterType::MaskImage; return reinterpret_cast(new CompiledFilter(std::move(filter))); } -*/ + +Rml::CompiledFilterHandle RenderInterface_DX11::CompileFilter(const Rml::String& name, const Rml::Dictionary& parameters) +{ + CompiledFilter filter = {}; + + if (name == "opacity") + { + filter.type = FilterType::Passthrough; + filter.blend_factor = Rml::Get(parameters, "value", 1.0f); + } + else if (name == "blur") + { + filter.type = FilterType::Blur; + filter.sigma = Rml::Get(parameters, "sigma", 1.0f); + } + else if (name == "drop-shadow") + { + filter.type = FilterType::DropShadow; + filter.sigma = Rml::Get(parameters, "sigma", 0.f); + filter.color = Rml::Get(parameters, "color", Rml::Colourb()).ToPremultiplied(); + filter.offset = Rml::Get(parameters, "offset", Rml::Vector2f(0.f)); + } + else if (name == "brightness") + { + filter.type = FilterType::ColorMatrix; + const float value = Rml::Get(parameters, "value", 1.0f); + filter.color_matrix = Rml::Matrix4f::Diag(value, value, value, 1.f); + } + else if (name == "contrast") + { + filter.type = FilterType::ColorMatrix; + const float value = Rml::Get(parameters, "value", 1.0f); + const float grayness = 0.5f - 0.5f * value; + filter.color_matrix = Rml::Matrix4f::Diag(value, value, value, 1.f); + filter.color_matrix.SetColumn(3, Rml::Vector4f(grayness, grayness, grayness, 1.f)); + } + else if (name == "invert") + { + filter.type = FilterType::ColorMatrix; + const float value = Rml::Math::Clamp(Rml::Get(parameters, "value", 1.0f), 0.f, 1.f); + const float inverted = 1.f - 2.f * value; + filter.color_matrix = Rml::Matrix4f::Diag(inverted, inverted, inverted, 1.f); + filter.color_matrix.SetColumn(3, Rml::Vector4f(value, value, value, 1.f)); + } + else if (name == "grayscale") + { + filter.type = FilterType::ColorMatrix; + const float value = Rml::Get(parameters, "value", 1.0f); + const float rev_value = 1.f - value; + const Rml::Vector3f gray = value * Rml::Vector3f(0.2126f, 0.7152f, 0.0722f); + // clang-format off + filter.color_matrix = Rml::Matrix4f::FromRows( + {gray.x + rev_value, gray.y, gray.z, 0.f}, + {gray.x, gray.y + rev_value, gray.z, 0.f}, + {gray.x, gray.y, gray.z + rev_value, 0.f}, + {0.f, 0.f, 0.f, 1.f} + ); + // clang-format on + } + else if (name == "sepia") + { + filter.type = FilterType::ColorMatrix; + const float value = Rml::Get(parameters, "value", 1.0f); + const float rev_value = 1.f - value; + const Rml::Vector3f r_mix = value * Rml::Vector3f(0.393f, 0.769f, 0.189f); + const Rml::Vector3f g_mix = value * Rml::Vector3f(0.349f, 0.686f, 0.168f); + const Rml::Vector3f b_mix = value * Rml::Vector3f(0.272f, 0.534f, 0.131f); + // clang-format off + filter.color_matrix = Rml::Matrix4f::FromRows( + {r_mix.x + rev_value, r_mix.y, r_mix.z, 0.f}, + {g_mix.x, g_mix.y + rev_value, g_mix.z, 0.f}, + {b_mix.x, b_mix.y, b_mix.z + rev_value, 0.f}, + {0.f, 0.f, 0.f, 1.f} + ); + // clang-format on + } + else if (name == "hue-rotate") + { + // Hue-rotation and saturation values based on: https://www.w3.org/TR/filter-effects-1/#attr-valuedef-type-huerotate + filter.type = FilterType::ColorMatrix; + const float value = Rml::Get(parameters, "value", 1.0f); + const float s = Rml::Math::Sin(value); + const float c = Rml::Math::Cos(value); + // clang-format off + filter.color_matrix = Rml::Matrix4f::FromRows( + {0.213f + 0.787f * c - 0.213f * s, 0.715f - 0.715f * c - 0.715f * s, 0.072f - 0.072f * c + 0.928f * s, 0.f}, + {0.213f - 0.213f * c + 0.143f * s, 0.715f + 0.285f * c + 0.140f * s, 0.072f - 0.072f * c - 0.283f * s, 0.f}, + {0.213f - 0.213f * c - 0.787f * s, 0.715f - 0.715f * c + 0.715f * s, 0.072f + 0.928f * c + 0.072f * s, 0.f}, + {0.f, 0.f, 0.f, 1.f} + ); + // clang-format on + } + else if (name == "saturate") + { + filter.type = FilterType::ColorMatrix; + const float value = Rml::Get(parameters, "value", 1.0f); + // clang-format off + filter.color_matrix = Rml::Matrix4f::FromRows( + {0.213f + 0.787f * value, 0.715f - 0.715f * value, 0.072f - 0.072f * value, 0.f}, + {0.213f - 0.213f * value, 0.715f + 0.285f * value, 0.072f - 0.072f * value, 0.f}, + {0.213f - 0.213f * value, 0.715f - 0.715f * value, 0.072f + 0.928f * value, 0.f}, + {0.f, 0.f, 0.f, 1.f} + ); + // clang-format on + } + + if (filter.type != FilterType::Invalid) + return reinterpret_cast(new CompiledFilter(std::move(filter))); + + Rml::Log::Message(Rml::Log::LT_WARNING, "Unsupported filter type '%s'.", name.c_str()); + return {}; +} + +void RenderInterface_DX11::ReleaseFilter(Rml::CompiledFilterHandle filter) +{ + delete reinterpret_cast(filter); +} enum class CompiledShaderType { Invalid = 0, Gradient, Creation }; struct CompiledShader { diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 3b7107a3c..a5f4d4511 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -112,7 +112,11 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::TextureHandle SaveLayerAsTexture() override; - // Rml::CompiledFilterHandle SaveLayerAsMaskImage() override; + Rml::CompiledFilterHandle SaveLayerAsMaskImage() override; + + Rml::CompiledFilterHandle CompileFilter(const Rml::String& name, const Rml::Dictionary& parameters) override; + void ReleaseFilter(Rml::CompiledFilterHandle filter) override; + Rml::CompiledShaderHandle CompileShader(const Rml::String& name, const Rml::Dictionary& parameters) override; void RenderShader(Rml::CompiledShaderHandle shader_handle, Rml::CompiledGeometryHandle geometry_handle, Rml::Vector2f translation, Rml::TextureHandle texture) override; From 565a4083e80154d2df0d0b92a502622d80100ca6 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 00:43:54 +0200 Subject: [PATCH 060/125] (fix): Convert filter rendering to D3D11 --- Backends/RmlUi_Renderer_DX11.cpp | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 1e6309579..2f1a7b326 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1856,15 +1856,16 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); const Gfx::RenderTargetData& destination = m_render_layers.GetBlendMask(); - glBindFramebuffer(GL_FRAMEBUFFER, destination.framebuffer); - BindTexture(source); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, nullptr); + Gfx::BindTexture(m_d3d_context, source); UseProgram(ProgramId::Passthrough); - glDisable(GL_BLEND); + ID3D11BlendState* blend_state_backup = m_current_blend_state; + SetBlendState(m_blend_state_disable); DrawFullscreenQuad(); - glEnable(GL_BLEND); - glBindFramebuffer(GL_FRAMEBUFFER, m_render_layers.GetTopLayer().framebuffer); + SetBlendState(blend_state_backup); + m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); CompiledFilter filter = {}; filter.type = FilterType::MaskImage; From 4b73b90b36c49415dba2a6c41d279f16c425de77 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 00:44:09 +0200 Subject: [PATCH 061/125] (fix): Bind cbuffer to fragment shader --- Backends/RmlUi_Renderer_DX11.cpp | 1 + 1 file changed, 1 insertion(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 2f1a7b326..afb9ba0e1 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1401,6 +1401,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm m_d3d_context->IASetInputLayout(m_vertex_layout); m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); + m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); // Bind vertex and index buffers, issue draw call uint32_t stride = sizeof(Rml::Vertex); From 35b94ec6c1ebc62489605852617c78bd8a5a543b Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 12:25:03 +0200 Subject: [PATCH 062/125] (fix): Correct BindTexture implementation --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index afb9ba0e1..8ced8013e 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -794,9 +794,9 @@ static void DestroyRenderTarget(RenderTargetData& rt) rt = {}; } -static void BindTexture(ID3D11DeviceContext* context, const RenderTargetData& rt, uint32_t slot = 1) +static void BindTexture(ID3D11DeviceContext* context, const RenderTargetData& rt, uint32_t slot = 0) { - context->PSSetShaderResources(0, slot, &rt.render_target_shader_resource_view); + context->PSSetShaderResources(slot, 1, &rt.render_target_shader_resource_view); } static bool CreateShaders(ID3D11Device* p_device, ProgramData& data) From adb96d4d03d0923fdcbc255a3e98fa1739aaaa2a Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 12:25:39 +0200 Subject: [PATCH 063/125] (chore): Snake case --- Backends/RmlUi_Renderer_DX11.cpp | 6 +++--- Backends/RmlUi_Renderer_DX11.h | 14 +++++++------- 2 files changed, 10 insertions(+), 10 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 8ced8013e..e3cbac44e 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1030,7 +1030,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* samplerDesc.MinLOD = 0; samplerDesc.MaxLOD = D3D11_FLOAT32_MAX; - HRESULT result = m_d3d_device->CreateSamplerState(&samplerDesc, &m_samplerState); + HRESULT result = m_d3d_device->CreateSamplerState(&samplerDesc, &m_sampler_state); RMLUI_DX_ASSERTMSG(result, "failed to CreateSamplerState"); if (FAILED(result)) { @@ -1089,7 +1089,7 @@ void RenderInterface_DX11::Cleanup() } // Cleans up all general resources - DX_CLEANUP_RESOURCE_IF_CREATED(m_samplerState); + DX_CLEANUP_RESOURCE_IF_CREATED(m_sampler_state); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_enable); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable_color); @@ -1389,7 +1389,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture; m_d3d_context->PSSetShaderResources(0, 1, &texture_view); } - m_d3d_context->PSSetSamplers(0, 1, &m_samplerState); + m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); } else { diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index a5f4d4511..c12eef8eb 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -172,7 +172,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { // Shaders ID3D11Buffer* m_shader_buffer = nullptr; bool m_cbuffer_dirty = true; - ID3D11SamplerState* m_samplerState = nullptr; + ID3D11SamplerState* m_sampler_state = nullptr; // Viewport dimensions int m_viewport_width = 0; @@ -252,16 +252,16 @@ class RenderInterface_DX11 : public Rml::RenderInterface { struct Blur { int _padding[18]; Rml::Vector4f weights; - Rml::Vector2f texelOffset; - Rml::Vector2f texCoordMin; - Rml::Vector2f texCoordMax; + Rml::Vector2f texel_offset; + Rml::Vector2f texcoord_min; + Rml::Vector2f texcoord_max; } blur; struct DropShadow { int _padding[18]; - Rml::Vector2f texCoordMin; - Rml::Vector2f texCoordMax; + Rml::Vector2f texcoord_min; + Rml::Vector2f texcoord_max; Rml::Vector4f color; - } dropShadow; + } drop_shadow; struct Creation { int _padding[18]; Rml::Vector2f dimensions; From 91262aa7351daed91a255039cf2083159b34b7b3 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 13:15:20 +0200 Subject: [PATCH 064/125] (feat): Implement filters and blurring --- Backends/RmlUi_Renderer_DX11.cpp | 418 ++++++++++++++++++++++++++++++- Backends/RmlUi_Renderer_DX11.h | 4 + 2 files changed, 418 insertions(+), 4 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e3cbac44e..5d1a48b64 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -930,6 +930,24 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* return; } #endif + + blend_desc.RenderTarget[0].BlendEnable = true; + blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_SRC_COLOR; + blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_ZERO; + blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_SRC_ALPHA; + blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_ZERO; + blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; + result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_color_filter); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); + return; + } + #endif } // Scissor regions require a rasterizer state. Cache one for scissor on and off @@ -1043,12 +1061,14 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* D3D11_DEPTH_STENCIL_DESC desc; ZeroMemory(&desc, sizeof(desc)); desc.DepthEnable = false; - desc.FrontFace.StencilFunc = D3D11_COMPARISON_EQUAL; + desc.StencilReadMask = D3D11_DEFAULT_STENCIL_READ_MASK; + desc.StencilWriteMask = 0; + desc.FrontFace.StencilFunc = D3D11_COMPARISON_LESS_EQUAL; desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; desc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; desc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; desc.BackFace = desc.FrontFace; - desc.StencilEnable = false; + desc.StencilEnable = true; // Disabled m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_disable); @@ -1092,6 +1112,7 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_sampler_state); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_enable); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_color_filter); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable_color); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_intersect); @@ -1749,8 +1770,7 @@ void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml:: BlitLayerToPostprocessPrimary(source_handle); // Render the filters, the PostprocessPrimary framebuffer is used for both input and output. - // @TODO: RenderFilters implementation - // RenderFilters(filters); + RenderFilters(filters); // Render to the destination layer. m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetLayer(destination_handle).render_target_view, @@ -1784,7 +1804,9 @@ Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() Rml::TextureHandle render_texture = GenerateTexture({}, bounds.Size()); if (!render_texture) + { return {}; + } BlitLayerToPostprocessPrimary(m_render_layers.GetTopLayerHandle()); @@ -2184,6 +2206,394 @@ void RenderInterface_DX11::ReleaseShader(Rml::CompiledShaderHandle shader_handle delete reinterpret_cast(shader_handle); } +static void SigmaToParameters(const float desired_sigma, int& out_pass_level, float& out_sigma) +{ + constexpr int max_num_passes = 10; + static_assert(max_num_passes < 31, ""); + constexpr float max_single_pass_sigma = 3.0f; + out_pass_level = Rml::Math::Clamp(Rml::Math::Log2(int(desired_sigma * (2.f / max_single_pass_sigma))), 0, max_num_passes); + out_sigma = Rml::Math::Clamp(desired_sigma / float(1 << out_pass_level), 0.0f, max_single_pass_sigma); +} + +static void SetTexCoordLimits(Rml::Vector2f& p_tex_coord_min, Rml::Vector2f& p_tex_coord_max, Rml::Rectanglei rectangle_flipped, + Rml::Vector2i framebuffer_size) +{ + // Offset by half-texel values so that texture lookups are clamped to fragment centers, thereby avoiding color + // bleeding from neighboring texels due to bilinear interpolation. + const Rml::Vector2f min = (Rml::Vector2f(rectangle_flipped.p0) + Rml::Vector2f(0.5f)) / Rml::Vector2f(framebuffer_size); + const Rml::Vector2f max = (Rml::Vector2f(rectangle_flipped.p1) - Rml::Vector2f(0.5f)) / Rml::Vector2f(framebuffer_size); + + p_tex_coord_min.x = min.x; + p_tex_coord_min.y = min.y; + p_tex_coord_max.x = max.x; + p_tex_coord_max.y = max.y; +} + +static void SetBlurWeights(Rml::Vector4f& p_weights, float sigma) +{ + constexpr int num_weights = BLUR_NUM_WEIGHTS; + float weights[num_weights]; + float normalization = 0.0f; + for (int i = 0; i < num_weights; i++) + { + if (Rml::Math::Absolute(sigma) < 0.1f) + weights[i] = float(i == 0); + else + weights[i] = Rml::Math::Exp(-float(i * i) / (2.0f * sigma * sigma)) / (Rml::Math::SquareRoot(2.f * Rml::Math::RMLUI_PI) * sigma); + + normalization += (i == 0 ? 1.f : 2.0f) * weights[i]; + } + for (int i = 0; i < num_weights; i++) + weights[i] /= normalization; + + p_weights.x = weights[0]; + p_weights.y = weights[1]; + p_weights.z = weights[2]; + p_weights.w = weights[3]; +} + +void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& source_destination, const Gfx::RenderTargetData& temp, + const Rml::Rectanglei window_flipped) +{ + // @TODO: Fix blur + return; + RMLUI_ASSERT(&source_destination != &temp && source_destination.width == temp.width && source_destination.height == temp.height); + RMLUI_ASSERT(window_flipped.Valid()); + + int pass_level = 0; + SigmaToParameters(sigma, pass_level, sigma); + + const Rml::Rectanglei original_scissor = m_scissor_state; + + // Begin by downscaling so that the blur pass can be done at a reduced resolution for large sigma. + Rml::Rectanglei scissor = window_flipped; + + UseProgram(ProgramId::Passthrough); + SetScissor(scissor, true); + + // Downscale by iterative half-scaling with bilinear filtering, to reduce aliasing. + D3D11_VIEWPORT d3dviewport; + d3dviewport.TopLeftX = 0; + d3dviewport.TopLeftY = 0; + d3dviewport.Width = source_destination.width / 2; + d3dviewport.Height = source_destination.height / 2; + d3dviewport.MinDepth = 0.0f; + d3dviewport.MaxDepth = 1.0f; + m_d3d_context->RSSetViewports(1, &d3dviewport); + + // Scale UVs if we have even dimensions, such that texture fetches align perfectly between texels, thereby producing a 50% blend of + // neighboring texels. + const Rml::Vector2f uv_scaling = {(source_destination.width % 2 == 1) ? (1.f - 1.f / float(source_destination.width)) : 1.f, + (source_destination.height % 2 == 1) ? (1.f - 1.f / float(source_destination.height)) : 1.f}; + + for (int i = 0; i < pass_level; i++) + { + scissor.p0 = (scissor.p0 + Rml::Vector2i(1)) / 2; + scissor.p1 = Rml::Math::Max(scissor.p1 / 2, scissor.p0); + const bool from_source = (i % 2 == 0); + Gfx::BindTexture(m_d3d_context, from_source ? source_destination : temp); + m_d3d_context->OMSetRenderTargets(1, &(from_source ? temp : source_destination).render_target_view, source_destination.depth_stencil_view); + SetScissor(scissor, true); + + DrawFullscreenQuad({}, uv_scaling); + } + + d3dviewport.TopLeftX = 0; + d3dviewport.TopLeftY = 0; + d3dviewport.Width = source_destination.width; + d3dviewport.Height = source_destination.height; + d3dviewport.MinDepth = 0.0f; + d3dviewport.MaxDepth = 1.0f; + m_d3d_context->RSSetViewports(1, &d3dviewport); + + // Ensure texture data end up in the temp buffer. Depending on the last downscaling, we might need to move it from the source_destination buffer. + const bool transfer_to_temp_buffer = (pass_level % 2 == 0); + if (transfer_to_temp_buffer) + { + Gfx::BindTexture(m_d3d_context, source_destination); + m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, nullptr); + DrawFullscreenQuad(); + } + + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return; + } + + // Get a pointer to the data in the constant buffer. + ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + + // Copy the data to the GPU + dataPtr->transform = m_transform; + dataPtr->translation = m_translation; + + SetBlurWeights(dataPtr->blur.weights, sigma); + + // Set up uniforms. + UseProgram(ProgramId::Blur); + SetBlurWeights(dataPtr->blur.weights, sigma); + SetTexCoordLimits(dataPtr->blur.texcoord_min, dataPtr->blur.texcoord_max, scissor, + {source_destination.width, source_destination.height}); + + auto SetTexelOffset = [dataPtr](Rml::Vector2f blur_direction, int texture_dimension) { + const Rml::Vector2f texel_offset = blur_direction * (1.0f / float(texture_dimension)); + dataPtr->blur.texel_offset = texel_offset; + }; + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + + // Blur render pass - vertical. + Gfx::BindTexture(m_d3d_context, temp); + m_d3d_context->OMSetRenderTargets(1, &source_destination.render_target_view, nullptr); + + SetTexelOffset({0.f, 1.f}, temp.height); + DrawFullscreenQuad(); + + // Blur render pass - horizontal. + Gfx::BindTexture(m_d3d_context, source_destination); + m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, nullptr); + + // Add a 1px transparent border around the blur region by first clearing with a padded scissor. This helps prevent + // artifacts when upscaling the blur result in the later step. On Intel and AMD, we have observed that during + // blitting with linear filtering, pixels outside the 'src' region can be blended into the output. On the other + // hand, it looks like Nvidia clamps the pixels to the source edge, which is what we really want. Regardless, we + // work around the issue with this extra step. + SetScissor(scissor.Extend(1), true); + float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; + m_d3d_context->ClearRenderTargetView(temp.render_target_view, clearColor); + SetScissor(scissor, true); + + SetTexelOffset({1.f, 0.f}, source_destination.width); + DrawFullscreenQuad(); + + // Blit the blurred image to the scissor region with upscaling. + SetScissor(window_flipped, true); + + const Rml::Vector2i src_min = scissor.p0; + const Rml::Vector2i src_max = scissor.p1; + const Rml::Vector2i dst_min = window_flipped.p0; + const Rml::Vector2i dst_max = window_flipped.p1; + + D3D11_BOX srcBox; + srcBox.left = src_min.x; + srcBox.top = src_min.y; + srcBox.front = 0; + srcBox.right = src_max.x; + srcBox.bottom = src_max.y; + srcBox.back = 1; + + m_d3d_context->CopySubresourceRegion(source_destination.render_target_texture, 0, dst_min.x, dst_min.y, 0, temp.render_target_texture, 0, + &srcBox); + + // The above upscale blit might be jittery at low resolutions (large pass levels). This is especially noticeable when moving an element with + // backdrop blur around or when trying to click/hover an element within a blurred region since it may be rendered at an offset. For more stable + // and accurate rendering we next upscale the blur image by an exact power-of-two. However, this may not fill the edges completely so we need to + // do the above first. Note that this strategy may sometimes result in visible seams. Alternatively, we could try to enlarge the window to the + // next power-of-two size and then downsample and blur that. + const Rml::Vector2i target_min = src_min * (1 << pass_level); + const Rml::Vector2i target_max = src_max * (1 << pass_level); + if (target_min != dst_min || target_max != dst_max) + { + Gfx::BindTexture(m_d3d_context, temp); + + D3D11_VIEWPORT viewport; + viewport.TopLeftX = target_min.x; + viewport.TopLeftY = target_min.y; + viewport.Width = target_max.x - target_min.x; + viewport.Height = target_max.y - target_min.y; + viewport.MinDepth = 0.0f; + viewport.MaxDepth = 1.0f; + m_d3d_context->RSSetViewports(1, &viewport); + UseProgram(ProgramId::Passthrough); + DrawFullscreenQuad(); + // Restore viewport + m_d3d_context->RSSetViewports(1, &d3dviewport); + + + // glBlitFramebuffer(src_min.x, src_min.y, src_max.x, src_max.y, target_min.x, target_min.y, target_max.x, target_max.y, GL_COLOR_BUFFER_BIT, + // GL_LINEAR); + } + + // Restore render state. + SetScissor(original_scissor); +} + +void RenderInterface_DX11::RenderFilters(Rml::Span filter_handles) +{ + m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); + for (const Rml::CompiledFilterHandle filter_handle : filter_handles) + { + const CompiledFilter& filter = *reinterpret_cast(filter_handle); + const FilterType type = filter.type; + + switch (type) + { + case FilterType::Passthrough: + { + UseProgram(ProgramId::Passthrough); + const float blend_factor[4] = {filter.blend_factor, filter.blend_factor, filter.blend_factor, filter.blend_factor}; + m_d3d_context->OMSetBlendState(m_blend_state_color_filter, blend_factor, 0xFFFFFFFF); + m_current_blend_state = m_blend_state_color_filter; + + const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); + Gfx::BindTexture(m_d3d_context, source); + m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, nullptr); + + DrawFullscreenQuad(); + + m_render_layers.SwapPostprocessPrimarySecondary(); + SetBlendState(m_blend_state_enable); + } + break; + case FilterType::Blur: + { + return; + ID3D11BlendState* blend_state_backup = m_current_blend_state; + SetBlendState(m_blend_state_disable); + + const Gfx::RenderTargetData& source_destination = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& temp = m_render_layers.GetPostprocessSecondary(); + + const Rml::Rectanglei window_flipped = VerticallyFlipped(m_scissor_state, m_viewport_height); + RenderBlur(filter.sigma, source_destination, temp, window_flipped); + + SetBlendState(blend_state_backup); + } + break; + case FilterType::DropShadow: + { + return; + + UseProgram(ProgramId::DropShadow); + ID3D11BlendState* blend_state_backup = m_current_blend_state; + SetBlendState(m_blend_state_disable); + + + const Gfx::RenderTargetData& primary = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& secondary = m_render_layers.GetPostprocessSecondary(); + Gfx::BindTexture(m_d3d_context, primary); + m_d3d_context->OMSetRenderTargets(1, &secondary.render_target_view, nullptr); + + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return; + } + + // Get a pointer to the data in the constant buffer. + ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + + Rml::Colourf color = ConvertToColorf(filter.color); + + // Copy the data to the GPU + dataPtr->transform = m_transform; + dataPtr->translation = m_translation; + dataPtr->drop_shadow.color.x = color.red; + dataPtr->drop_shadow.color.y = color.green; + dataPtr->drop_shadow.color.z = color.blue; + dataPtr->drop_shadow.color.w = color.alpha; + + const Rml::Rectanglei window_flipped = VerticallyFlipped(m_scissor_state, m_viewport_height); + SetTexCoordLimits(dataPtr->drop_shadow.texcoord_min, dataPtr->drop_shadow.texcoord_max, window_flipped, + {primary.width, primary.height}); + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + + const Rml::Vector2f uv_offset = filter.offset / Rml::Vector2f(-(float)m_viewport_width, (float)m_viewport_height); + DrawFullscreenQuad(uv_offset); + + if (filter.sigma >= 0.5f) + { + const Gfx::RenderTargetData& tertiary = m_render_layers.GetPostprocessTertiary(); + RenderBlur(filter.sigma, secondary, tertiary, window_flipped); + } + + UseProgram(ProgramId::Passthrough); + BindTexture(m_d3d_context, primary); + SetBlendState(blend_state_backup); + DrawFullscreenQuad(); + + m_render_layers.SwapPostprocessPrimarySecondary(); + } + break; + case FilterType::ColorMatrix: + { + return; + + UseProgram(ProgramId::ColorMatrix); + ID3D11BlendState* blend_state_backup = m_current_blend_state; + SetBlendState(m_blend_state_disable); + + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + return; + } + + // Get a pointer to the data in the constant buffer. + ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + + // Copy the data to the GPU + dataPtr->transform = m_transform; + dataPtr->translation = m_translation; + memset(&dataPtr->color_matrix.color_matrix, 0, sizeof(dataPtr->color_matrix.color_matrix)); + memcpy_s(&dataPtr->color_matrix.color_matrix, sizeof(dataPtr->color_matrix.color_matrix), filter.color_matrix.data(), + sizeof(dataPtr->color_matrix.color_matrix)); + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + + const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); + Gfx::BindTexture(m_d3d_context, source); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, nullptr); + + DrawFullscreenQuad(); + + m_render_layers.SwapPostprocessPrimarySecondary(); + SetBlendState(blend_state_backup); + } + break; + case FilterType::MaskImage: + { + UseProgram(ProgramId::BlendMask); + ID3D11BlendState* blend_state_backup = m_current_blend_state; + SetBlendState(m_blend_state_disable); + + const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& blend_mask = m_render_layers.GetBlendMask(); + const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); + + Gfx::BindTexture(m_d3d_context, source, 0); + Gfx::BindTexture(m_d3d_context, blend_mask, 1); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, nullptr); + + DrawFullscreenQuad(); + + m_render_layers.SwapPostprocessPrimarySecondary(); + SetBlendState(blend_state_backup); + } + break; + case FilterType::Invalid: + { + Rml::Log::Message(Rml::Log::LT_WARNING, "Unhandled render filter %d.", (int)type); + } + break; + } + } +} + void RenderInterface_DX11::UpdateConstantBuffer() { if (m_cbuffer_dirty) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index c12eef8eb..62bb2520f 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -146,12 +146,15 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void UseProgram(ProgramId program_id); void BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle); + void RenderFilters(Rml::Span filter_handles); void SetScissor(Rml::Rectanglei region, bool vertically_flip = false); void DrawFullscreenQuad(); void DrawFullscreenQuad(Rml::Vector2f uv_offset, Rml::Vector2f uv_scaling = Rml::Vector2f(1.f)); + void RenderBlur(float sigma, const Gfx::RenderTargetData& source_destination, const Gfx::RenderTargetData& temp, Rml::Rectanglei window_flipped); + private: // D3D11 core resources ID3D11Device* m_d3d_device = nullptr; @@ -186,6 +189,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11BlendState* m_blend_state_enable = nullptr; ID3D11BlendState* m_blend_state_disable = nullptr; ID3D11BlendState* m_blend_state_disable_color = nullptr; + ID3D11BlendState* m_blend_state_color_filter = nullptr; ProgramId active_program = {}; Rml::Rectanglei m_scissor_state = Rml::Rectanglei::MakeInvalid(); From 69e44604e6c25d9c88b0934262a92390b8a75af0 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 13:16:32 +0200 Subject: [PATCH 065/125] (feat): Implement debug markers Temp to fix the rendering bugs --- Backends/RmlUi_Renderer_DX11.cpp | 42 ++++++++++++++++++++++++++++++++ Backends/RmlUi_Renderer_DX11.h | 12 +++++++++ 2 files changed, 54 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 5d1a48b64..3575519f5 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -874,6 +874,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* // Assign D3D resources m_d3d_device = p_d3d_device; m_d3d_context = p_d3d_device_context; + m_d3d_context->QueryInterface(__uuidof(ID3DUserDefinedAnnotation), (void**)&m_debug); + DebugScope::AttachDebugLayer(m_debug); m_render_layers.SetD3DResources(m_d3d_device); // Pre-cache quad for blitting @@ -1121,6 +1123,7 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); + DX_CLEANUP_RESOURCE_IF_CREATED(m_debug); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) @@ -1741,6 +1744,7 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle) { + DebugScope scope(L"BlitLayerToPostprocessPrimary"); const Gfx::RenderTargetData& source = m_render_layers.GetLayer(layer_handle); const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessPrimary(); @@ -1756,12 +1760,14 @@ Rml::LayerHandle RenderInterface_DX11::PushLayer() float colors[4] = {0, 0, 0, 0}; m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, colors); + m_debug->BeginEvent(L"BeginLayer"); return layer_handle; } void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml::LayerHandle destination_handle, Rml::BlendMode blend_mode, Rml::Span filters) { + DebugScope scope(L"CompositeLayers"); using Rml::BlendMode; // Blit source layer to postprocessing buffer. Do this regardless of whether we actually have any filters to be @@ -1794,11 +1800,13 @@ void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml:: void RenderInterface_DX11::PopLayer() { m_render_layers.PopLayer(); + m_debug->EndEvent(); m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); } Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() { + DebugScope scope(L"SaveLayerAsTexture"); RMLUI_ASSERT(m_scissor_state.Valid()); const Rml::Rectanglei bounds = m_scissor_state; @@ -1874,6 +1882,7 @@ struct CompiledFilter { Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() { + DebugScope scope(L"SaveLayerAsMaskImage"); BlitLayerToPostprocessPrimary(m_render_layers.GetTopLayerHandle()); const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); @@ -2096,6 +2105,7 @@ Rml::CompiledShaderHandle RenderInterface_DX11::CompileShader(const Rml::String& void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, Rml::CompiledGeometryHandle geometry_handle, Rml::Vector2f translation, Rml::TextureHandle /*texture*/) { + DebugScope scope(L"RenderShader"); RMLUI_ASSERT(shader_handle && geometry_handle); const CompiledShader& shader = *reinterpret_cast(shader_handle); const CompiledShaderType type = shader.type; @@ -2110,6 +2120,7 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, { case CompiledShaderType::Gradient: { + DebugScope scope_gradient(L"Gradient"); RMLUI_ASSERT(shader.stop_positions.size() == shader.stop_colors.size()); const int num_stops = (int)shader.stop_positions.size(); @@ -2157,6 +2168,7 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, break; case CompiledShaderType::Creation: { + DebugScope scope_creation(L"Creation"); const double time = Rml::GetSystemInterface()->GetElapsedTime(); UseProgram(ProgramId::Creation); @@ -2257,6 +2269,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& { // @TODO: Fix blur return; + DebugScope scope(L"RenderBlur"); RMLUI_ASSERT(&source_destination != &temp && source_destination.width == temp.width && source_destination.height == temp.height); RMLUI_ASSERT(window_flipped.Valid()); @@ -2424,6 +2437,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& void RenderInterface_DX11::RenderFilters(Rml::Span filter_handles) { + DebugScope scope(L"RenderFilters"); m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); for (const Rml::CompiledFilterHandle filter_handle : filter_handles) { @@ -2434,6 +2448,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetBlendState(m_blend_state_color_filter, blend_factor, 0xFFFFFFFF); @@ -2469,6 +2484,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanEndEvent(); +} +RenderInterface_DX11::DebugScope::DebugScope(LPCWSTR name) +{ + m_debug_scope_debug_device->BeginEvent(name); +} +void RenderInterface_DX11::DebugScope::operator()(LPCWSTR name) const +{ + m_debug_scope_debug_device->BeginEvent(name); +} +void RenderInterface_DX11::DebugScope::AttachDebugLayer(ID3DUserDefinedAnnotation* debug) +{ + m_debug_scope_debug_device = debug; +} #endif // RMLUI_PLATFORM_WIN32 diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 62bb2520f..c948d4f6a 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -36,6 +36,7 @@ #include #include + #include #ifdef RMLUI_DEBUG #define RMLUI_DX_ASSERTMSG(statement, msg) RMLUI_ASSERTMSG(SUCCEEDED(statement), msg) @@ -328,6 +329,17 @@ class RenderInterface_DX11 : public Rml::RenderInterface { }; RenderLayerStack m_render_layers; + + ID3DUserDefinedAnnotation* m_debug = nullptr; + + class DebugScope { + private: + public: + DebugScope(LPCWSTR scopeName); + void operator()(LPCWSTR scopeName) const; + ~DebugScope(); + static void AttachDebugLayer(ID3DUserDefinedAnnotation* debug); + }; }; #endif // RMLUI_PLATFORM_WIN32 From 87e061134a277939ecf62fb7f09ab5bdf1895515 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 27 Sep 2024 13:18:00 +0200 Subject: [PATCH 066/125] (chore): Spaces not tabs --- Backends/RmlUi_Renderer_DX11.h | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index c948d4f6a..8cda74876 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -333,13 +333,13 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3DUserDefinedAnnotation* m_debug = nullptr; class DebugScope { - private: - public: - DebugScope(LPCWSTR scopeName); - void operator()(LPCWSTR scopeName) const; - ~DebugScope(); - static void AttachDebugLayer(ID3DUserDefinedAnnotation* debug); - }; + private: + public: + DebugScope(LPCWSTR scopeName); + void operator()(LPCWSTR scopeName) const; + ~DebugScope(); + static void AttachDebugLayer(ID3DUserDefinedAnnotation* debug); + }; }; #endif // RMLUI_PLATFORM_WIN32 From ba0dbb5169704412a5d55813047fe2b0f901999e Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 28 Sep 2024 13:22:57 +0200 Subject: [PATCH 067/125] (chore): Simplify shader structure names --- Backends/RmlUi_Renderer_DX11.cpp | 133 +++++++++++++++---------------- 1 file changed, 66 insertions(+), 67 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 3575519f5..31b739b16 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -62,16 +62,16 @@ constexpr const char shader_vert_main[] = RMLUI_SHADER_HEADER R"( struct VS_Input { - float2 inPosition : POSITION; - float4 inColor : COLOR; - float2 inTexCoord : TEXCOORD; + float2 position : POSITION; + float4 color : COLOR; + float2 uv : TEXCOORD; }; struct PS_INPUT { - float4 outPosition : SV_Position; - float4 outColor : COLOR; - float2 outUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; cbuffer SharedConstantBuffer : register(b0) @@ -86,16 +86,16 @@ PS_INPUT VSMain(const VS_Input IN) { PS_INPUT result = (PS_INPUT)0; - float2 translatedPos = IN.inPosition + m_translate; + float2 translatedPos = IN.position + m_translate; float4 resPos = mul(m_transform, float4(translatedPos.x, translatedPos.y, 0.0, 1.0)); - result.outPosition = resPos; - result.outColor = IN.inColor; - result.outUV = IN.inTexCoord; + result.position = resPos; + result.color = IN.color; + result.uv = IN.uv; #if defined(RMLUI_PREMULTIPLIED_ALPHA) // Pre-multiply vertex colors with their alpha. - result.outColor.rgb = result.outColor.rgb * result.outColor.a; + result.color.rgb = result.color.rgb * result.color.a; #endif return result; @@ -105,9 +105,9 @@ PS_INPUT VSMain(const VS_Input IN) constexpr const char shader_frag_texture[] = RMLUI_SHADER_HEADER R"( struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; Texture2D g_InputTexture : register(t0); @@ -115,21 +115,21 @@ SamplerState g_SamplerLinear : register(s0); float4 PSMain(const PS_Input IN) : SV_TARGET { - return IN.inputColor * g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); + return IN.color * g_InputTexture.Sample(g_SamplerLinear, IN.uv); }; )"; constexpr const char shader_frag_color[] = RMLUI_SHADER_HEADER R"( struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; float4 PSMain(const PS_Input IN) : SV_TARGET { - return IN.inputColor; + return IN.csolor; }; )"; @@ -172,9 +172,9 @@ cbuffer SharedConstantBuffer : register(b0) struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; #define glsl_mod(x,y) (((x)-(y)*floor((x)/(y)))) @@ -194,16 +194,16 @@ float4 PSMain(const PS_Input IN) : SV_TARGET if (m_func == LINEAR || m_func == REPEATING_LINEAR) { float dist_square = dot(m_v, m_v); - float2 V = IN.inputPos.xy - m_p; + float2 V = IN.uv.xy - m_p; t = dot(m_v, V) / dist_square; } else if (m_func == RADIAL || m_func == REPEATING_RADIAL) { - float2 V = IN.inputPos.xy - m_p; + float2 V = IN.uv.xy - m_p; t = length(m_v * V); } else if (m_func == CONIC || m_func == REPEATING_CONIC) { float2x2 R = float2x2(m_v.x, -m_v.y, m_v.y, m_v.x); - float2 V = mul(R, (IN.inputPos.xy - m_p)); + float2 V = mul(R, (IN.uv.xy - m_p)); t = 0.5 + atan2(V.y, -V.x) / (2.0 * PI); } @@ -213,7 +213,7 @@ float4 PSMain(const PS_Input IN) : SV_TARGET t = t0 + glsl_mod(t - t0, t1 - t0); } - return IN.inputColor * lerp_stop_colors(t); + return IN.color * lerp_stop_colors(t); }; )"; @@ -221,9 +221,9 @@ float4 PSMain(const PS_Input IN) : SV_TARGET static const char shader_frag_creation[] = RMLUI_SHADER_HEADER R"( struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; cbuffer SharedConstantBuffer : register(b0) @@ -244,7 +244,7 @@ float4 PSMain(const PS_Input IN) : SV_TARGET float3 c; float l; for (int i = 0; i < 3; i++) { - float2 p = IN.inputUV; + float2 p = IN.uv; float2 uv = p; p -= .5; p.x *= m_dimensions.x / m_dimensions.y; @@ -253,23 +253,23 @@ float4 PSMain(const PS_Input IN) : SV_TARGET uv += p / l * (sin(z) + 1.) * abs(sin(l * 9. - z - z)); c[i] = .01 / length(glsl_mod(uv, 1.) - .5); } - return float4(c / l, IN.inputColor.a); + return float4(c / l, IN.color.a); }; )"; constexpr const char shader_vert_passthrough[] = RMLUI_SHADER_HEADER R"( struct VS_Input { - float2 inPosition : POSITION; - float4 inColor : COLOR; - float2 inTexCoord : TEXCOORD; + float2 position : POSITION; + float4 color : COLOR; + float2 uv : TEXCOORD; }; struct PS_Input { - float4 outPosition : SV_Position; - float4 outColor : COLOR; - float2 outUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; cbuffer ConstantBuffer : register(b0) @@ -284,8 +284,8 @@ PS_Input VSMain(const VS_Input IN) { PS_Input result = (PS_Input)0; - result.outPosition = float4(IN.inPosition.xy, 0.0f, 1.0f); - result.outUV = float2(IN.inTexCoord.x, 1.0f - IN.inTexCoord.y); + result.position = float4(IN.position.xy, 0.0f, 1.0f); + result.uv = float2(IN.uv.x, 1.0f - IN.uv.y); return result; }; @@ -294,9 +294,9 @@ PS_Input VSMain(const VS_Input IN) constexpr const char shader_frag_passthrough[] = RMLUI_SHADER_HEADER R"( struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; Texture2D g_InputTexture : register(t0); @@ -304,7 +304,7 @@ SamplerState g_SamplerLinear : register(s0); float4 PSMain(const PS_Input IN) : SV_TARGET { - return g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); + return g_InputTexture.Sample(g_SamplerLinear, IN.uv); }; )"; @@ -321,9 +321,9 @@ cbuffer ConstantBuffer : register(b0) struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; float4 PSMain(const PS_Input IN) : SV_TARGET @@ -334,7 +334,7 @@ float4 PSMain(const PS_Input IN) : SV_TARGET // In the general case we should do the matrix transformation in non-premultiplied space. However, without alpha // transformations, we can do it directly in premultiplied space to avoid the extra division and multiplication // steps. In this space, the constant term needs to be multiplied by the alpha value, instead of unity. - float4 texColor = g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); + float4 texColor = g_InputTexture.Sample(g_SamplerLinear, IN.uv); float3 transformedColor = mul(m_color_matrix, texColor).rgb; return float4(transformedColor, texColor.a); }; @@ -347,15 +347,15 @@ Texture2D g_MaskTexture : register(t1); struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; float4 PSMain(const PS_Input IN) : SV_TARGET { - float4 texColor = g_InputTexture.Sample(g_SamplerLinear, IN.inputUV); - float maskAlpha = g_MaskTexture.Sample(g_SamplerLinear, IN.inputUV).a; + float4 texColor = g_InputTexture.Sample(g_SamplerLinear, IN.uv); + float maskAlpha = g_MaskTexture.Sample(g_SamplerLinear, IN.uv).a; return texColor * maskAlpha; }; )"; @@ -366,16 +366,15 @@ float4 PSMain(const PS_Input IN) : SV_TARGET static const char shader_vert_blur[] = RMLUI_SHADER_BLUR_HEADER R"( struct VS_Input { - float2 inPosition : POSITION; - float4 inColor : COLOR; - float2 inTexCoord : TEXCOORD; + float2 position : POSITION; + float4 color : COLOR; + float2 uv : TEXCOORD; }; struct PS_INPUT { - float4 outPosition : SV_Position; - float4 outColor : COLOR; - float2 outUV : TEXCOORD; + float4 position : SV_Position; + float2 uv[BLUR_SIZE] : TEXCOORD; }; cbuffer SharedConstantBuffer : register(b0) @@ -423,8 +422,8 @@ cbuffer ConstantBuffer : register(b0) struct PS_Input { - float4 inputPos : SV_Position; - float2 inputUV[BLUR_SIZE] : TEXCOORD; + float4 position : SV_Position; + float2 uv[BLUR_SIZE] : TEXCOORD; }; float4 PSMain(const PS_Input IN) : SV_TARGET @@ -432,8 +431,8 @@ float4 PSMain(const PS_Input IN) : SV_TARGET float4 color = float4(0.0, 0.0, 0.0, 0.0); for(int i = 0; i < BLUR_SIZE; i++) { - float2 in_region = step(m_texCoordMin, IN.inputUV[i]) * step(IN.inputUV[i], m_texCoordMax); - color += g_InputTexture.Sample(g_SamplerLinear, IN.inputUV[i]) * in_region.x * in_region.y * m_weights[abs(i - BLUR_NUM_WEIGHTS + 1)]; + float2 in_region = step(m_texCoordMin, IN.uv[i]) * step(IN.uv[i], m_texCoordMax); + color += g_InputTexture.Sample(g_SamplerLinear, IN.uv[i]) * in_region.x * in_region.y * m_weights[abs(i - BLUR_NUM_WEIGHTS + 1)]; } return color; }; @@ -456,15 +455,15 @@ cbuffer ConstantBuffer : register(b0) struct PS_Input { - float4 inputPos : SV_Position; - float4 inputColor : COLOR; - float2 inputUV : TEXCOORD; + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; }; float4 PSMain(const PS_Input IN) : SV_TARGET { - float2 in_region = step(m_texCoordMin, IN.inputUV) * step(IN.inputUV, m_texCoordMax); - return g_InputTexture.Sample(g_SamplerLinear, IN.inputUV).a * in_region.x * in_region.y * m_color; + float2 in_region = step(m_texCoordMin, IN.uv) * step(IN.uv, m_texCoordMax); + return g_InputTexture.Sample(g_SamplerLinear, IN.uv).a * in_region.x * in_region.y * m_color; }; )"; From 2768cfa2e26d2001cc7c2c48c39c40c1a02c53c3 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 28 Sep 2024 13:23:23 +0200 Subject: [PATCH 068/125] (fix): Correct blur using incorrect vertex shader --- Backends/RmlUi_Renderer_DX11.cpp | 22 +++++++++------------- 1 file changed, 9 insertions(+), 13 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 31b739b16..1379b34b7 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -381,25 +381,21 @@ cbuffer SharedConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; - float2 _padding; - float4 _padding2[21]; // Padding so that cbuffer aligns with the largest one (gradient) + float4 m_weights; + float2 m_texelOffset; + float2 m_texCoordMin; + float2 m_texCoordMax; + float4 _padding[19]; // Padding so that cbuffer aligns with the largest one (gradient) }; PS_INPUT VSMain(const VS_Input IN) { PS_INPUT result = (PS_INPUT)0; - float2 translatedPos = IN.inPosition + m_translate; - float4 resPos = mul(m_transform, float4(translatedPos.x, translatedPos.y, 0.0, 1.0)); - - result.outPosition = resPos; - result.outColor = IN.inColor; - result.outUV = IN.inTexCoord; - -#if defined(RMLUI_PREMULTIPLIED_ALPHA) - // Pre-multiply vertex colors with their alpha. - result.outColor.rgb = result.outColor.rgb * result.outColor.a; -#endif + for (int i = 0; i < BLUR_SIZE; i++) { + resukt.uv[i] = IN.uv - float(i - BLUR_NUM_WEIGHTS + 1) * m_texelOffset; + } + result.position = float(IN.position, 1.0); return result; }; From 50b203d7f5f728c7f0fed21a22c784a73492c2d2 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 28 Sep 2024 13:38:02 +0200 Subject: [PATCH 069/125] (fix): Correct typos in shaders --- Backends/RmlUi_Renderer_DX11.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 1379b34b7..83df5eaf9 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -129,7 +129,7 @@ struct PS_Input float4 PSMain(const PS_Input IN) : SV_TARGET { - return IN.csolor; + return IN.color; }; )"; @@ -393,9 +393,9 @@ PS_INPUT VSMain(const VS_Input IN) PS_INPUT result = (PS_INPUT)0; for (int i = 0; i < BLUR_SIZE; i++) { - resukt.uv[i] = IN.uv - float(i - BLUR_NUM_WEIGHTS + 1) * m_texelOffset; + result.uv[i] = IN.uv - float(i - BLUR_NUM_WEIGHTS + 1) * m_texelOffset; } - result.position = float(IN.position, 1.0); + result.position = float4(IN.position.xy, 1.0, 1.0); return result; }; From a4a19c5025bef1be5aa15b68bd835ebc4730083e Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 28 Sep 2024 13:38:12 +0200 Subject: [PATCH 070/125] (fix): Unbind slot 1 after use --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 83df5eaf9..76b8881d1 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2594,6 +2594,10 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(1, 1, null_shader_resource_view); + m_render_layers.SwapPostprocessPrimarySecondary(); SetBlendState(blend_state_backup); } From 40dacf81893a721527f0e8d214374a2488af0ad4 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 28 Sep 2024 16:55:50 +0200 Subject: [PATCH 071/125] (fix): Correct clip mask implementation --- Backends/RmlUi_Renderer_DX11.cpp | 38 +++++++++++++++++++++----------- Backends/RmlUi_Renderer_DX11.h | 1 + 2 files changed, 26 insertions(+), 13 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 76b8881d1..fa0c325c3 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1060,7 +1060,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* desc.DepthEnable = false; desc.StencilReadMask = D3D11_DEFAULT_STENCIL_READ_MASK; desc.StencilWriteMask = 0; - desc.FrontFace.StencilFunc = D3D11_COMPARISON_LESS_EQUAL; + desc.FrontFace.StencilFunc = D3D11_COMPARISON_ALWAYS; desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; desc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; desc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; @@ -1087,6 +1087,15 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* desc.BackFace = desc.FrontFace; // Intersect m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_stencil_intersect); + + desc.StencilWriteMask = 0; + desc.FrontFace.StencilFunc = D3D11_COMPARISON_EQUAL; + desc.FrontFace.StencilPassOp = D3D11_STENCIL_OP_KEEP; + desc.FrontFace.StencilFailOp = D3D11_STENCIL_OP_KEEP; + desc.FrontFace.StencilDepthFailOp = D3D11_STENCIL_OP_KEEP; + desc.BackFace = desc.FrontFace; + // Stencil test + m_d3d_device->CreateDepthStencilState(&desc, &m_depth_stencil_state_stencil_test); } } @@ -1114,6 +1123,7 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_intersect); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_set); + DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_test); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); @@ -1170,8 +1180,10 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); + SetBlendState(m_blend_state_enable); m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); // Disable scissor + m_stencil_ref_value = 0; // Reset stencil m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); Clear(); @@ -2656,13 +2668,14 @@ void RenderInterface_DX11::EnableClipMask(bool enable) clipMaskName = L"DisableClipMask"; DebugScope scope(clipMaskName); - if (enable != m_is_stencil_enabled) + m_is_stencil_enabled = enable; + if (enable) { - m_is_stencil_enabled = enable; - if (!enable) - { - m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); - } + m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_stencil_test, m_stencil_ref_value); + } + else + { + m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); } } @@ -2674,7 +2687,7 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm ID3D11DepthStencilState* stencil_state = m_depth_stencil_state_disable; - UINT stencil_test_value = 0; + UINT stencil_test_value = m_stencil_ref_value; switch (operation) { case ClipMaskOperation::Set: @@ -2702,7 +2715,7 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm m_d3d_context->OMSetBlendState(m_blend_state_disable_color, blendFactor, 0xffffffff); m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); - m_d3d_context->OMSetDepthStencilState(stencil_state, stencil_test_value); + m_d3d_context->OMSetDepthStencilState(stencil_state, 1); bool clear_stencil = (operation == ClipMaskOperation::Set || operation == ClipMaskOperation::SetInverse); if (clear_stencil) @@ -2713,14 +2726,13 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm m_d3d_context->ClearDepthStencilView(rtData.depth_stencil_view, D3D11_CLEAR_STENCIL, 1.0f, 0); } - // @TODO: Stencil buffer is inverted? RenderGeometry(geometry, translation, {}); - // Restore state - // @performance Is this even necessary? - m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); // Restore blend state m_d3d_context->OMSetBlendState(m_current_blend_state, blendFactor, 0xffffffff); + // Restore state + m_stencil_ref_value = stencil_test_value; + m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_stencil_test, m_stencil_ref_value); } RenderInterface_DX11::RenderLayerStack::RenderLayerStack() diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 8cda74876..b7deeb9b9 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -172,6 +172,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11DepthStencilState* m_depth_stencil_state_disable = nullptr; // Stencil Off ID3D11DepthStencilState* m_depth_stencil_state_stencil_set = nullptr; // Clipmask Set/Inverse ID3D11DepthStencilState* m_depth_stencil_state_stencil_intersect = nullptr; // Clipmask Intersect. Can have several test values + ID3D11DepthStencilState* m_depth_stencil_state_stencil_test = nullptr; // Clipmask test // Shaders ID3D11Buffer* m_shader_buffer = nullptr; From 7f1c779afaab744ddf93bdffb9b194de4238c137 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:38:26 +0200 Subject: [PATCH 072/125] (chore): Match comment locations with GL3 --- Backends/RmlUi_Renderer_DX11.cpp | 6 +++--- Backends/RmlUi_Renderer_DX11.h | 8 ++++---- 2 files changed, 7 insertions(+), 7 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index fa0c325c3..4485b193f 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -878,8 +878,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* Rml::MeshUtilities::GenerateQuad(mesh, Rml::Vector2f(-1), Rml::Vector2f(2), {}); m_fullscreen_quad_geometry = RenderInterface_DX11::CompileGeometry(mesh.vertices, mesh.indices); - // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. - // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). if (!m_blend_state_enable) { D3D11_BLEND_DESC blend_desc{}; @@ -897,6 +895,8 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* } #endif + // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. + // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). blend_desc.RenderTarget[0].BlendEnable = true; blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE; blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA; @@ -1409,7 +1409,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm if (texture == TexturePostprocess) { - + // Do nothing. } else if (texture) { diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index b7deeb9b9..f418f60d4 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -318,13 +318,13 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void DestroyRenderTargets(); const Gfx::RenderTargetData& EnsureRenderTargetPostprocess(int index); - int width = 0, height = 0; + int m_width = 0, m_height = 0; // The number of active layers is manually tracked since we re-use the render targets stored in the fb_layers stack. - int layers_size = 0; + int m_layers_size = 0; - Rml::Vector rt_layers; - Rml::Vector rt_postprocess; + Rml::Vector m_rt_layers; + Rml::Vector m_rt_postprocess; ID3D11Device* m_d3d_device = nullptr; }; From cf2397bfb1030fb21f81357cdedc08175fa52b37 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:38:59 +0200 Subject: [PATCH 073/125] (chore): Correct RenderGeometry having side-effects --- Backends/RmlUi_Renderer_DX11.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 4485b193f..99de633ac 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1401,7 +1401,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm { DX11_GeometryData geometryData = m_geometry_cache[handle]; - if (m_translation != translation) + if (texture != TexturePostprocess && m_translation != translation) { m_translation = translation; m_cbuffer_dirty = true; @@ -1421,15 +1421,15 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm m_d3d_context->PSSetShaderResources(0, 1, &texture_view); } m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); + UpdateConstantBuffer(); } else { // No texture, use color UseProgram(ProgramId::Color); + UpdateConstantBuffer(); } - UpdateConstantBuffer(); - m_d3d_context->IASetInputLayout(m_vertex_layout); m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); From 80bae7207f703ebc6d704805869389fae9ad31dc Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:40:24 +0200 Subject: [PATCH 074/125] (fix): Bind depth stencil to match GL3 behaviour --- Backends/RmlUi_Renderer_DX11.cpp | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 99de633ac..1339f2edb 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -342,8 +342,8 @@ float4 PSMain(const PS_Input IN) : SV_TARGET static const char shader_frag_blend_mask[] = RMLUI_SHADER_HEADER R"( Texture2D g_InputTexture : register(t0); -SamplerState g_SamplerLinear : register(s0); Texture2D g_MaskTexture : register(t1); +SamplerState g_SamplerLinear : register(s0); struct PS_Input { @@ -1895,7 +1895,7 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); const Gfx::RenderTargetData& destination = m_render_layers.GetBlendMask(); - m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); Gfx::BindTexture(m_d3d_context, source); UseProgram(ProgramId::Passthrough); ID3D11BlendState* blend_state_backup = m_current_blend_state; @@ -2274,7 +2274,6 @@ static void SetBlurWeights(Rml::Vector4f& p_weights, float sigma) void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& source_destination, const Gfx::RenderTargetData& temp, const Rml::Rectanglei window_flipped) { - // @TODO: Fix blur return; DebugScope scope(L"RenderBlur"); RMLUI_ASSERT(&source_destination != &temp && source_destination.width == temp.width && source_destination.height == temp.height); @@ -2312,7 +2311,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& scissor.p1 = Rml::Math::Max(scissor.p1 / 2, scissor.p0); const bool from_source = (i % 2 == 0); Gfx::BindTexture(m_d3d_context, from_source ? source_destination : temp); - m_d3d_context->OMSetRenderTargets(1, &(from_source ? temp : source_destination).render_target_view, source_destination.depth_stencil_view); + m_d3d_context->OMSetRenderTargets(1, &(from_source ? temp : source_destination).render_target_view, + (from_source ? temp : source_destination).depth_stencil_view); SetScissor(scissor, true); DrawFullscreenQuad({}, uv_scaling); @@ -2331,7 +2331,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& if (transfer_to_temp_buffer) { Gfx::BindTexture(m_d3d_context, source_destination); - m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, temp.depth_stencil_view); DrawFullscreenQuad(); } @@ -2368,14 +2368,14 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // Blur render pass - vertical. Gfx::BindTexture(m_d3d_context, temp); - m_d3d_context->OMSetRenderTargets(1, &source_destination.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &source_destination.render_target_view, source_destination.depth_stencil_view); SetTexelOffset({0.f, 1.f}, temp.height); DrawFullscreenQuad(); // Blur render pass - horizontal. Gfx::BindTexture(m_d3d_context, source_destination); - m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, temp.depth_stencil_view); // Add a 1px transparent border around the blur region by first clearing with a padded scissor. This helps prevent // artifacts when upscaling the blur result in the later step. On Intel and AMD, we have observed that during @@ -2465,7 +2465,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetSamplers(0, 1, &m_sampler_state); - m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); DrawFullscreenQuad(); @@ -2501,7 +2501,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetRenderTargets(1, &secondary.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &secondary.render_target_view, secondary.depth_stencil_view); D3D11_MAPPED_SUBRESOURCE mappedResource{}; // Lock the constant buffer so it can be written to. @@ -2581,7 +2581,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetRenderTargets(1, &destination.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); DrawFullscreenQuad(); @@ -2602,7 +2602,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetRenderTargets(1, &destination.render_target_view, nullptr); + m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); DrawFullscreenQuad(); From cf869fce03d82564b8ccab6798d44db720811466 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:40:39 +0200 Subject: [PATCH 075/125] (fix): Don't clear alpha channel --- Backends/RmlUi_Renderer_DX11.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 1339f2edb..24fe4c94f 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2383,7 +2383,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // hand, it looks like Nvidia clamps the pixels to the source edge, which is what we really want. Regardless, we // work around the issue with this extra step. SetScissor(scissor.Extend(1), true); - float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; + float clearColor[4] = {0.0f, 0.0f, 0.0f, 0.0f}; m_d3d_context->ClearRenderTargetView(temp.render_target_view, clearColor); SetScissor(scissor, true); From 04df677870a6d5a71d0071c31a7ae09cbe07a0d6 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:41:40 +0200 Subject: [PATCH 076/125] (chore): Make RenderLayerStack use m_ for variables --- Backends/RmlUi_Renderer_DX11.cpp | 53 ++++++++++++++++---------------- 1 file changed, 27 insertions(+), 26 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 24fe4c94f..eb0aa4cc7 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2737,7 +2737,7 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm RenderInterface_DX11::RenderLayerStack::RenderLayerStack() { - rt_postprocess.resize(4); + m_rt_postprocess.resize(4); } RenderInterface_DX11::RenderLayerStack::~RenderLayerStack() @@ -2753,32 +2753,33 @@ void RenderInterface_DX11::RenderLayerStack::SetD3DResources(ID3D11Device* devic Rml::LayerHandle RenderInterface_DX11::RenderLayerStack::PushLayer() { - RMLUI_ASSERT(layers_size <= (int)rt_layers.size()); + RMLUI_ASSERT(m_layers_size <= (int)m_rt_layers.size()); - if (layers_size == (int)rt_layers.size()) + if (m_layers_size == (int)m_rt_layers.size()) { // All framebuffers should share a single stencil buffer. - ID3D11DepthStencilView* shared_depth_stencil = (rt_layers.empty() ? nullptr : rt_layers.front().depth_stencil_view); + ID3D11DepthStencilView* shared_depth_stencil = (m_rt_layers.empty() ? nullptr : m_rt_layers.front().depth_stencil_view); - rt_layers.push_back(Gfx::RenderTargetData{}); - Gfx::CreateRenderTarget(m_d3d_device, rt_layers.back(), width, height, NUM_MSAA_SAMPLES, Gfx::RenderTargetAttachment::DepthStencil, shared_depth_stencil); + m_rt_layers.push_back(Gfx::RenderTargetData{}); + Gfx::CreateRenderTarget(m_d3d_device, m_rt_layers.back(), m_width, m_height, NUM_MSAA_SAMPLES, Gfx::RenderTargetAttachment::DepthStencil, + shared_depth_stencil); } - layers_size += 1; + m_layers_size += 1; return GetTopLayerHandle(); } void RenderInterface_DX11::RenderLayerStack::PopLayer() { - RMLUI_ASSERT(layers_size > 0); - layers_size -= 1; + RMLUI_ASSERT(m_layers_size > 0); + m_layers_size -= 1; } const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::GetLayer(Rml::LayerHandle layer) const { - RMLUI_ASSERT((size_t)layer < (size_t)layers_size); - return rt_layers[layer]; + RMLUI_ASSERT((size_t)layer < (size_t)m_layers_size); + return m_rt_layers[layer]; } const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::GetTopLayer() const @@ -2788,23 +2789,23 @@ const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::GetTopLayer Rml::LayerHandle RenderInterface_DX11::RenderLayerStack::GetTopLayerHandle() const { - RMLUI_ASSERT(layers_size > 0); - return static_cast(layers_size - 1); + RMLUI_ASSERT(m_layers_size > 0); + return static_cast(m_layers_size - 1); } void RenderInterface_DX11::RenderLayerStack::SwapPostprocessPrimarySecondary() { - std::swap(rt_postprocess[0], rt_postprocess[1]); + std::swap(m_rt_postprocess[0], m_rt_postprocess[1]); } void RenderInterface_DX11::RenderLayerStack::BeginFrame(int new_width, int new_height) { - RMLUI_ASSERT(layers_size == 0); + RMLUI_ASSERT(m_layers_size == 0); - if (new_width != width || new_height != height) + if (new_width != m_width || new_height != m_height) { - width = new_width; - height = new_height; + m_width = new_width; + m_height = new_height; DestroyRenderTargets(); } @@ -2814,29 +2815,29 @@ void RenderInterface_DX11::RenderLayerStack::BeginFrame(int new_width, int new_h void RenderInterface_DX11::RenderLayerStack::EndFrame() { - RMLUI_ASSERT(layers_size == 1); + RMLUI_ASSERT(m_layers_size == 1); PopLayer(); } void RenderInterface_DX11::RenderLayerStack::DestroyRenderTargets() { - RMLUI_ASSERTMSG(layers_size == 0, "Do not call this during frame rendering, that is, between BeginFrame() and EndFrame()."); + RMLUI_ASSERTMSG(m_layers_size == 0, "Do not call this during frame rendering, that is, between BeginFrame() and EndFrame()."); - for (Gfx::RenderTargetData& fb : rt_layers) + for (Gfx::RenderTargetData& fb : m_rt_layers) Gfx::DestroyRenderTarget(fb); - rt_layers.clear(); + m_rt_layers.clear(); - for (Gfx::RenderTargetData& fb : rt_postprocess) + for (Gfx::RenderTargetData& fb : m_rt_postprocess) Gfx::DestroyRenderTarget(fb); } const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::EnsureRenderTargetPostprocess(int index) { - RMLUI_ASSERT(index < (int)rt_postprocess.size()) - Gfx::RenderTargetData& rt = rt_postprocess[index]; + RMLUI_ASSERT(index < (int)m_rt_postprocess.size()) + Gfx::RenderTargetData& rt = m_rt_postprocess[index]; if (!rt.render_target_view) - Gfx::CreateRenderTarget(m_d3d_device, rt, width, height, 0, Gfx::RenderTargetAttachment::None, 0); + Gfx::CreateRenderTarget(m_d3d_device, rt, m_width, m_height, 0, Gfx::RenderTargetAttachment::None, 0); return rt; } From 42a408271110c04bd86e1e37d3d0665ac0a39e9a Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:42:13 +0200 Subject: [PATCH 077/125] (fix): Unbind resource views after rendering filters --- Backends/RmlUi_Renderer_DX11.cpp | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index eb0aa4cc7..7166563cb 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2446,6 +2446,10 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetSamplers(0, 1, &m_sampler_state); + + // To unbind slots when necessary + ID3D11ShaderResourceView* null_shader_resource_views[2] = {nullptr, nullptr}; + for (const Rml::CompiledFilterHandle filter_handle : filter_handles) { const CompiledFilter& filter = *reinterpret_cast(filter_handle); @@ -2585,6 +2589,8 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(0, 2, null_shader_resource_views); + m_render_layers.SwapPostprocessPrimarySecondary(); SetBlendState(blend_state_backup); } @@ -2606,9 +2612,8 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(1, 1, null_shader_resource_view); + // Unbind textures + m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); m_render_layers.SwapPostprocessPrimarySecondary(); SetBlendState(blend_state_backup); From 2d164ec3fa675159e5d466abbf7960cd2f9fcdec Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:42:29 +0200 Subject: [PATCH 078/125] (chore): Cast bounds to float --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 7166563cb..ff7c60149 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2423,8 +2423,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& D3D11_VIEWPORT viewport; viewport.TopLeftX = target_min.x; viewport.TopLeftY = target_min.y; - viewport.Width = target_max.x - target_min.x; - viewport.Height = target_max.y - target_min.y; + viewport.Width = (float)(target_max.x - target_min.x); + viewport.Height = (float)(target_max.y - target_min.y); viewport.MinDepth = 0.0f; viewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &viewport); From a7eb5f7d4379cb274ee6841de9c9753324e780d4 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:43:03 +0200 Subject: [PATCH 079/125] (fix): Fix debug layer incorrect marking event at wrong spot --- Backends/RmlUi_Renderer_DX11.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index ff7c60149..13d93d1f6 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1761,13 +1761,13 @@ void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_ Rml::LayerHandle RenderInterface_DX11::PushLayer() { + m_debug->BeginEvent(L"BeginLayer"); const Rml::LayerHandle layer_handle = m_render_layers.PushLayer(); m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); float colors[4] = {0, 0, 0, 0}; m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, colors); - m_debug->BeginEvent(L"BeginLayer"); return layer_handle; } From a0f55434af03454500d5789810b0b806d5c835b5 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:43:18 +0200 Subject: [PATCH 080/125] (fix): Mark cbuffer as dirty when altered --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 13d93d1f6..cd2d0ff33 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2534,6 +2534,8 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanUnmap(m_shader_buffer, 0); + + m_cbuffer_dirty = true; const Rml::Vector2f uv_offset = filter.offset / Rml::Vector2f(-(float)m_viewport_width, (float)m_viewport_height); DrawFullscreenQuad(uv_offset); @@ -2582,6 +2584,8 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanUnmap(m_shader_buffer, 0); + m_cbuffer_dirty = true; + const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); Gfx::BindTexture(m_d3d_context, source); From de5755a7f8bac48261c416c9198475e3090fe82f Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:43:41 +0200 Subject: [PATCH 081/125] (chore): Handle transposing matrices? Doesn't actually work --- Backends/RmlUi_Renderer_DX11.cpp | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index cd2d0ff33..04ab2dbfe 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2574,12 +2574,13 @@ void RenderInterface_DX11::RenderFilters(Rml::Span::value; + // Copy the data to the GPU dataPtr->transform = m_transform; dataPtr->translation = m_translation; memset(&dataPtr->color_matrix.color_matrix, 0, sizeof(dataPtr->color_matrix.color_matrix)); - memcpy_s(&dataPtr->color_matrix.color_matrix, sizeof(dataPtr->color_matrix.color_matrix), filter.color_matrix.data(), - sizeof(dataPtr->color_matrix.color_matrix)); + memcpy_s(&dataPtr->color_matrix.color_matrix, sizeof(dataPtr->color_matrix.color_matrix), filter.color_matrix.data(), sizeof(dataPtr->color_matrix.color_matrix)); // Upload to the GPU. m_d3d_context->Unmap(m_shader_buffer, 0); From 5e7207bd5c3acd69176f001a1a58bbfcd98d69b2 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:45:10 +0200 Subject: [PATCH 082/125] (feat): Begin implementing BlitRenderTarget This is to match the behaviour of glBlitFramebuffers, as on DX11 using CopySubresourceRegion breaks rendering with layers it seems --- Backends/RmlUi_Renderer_DX11.cpp | 39 +++++++++++++++++++++++--------- Backends/RmlUi_Renderer_DX11.h | 3 +++ 2 files changed, 31 insertions(+), 11 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 04ab2dbfe..d41f4f583 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1749,6 +1749,27 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) m_cbuffer_dirty = true; } +void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, const Gfx::RenderTargetData& dest, int srcX0, int srcY0, int srcX1, + int srcY1, int dstX0, int dstY0, int dstX1, int dstY1) +{ + int src_width = srcX1 - srcX0; + int src_height = srcY1 - srcY0; + int dest_width = dstX1 - dstX0; + int dest_height = dstY1 - dstY0; + + RMLUI_ASSERTMSG(src_width > 0, "Invalid source rectangle (width)!"); + RMLUI_ASSERTMSG(src_height > 0, "Invalid source rectangle (height)!"); + RMLUI_ASSERTMSG(dest_width > 0, "Invalid destination rectangle (width)!"); + RMLUI_ASSERTMSG(dest_height > 0, "Invalid destination rectangle (height)!"); + + // @TODO: Use viewport to define write region + // @TODO: Lerp UVs such that they match with the write region + // @TODO: MSAA resolve + + // @TODO: Create temporary for MSAA resolve ? + m_d3d_context->ResolveSubresource(dest.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); +} + void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle) { DebugScope scope(L"BlitLayerToPostprocessPrimary"); @@ -1756,7 +1777,7 @@ void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_ const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessPrimary(); // Blit and resolve MSAA. Any active scissor state will restrict the size of the blit region. - m_d3d_context->ResolveSubresource(destination.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); + BlitRenderTarget(source, destination, 0, 0, source.width, source.height, 0, 0, destination.width, destination.height); } Rml::LayerHandle RenderInterface_DX11::PushLayer() @@ -1831,16 +1852,12 @@ Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); // Flip the image vertically, as that convention is used for textures, and move to origin. - D3D11_BOX source_box{}; - source_box.left = bounds.Left(); - source_box.right = bounds.Right(); - source_box.top = source.height - bounds.Bottom(); - source_box.bottom = source.height - bounds.Top(); - source_box.front = 0; - source_box.back = 1; - - // Perform a copy from the source render target to the destination render target - m_d3d_context->CopySubresourceRegion(destination.render_target_texture, 0, 0, 0, 0, source.render_target_texture, 0, &source_box); + BlitRenderTarget(source, destination, + bounds.Left(), source.height - bounds.Bottom(), // src0 + bounds.Right(), source.height - bounds.Top(), // src1 + 0, bounds.Height(), // dst0 + bounds.Width(), 0 // dst1 + ); // Now we need to copy the destination texture to the final texture for rendering // Bind the destination texture as the source for copying to the final texture (render_texture) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index f418f60d4..6424c789b 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -146,6 +146,9 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void UpdateConstantBuffer(); void UseProgram(ProgramId program_id); + // Mirrors the behaviour of glBlitFramebuffer. It supports sampling from a rect from source, writing to a rect at dest, and does MSAA resolve. + void BlitRenderTarget(const Gfx::RenderTargetData& source, const Gfx::RenderTargetData& dest, int srcX0, int srcY0, int srcX1, int srcY1, + int dstX0, int dstY0, int dstX1, int dstY1); void BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle); void RenderFilters(Rml::Span filter_handles); From 9225739496eb82d94baae54847bb4778b6952e51 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:56:22 +0200 Subject: [PATCH 083/125] (feat): Add temporary render target for blitting --- Backends/RmlUi_Renderer_DX11.cpp | 5 ++++- Backends/RmlUi_Renderer_DX11.h | 1 + 2 files changed, 5 insertions(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index d41f4f583..f7564950c 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1762,10 +1762,13 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, RMLUI_ASSERTMSG(dest_width > 0, "Invalid destination rectangle (width)!"); RMLUI_ASSERTMSG(dest_height > 0, "Invalid destination rectangle (height)!"); + // @TODO: Use viewport to define write region // @TODO: Lerp UVs such that they match with the write region // @TODO: MSAA resolve + const Gfx::RenderTargetData& temporary_rt = m_render_layers.GetTemporary(); + // @TODO: Create temporary for MSAA resolve ? m_d3d_context->ResolveSubresource(dest.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); } @@ -2764,7 +2767,7 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm RenderInterface_DX11::RenderLayerStack::RenderLayerStack() { - m_rt_postprocess.resize(4); + m_rt_postprocess.resize(5); } RenderInterface_DX11::RenderLayerStack::~RenderLayerStack() diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 6424c789b..a8d508399 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -311,6 +311,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { const Gfx::RenderTargetData& GetPostprocessSecondary() { return EnsureRenderTargetPostprocess(1); } const Gfx::RenderTargetData& GetPostprocessTertiary() { return EnsureRenderTargetPostprocess(2); } const Gfx::RenderTargetData& GetBlendMask() { return EnsureRenderTargetPostprocess(3); } + const Gfx::RenderTargetData& GetTemporary() { return EnsureRenderTargetPostprocess(4); } void SwapPostprocessPrimarySecondary(); From a13dceeaddf6bf4fe0a67466b37951748cac4d39 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 11:56:38 +0200 Subject: [PATCH 084/125] (chore): Assert on invalid coords for blitting --- Backends/RmlUi_Renderer_DX11.cpp | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index f7564950c..804026196 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1757,6 +1757,17 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, int dest_width = dstX1 - dstX0; int dest_height = dstY1 - dstY0; + // All coords must be greater than 0 + RMLUI_ASSERTMSG(srcX0 > 0, "Invalid source coordinate (srcX0)!"); + RMLUI_ASSERTMSG(srcX1 > 0, "Invalid source coordinate (srcX1)!"); + RMLUI_ASSERTMSG(srcY0 > 0, "Invalid source coordinate (srcY0)!"); + RMLUI_ASSERTMSG(srcY1 > 0, "Invalid source coordinate (srcY1)!"); + RMLUI_ASSERTMSG(dstX0 > 0, "Invalid destination coordinate (dstX0)!"); + RMLUI_ASSERTMSG(dstX1 > 0, "Invalid destination coordinate (dstX1)!"); + RMLUI_ASSERTMSG(dstY0 > 0, "Invalid destination coordinate (dstY0)!"); + RMLUI_ASSERTMSG(dstY1 > 0, "Invalid destination coordinate (dstY1)!"); + + // Width and height must be greater than 0 RMLUI_ASSERTMSG(src_width > 0, "Invalid source rectangle (width)!"); RMLUI_ASSERTMSG(src_height > 0, "Invalid source rectangle (height)!"); RMLUI_ASSERTMSG(dest_width > 0, "Invalid destination rectangle (width)!"); From ca36f572e6f375e5cc4502d5eaf0668d5da367a9 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 29 Sep 2024 23:52:16 +0200 Subject: [PATCH 085/125] (feat): Add a custom blit implementation glBlitFramebuffer doesn't have a direct equivalent, so we have to mimic it's behaviour another way --- Backends/RmlUi_Renderer_DX11.cpp | 153 +++++++++++++++++++++++++++---- Backends/RmlUi_Renderer_DX11.h | 15 +++ 2 files changed, 150 insertions(+), 18 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 804026196..003134386 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -463,6 +463,53 @@ float4 PSMain(const PS_Input IN) : SV_TARGET }; )"; +constexpr const char shader_vert_blit[] = RMLUI_SHADER_HEADER R"( +struct VS_Input +{ + float2 position : POSITION; + float4 color : COLOR; + float2 uv : TEXCOORD; +}; + +struct PS_Input +{ + float4 position : SV_Position; + float4 color : COLOR; + float2 uv : TEXCOORD; +}; + +cbuffer ConstantBuffer : register(b0) +{ + float4x4 m_transform; + float2 m_source_min; + float2 m_source_max; + float2 m_destination_min; + float2 m_destination_max; + float4 _padding2[20]; // Padding so that cbuffer aligns with the largest one (gradient) +}; + +PS_Input VSMain(const VS_Input IN) +{ + PS_Input result = (PS_Input)0; + + result.position = float4(IN.position.xy, 0.0f, 1.0f); + result.uv = float2(IN.uv.x, 1.0f - IN.uv.y); + + // Correct coordinates to match source and dest bounds + // For source we lerp the UVs + result.uv.x = lerp(m_source_min.x, m_source_max.x, result.uv.x); + result.uv.y = lerp(m_source_min.y, m_source_max.y, result.uv.y); + + // For destination we lerp the positions. Since the clip space box has a range of -1 to +1, we have to remap -1 to +1 to 0 to 1, lerp, then remap back to -1 to +1 + result.position.xy = result.position.xy * 0.5f + 0.5f; // Remap to 0-+1 + result.position.x = lerp(m_destination_min.x, m_destination_max.x, result.position.x); + result.position.y = lerp(m_destination_min.y, m_destination_max.y, result.position.y); + result.position.xy = (result.position.xy - 0.5f) * 2.0f; // Remap to -1-+1 + + return result; +}; +)"; + enum class ProgramId { None, Color, @@ -474,12 +521,14 @@ enum class ProgramId { BlendMask, Blur, DropShadow, + Blit, Count, }; enum class VertShaderId { Main, Passthrough, Blur, + Blit, Count, }; enum class FragShaderId { @@ -525,6 +574,7 @@ static const VertShaderDefinition vert_shader_definitions[] = { {VertShaderId::Main, "main", shader_vert_main, sizeof(shader_vert_main)}, {VertShaderId::Passthrough, "passthrough", shader_vert_passthrough, sizeof(shader_vert_passthrough)}, {VertShaderId::Blur, "blur", shader_vert_blur, sizeof(shader_vert_blur)}, + {VertShaderId::Blit, "blit", shader_vert_blit, sizeof(shader_vert_blit)}, }; static const FragShaderDefinition frag_shader_definitions[] = { {FragShaderId::Color, "color", shader_frag_color, sizeof(shader_frag_color)}, @@ -547,6 +597,7 @@ static const ProgramDefinition program_definitions[] = { {ProgramId::BlendMask, "blend_mask", VertShaderId::Passthrough, FragShaderId::BlendMask}, {ProgramId::Blur, "blur", VertShaderId::Blur, FragShaderId::Blur}, {ProgramId::DropShadow, "drop_shadow", VertShaderId::Passthrough, FragShaderId::DropShadow}, + {ProgramId::Blit, "blur", VertShaderId::Blit, FragShaderId::Passthrough}, }; // clang-format on @@ -1682,9 +1733,15 @@ void RenderInterface_DX11::SetScissor(Rml::Rectanglei region, bool vertically_fl if (region.Valid() != m_scissor_state.Valid()) { if (region.Valid()) + { m_d3d_context->RSSetState(m_rasterizer_state_scissor_enabled); + m_scissor_enabled = true; + } else + { m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); + m_scissor_enabled = false; + } } if (region.Valid() && vertically_flip) @@ -1752,36 +1809,96 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, const Gfx::RenderTargetData& dest, int srcX0, int srcY0, int srcX1, int srcY1, int dstX0, int dstY0, int dstX1, int dstY1) { + DebugScope scope(L"BlitRenderTarget"); + int src_width = srcX1 - srcX0; int src_height = srcY1 - srcY0; int dest_width = dstX1 - dstX0; int dest_height = dstY1 - dstY0; // All coords must be greater than 0 - RMLUI_ASSERTMSG(srcX0 > 0, "Invalid source coordinate (srcX0)!"); - RMLUI_ASSERTMSG(srcX1 > 0, "Invalid source coordinate (srcX1)!"); - RMLUI_ASSERTMSG(srcY0 > 0, "Invalid source coordinate (srcY0)!"); - RMLUI_ASSERTMSG(srcY1 > 0, "Invalid source coordinate (srcY1)!"); - RMLUI_ASSERTMSG(dstX0 > 0, "Invalid destination coordinate (dstX0)!"); - RMLUI_ASSERTMSG(dstX1 > 0, "Invalid destination coordinate (dstX1)!"); - RMLUI_ASSERTMSG(dstY0 > 0, "Invalid destination coordinate (dstY0)!"); - RMLUI_ASSERTMSG(dstY1 > 0, "Invalid destination coordinate (dstY1)!"); + RMLUI_ASSERTMSG(srcX0 >= 0, "Invalid source coordinate (srcX0)!"); + RMLUI_ASSERTMSG(srcX1 >= 0, "Invalid source coordinate (srcX1)!"); + RMLUI_ASSERTMSG(srcY0 >= 0, "Invalid source coordinate (srcY0)!"); + RMLUI_ASSERTMSG(srcY1 >= 0, "Invalid source coordinate (srcY1)!"); + RMLUI_ASSERTMSG(dstX0 >= 0, "Invalid destination coordinate (dstX0)!"); + RMLUI_ASSERTMSG(dstX1 >= 0, "Invalid destination coordinate (dstX1)!"); + RMLUI_ASSERTMSG(dstY0 >= 0, "Invalid destination coordinate (dstY0)!"); + RMLUI_ASSERTMSG(dstY1 >= 0, "Invalid destination coordinate (dstY1)!"); // Width and height must be greater than 0 - RMLUI_ASSERTMSG(src_width > 0, "Invalid source rectangle (width)!"); - RMLUI_ASSERTMSG(src_height > 0, "Invalid source rectangle (height)!"); - RMLUI_ASSERTMSG(dest_width > 0, "Invalid destination rectangle (width)!"); - RMLUI_ASSERTMSG(dest_height > 0, "Invalid destination rectangle (height)!"); - + RMLUI_ASSERTMSG(src_width != 0, "Invalid source rectangle (width)!"); + RMLUI_ASSERTMSG(src_height != 0, "Invalid source rectangle (height)!"); + RMLUI_ASSERTMSG(dest_width != 0, "Invalid destination rectangle (width)!"); + RMLUI_ASSERTMSG(dest_height != 0, "Invalid destination rectangle (height)!"); - // @TODO: Use viewport to define write region - // @TODO: Lerp UVs such that they match with the write region - // @TODO: MSAA resolve + ID3D11RenderTargetView* original_render_target_view = nullptr; + ID3D11DepthStencilView* original_depth_stencil_view = nullptr; + m_d3d_context->OMGetRenderTargets(1, &original_render_target_view, &original_depth_stencil_view); + // Bind temporary as the render target const Gfx::RenderTargetData& temporary_rt = m_render_layers.GetTemporary(); - // @TODO: Create temporary for MSAA resolve ? - m_d3d_context->ResolveSubresource(dest.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); + // Unbind existing textures to prevent warning spam + ID3D11ShaderResourceView* null_shader_resource_views[2] = {nullptr, nullptr}; + m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); + + // Resolve from source to the temporary first + m_d3d_context->ResolveSubresource(temporary_rt.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); + + m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); + + // Bind destination as our final render texture + m_d3d_context->OMSetRenderTargets(1, &dest.render_target_view, dest.depth_stencil_view); + + // Draw a quad into temporary with source bound as the texture, and the UVs lerping to match the new dimensions + // We want to map UV 0 - 1 to src min-max + // We want to map pos -1 - +1 to dst min max + UseProgram(ProgramId::Blit); + + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + // Restore state before exiting + m_d3d_context->OMSetRenderTargets(1, &original_render_target_view, original_depth_stencil_view); + original_render_target_view->Release(); + original_depth_stencil_view->Release(); + return; + } + + // Get a pointer to the data in the constant buffer. + ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + + dataPtr->transform = m_transform; + + // To blit we need to map the UVs from a 0 to 1 range to a src0 to src1 range + dataPtr->blit.src_x_min = (float) srcX0 / (float) source.width; // Map to 0 + dataPtr->blit.src_y_min = (float) srcY0 / (float) source.height; // Map to 0 + dataPtr->blit.src_x_max = (float) srcX1 / (float) source.width; // Map to +1 + dataPtr->blit.src_y_max = (float) srcY1 / (float) source.height; // Map to +1 + + // To blit we need to map the position from a -1 to 1 range to a dst0 dst1 range + dataPtr->blit.dst_x_min = (float) dstX0 / (float) dest.width; // Map to -1 + dataPtr->blit.dst_y_min = (float) dstY0 / (float) dest.height; // Map to -1 + dataPtr->blit.dst_x_max = (float) dstX1 / (float) dest.width; // Map to +1 + dataPtr->blit.dst_y_max = (float) dstY1 / (float) dest.height; // Map to +1 + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + + // Bind texture + m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); + Gfx::BindTexture(m_d3d_context, temporary_rt); + + DrawFullscreenQuad(); + + // Restore state + m_d3d_context->OMSetRenderTargets(1, &original_render_target_view, original_depth_stencil_view); + m_d3d_context->RSSetState(m_scissor_enabled ? m_rasterizer_state_scissor_enabled : m_rasterizer_state_scissor_disabled); + original_render_target_view->Release(); + original_depth_stencil_view->Release(); } void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index a8d508399..e9a0597c6 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -147,6 +147,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void UseProgram(ProgramId program_id); // Mirrors the behaviour of glBlitFramebuffer. It supports sampling from a rect from source, writing to a rect at dest, and does MSAA resolve. + // Coordinates are in pixels. void BlitRenderTarget(const Gfx::RenderTargetData& source, const Gfx::RenderTargetData& dest, int srcX0, int srcY0, int srcX1, int srcY1, int dstX0, int dstY0, int dstX1, int dstY1); void BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle); @@ -165,6 +166,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11DeviceContext* m_d3d_context = nullptr; IDXGISwapChain* m_bound_swapchain = nullptr; ID3D11RenderTargetView* m_bound_render_target = nullptr; + bool m_scissor_enabled = false; ID3D11RasterizerState* m_rasterizer_state_scissor_enabled = nullptr; ID3D11RasterizerState* m_rasterizer_state_scissor_disabled = nullptr; @@ -276,6 +278,19 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::Vector2f dimensions; float value; } creation; + struct Blit { + int _padding[16]; + + float src_x_min; + float src_y_min; + float src_x_max; + float src_y_max; + + float dst_x_min; + float dst_y_min; + float dst_x_max; + float dst_y_max; + } blit; }; #pragma pack() From baaf51f31eadf161696febc01598bc5b8ccd2068 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 30 Sep 2024 14:10:53 +0200 Subject: [PATCH 086/125] (fix): Fix incorrect blend factor --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 003134386..1663af98f 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1794,7 +1794,7 @@ void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blend_state) { if (blend_state != m_current_blend_state) { - const float blend_factor[4] = {0.f, 0.f, 0.f, 0.f}; + const float blend_factor[4] = {1.0f, 1.0f, 1.0f, 1.0f}; m_d3d_context->OMSetBlendState(blend_state, blend_factor, 0xFFFFFFFF); m_current_blend_state = blend_state; } @@ -2869,7 +2869,7 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm } // Disable writing to the color of the render target - float blendFactor[4] = {0.f, 0.f, 0.f, 0.f}; + float blendFactor[4] = {1.0f, 1.0f, 1.0f, 1.0f}; m_d3d_context->OMSetBlendState(m_blend_state_disable_color, blendFactor, 0xffffffff); m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); From dc437f239e8d9fc1224b1dbeb2dd30da81725eef Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 30 Sep 2024 14:11:11 +0200 Subject: [PATCH 087/125] (fix): Correct copy to copy bounds --- Backends/RmlUi_Renderer_DX11.cpp | 8 +- Backends/stb_image.h | 9310 ++++++++++++++++++++++++++++++ 2 files changed, 9314 insertions(+), 4 deletions(-) create mode 100644 Backends/stb_image.h diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 1663af98f..01999a24a 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1999,10 +1999,10 @@ Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() // Copy the destination texture to the final texture resource D3D11_BOX copy_box{}; - copy_box.left = 0; - copy_box.right = bounds.Width(); - copy_box.top = 0; - copy_box.bottom = bounds.Height(); + copy_box.left = bounds.Left(); + copy_box.right = bounds.Right(); + copy_box.top = bounds.Top(); + copy_box.bottom = bounds.Bottom(); copy_box.front = 0; copy_box.back = 1; diff --git a/Backends/stb_image.h b/Backends/stb_image.h new file mode 100644 index 000000000..7a9c7bfac --- /dev/null +++ b/Backends/stb_image.h @@ -0,0 +1,9310 @@ +/* stb_image - v2.30 - public domain image loader - http://nothings.org/stb + no warranty implied; use at your own risk + + Do this: + #define STB_IMAGE_IMPLEMENTATION + before you include this file in *one* C or C++ file to create the implementation. + + // i.e. it should look like this: + #include ... + #include ... + #include ... + #define STB_IMAGE_IMPLEMENTATION + #include "stb_image.h" + + You can #define STBI_ASSERT(x) before the #include to avoid using assert.h. + And #define STBI_MALLOC, STBI_REALLOC, and STBI_FREE to avoid using malloc,realloc,free + + + QUICK NOTES: + Primarily of interest to game developers and other people who can + avoid problematic images and only need the trivial interface + + JPEG baseline & progressive (12 bpc/arithmetic not supported, same as stock IJG lib) + PNG 1/2/4/8/16-bit-per-channel + + TGA (not sure what subset, if a subset) + BMP non-1bpp, non-RLE + PSD (composited view only, no extra channels, 8/16 bit-per-channel) + + GIF (*comp always reports as 4-channel) + HDR (radiance rgbE format) + PIC (Softimage PIC) + PNM (PPM and PGM binary only) + + Animated GIF still needs a proper API, but here's one way to do it: + http://gist.github.com/urraka/685d9a6340b26b830d49 + + - decode from memory or through FILE (define STBI_NO_STDIO to remove code) + - decode from arbitrary I/O callbacks + - SIMD acceleration on x86/x64 (SSE2) and ARM (NEON) + + Full documentation under "DOCUMENTATION" below. + + +LICENSE + + See end of file for license information. + +RECENT REVISION HISTORY: + + 2.30 (2024-05-31) avoid erroneous gcc warning + 2.29 (2023-05-xx) optimizations + 2.28 (2023-01-29) many error fixes, security errors, just tons of stuff + 2.27 (2021-07-11) document stbi_info better, 16-bit PNM support, bug fixes + 2.26 (2020-07-13) many minor fixes + 2.25 (2020-02-02) fix warnings + 2.24 (2020-02-02) fix warnings; thread-local failure_reason and flip_vertically + 2.23 (2019-08-11) fix clang static analysis warning + 2.22 (2019-03-04) gif fixes, fix warnings + 2.21 (2019-02-25) fix typo in comment + 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs + 2.19 (2018-02-11) fix warning + 2.18 (2018-01-30) fix warnings + 2.17 (2018-01-29) bugfix, 1-bit BMP, 16-bitness query, fix warnings + 2.16 (2017-07-23) all functions have 16-bit variants; optimizations; bugfixes + 2.15 (2017-03-18) fix png-1,2,4; all Imagenet JPGs; no runtime SSE detection on GCC + 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs + 2.13 (2016-12-04) experimental 16-bit API, only for PNG so far; fixes + 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes + 2.11 (2016-04-02) 16-bit PNGS; enable SSE2 in non-gcc x64 + RGB-format JPEG; remove white matting in PSD; + allocate large structures on the stack; + correct channel count for PNG & BMP + 2.10 (2016-01-22) avoid warning introduced in 2.09 + 2.09 (2016-01-16) 16-bit TGA; comments in PNM files; STBI_REALLOC_SIZED + + See end of file for full revision history. + + + ============================ Contributors ========================= + + Image formats Extensions, features + Sean Barrett (jpeg, png, bmp) Jetro Lauha (stbi_info) + Nicolas Schulz (hdr, psd) Martin "SpartanJ" Golini (stbi_info) + Jonathan Dummer (tga) James "moose2000" Brown (iPhone PNG) + Jean-Marc Lienher (gif) Ben "Disch" Wenger (io callbacks) + Tom Seddon (pic) Omar Cornut (1/2/4-bit PNG) + Thatcher Ulrich (psd) Nicolas Guillemot (vertical flip) + Ken Miller (pgm, ppm) Richard Mitton (16-bit PSD) + github:urraka (animated gif) Junggon Kim (PNM comments) + Christopher Forseth (animated gif) Daniel Gibson (16-bit TGA) + socks-the-fox (16-bit PNG) + Jeremy Sawicki (handle all ImageNet JPGs) + Optimizations & bugfixes Mikhail Morozov (1-bit BMP) + Fabian "ryg" Giesen Anael Seghezzi (is-16-bit query) + Arseny Kapoulkine Simon Breuss (16-bit PNM) + John-Mark Allen + Carmelo J Fdez-Aguera + + Bug & warning fixes + Marc LeBlanc David Woo Guillaume George Martins Mozeiko + Christpher Lloyd Jerry Jansson Joseph Thomson Blazej Dariusz Roszkowski + Phil Jordan Dave Moore Roy Eltham + Hayaki Saito Nathan Reed Won Chun + Luke Graham Johan Duparc Nick Verigakis the Horde3D community + Thomas Ruf Ronny Chevalier github:rlyeh + Janez Zemva John Bartholomew Michal Cichon github:romigrou + Jonathan Blow Ken Hamada Tero Hanninen github:svdijk + Eugene Golushkov Laurent Gomila Cort Stratton github:snagar + Aruelien Pocheville Sergio Gonzalez Thibault Reuille github:Zelex + Cass Everitt Ryamond Barbiero github:grim210 + Paul Du Bois Engin Manap Aldo Culquicondor github:sammyhw + Philipp Wiesemann Dale Weiler Oriol Ferrer Mesia github:phprus + Josh Tobin Neil Bickford Matthew Gregan github:poppolopoppo + Julian Raschke Gregory Mullen Christian Floisand github:darealshinji + Baldur Karlsson Kevin Schmidt JR Smith github:Michaelangel007 + Brad Weinberger Matvey Cherevko github:mosra + Luca Sas Alexander Veselov Zack Middleton [reserved] + Ryan C. Gordon [reserved] [reserved] + DO NOT ADD YOUR NAME HERE + + Jacko Dirks + + To add your name to the credits, pick a random blank space in the middle and fill it. + 80% of merge conflicts on stb PRs are due to people adding their name at the end + of the credits. +*/ + +#ifndef STBI_INCLUDE_STB_IMAGE_H + #define STBI_INCLUDE_STB_IMAGE_H + +// DOCUMENTATION +// +// Limitations: +// - no 12-bit-per-channel JPEG +// - no JPEGs with arithmetic coding +// - GIF always returns *comp=4 +// +// Basic usage (see HDR discussion below for HDR usage): +// int x,y,n; +// unsigned char *data = stbi_load(filename, &x, &y, &n, 0); +// // ... process data if not NULL ... +// // ... x = width, y = height, n = # 8-bit components per pixel ... +// // ... replace '0' with '1'..'4' to force that many components per pixel +// // ... but 'n' will always be the number that it would have been if you said 0 +// stbi_image_free(data); +// +// Standard parameters: +// int *x -- outputs image width in pixels +// int *y -- outputs image height in pixels +// int *channels_in_file -- outputs # of image components in image file +// int desired_channels -- if non-zero, # of image components requested in result +// +// The return value from an image loader is an 'unsigned char *' which points +// to the pixel data, or NULL on an allocation failure or if the image is +// corrupt or invalid. The pixel data consists of *y scanlines of *x pixels, +// with each pixel consisting of N interleaved 8-bit components; the first +// pixel pointed to is top-left-most in the image. There is no padding between +// image scanlines or between pixels, regardless of format. The number of +// components N is 'desired_channels' if desired_channels is non-zero, or +// *channels_in_file otherwise. If desired_channels is non-zero, +// *channels_in_file has the number of components that _would_ have been +// output otherwise. E.g. if you set desired_channels to 4, you will always +// get RGBA output, but you can check *channels_in_file to see if it's trivially +// opaque because e.g. there were only 3 channels in the source image. +// +// An output image with N components has the following components interleaved +// in this order in each pixel: +// +// N=#comp components +// 1 grey +// 2 grey, alpha +// 3 red, green, blue +// 4 red, green, blue, alpha +// +// If image loading fails for any reason, the return value will be NULL, +// and *x, *y, *channels_in_file will be unchanged. The function +// stbi_failure_reason() can be queried for an extremely brief, end-user +// unfriendly explanation of why the load failed. Define STBI_NO_FAILURE_STRINGS +// to avoid compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly +// more user-friendly ones. +// +// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized. +// +// To query the width, height and component count of an image without having to +// decode the full file, you can use the stbi_info family of functions: +// +// int x,y,n,ok; +// ok = stbi_info(filename, &x, &y, &n); +// // returns ok=1 and sets x, y, n if image is a supported format, +// // 0 otherwise. +// +// Note that stb_image pervasively uses ints in its public API for sizes, +// including sizes of memory buffers. This is now part of the API and thus +// hard to change without causing breakage. As a result, the various image +// loaders all have certain limits on image size; these differ somewhat +// by format but generally boil down to either just under 2GB or just under +// 1GB. When the decoded image would be larger than this, stb_image decoding +// will fail. +// +// Additionally, stb_image will reject image files that have any of their +// dimensions set to a larger value than the configurable STBI_MAX_DIMENSIONS, +// which defaults to 2**24 = 16777216 pixels. Due to the above memory limit, +// the only way to have an image with such dimensions load correctly +// is for it to have a rather extreme aspect ratio. Either way, the +// assumption here is that such larger images are likely to be malformed +// or malicious. If you do need to load an image with individual dimensions +// larger than that, and it still fits in the overall size limit, you can +// #define STBI_MAX_DIMENSIONS on your own to be something larger. +// +// =========================================================================== +// +// UNICODE: +// +// If compiling for Windows and you wish to use Unicode filenames, compile +// with +// #define STBI_WINDOWS_UTF8 +// and pass utf8-encoded filenames. Call stbi_convert_wchar_to_utf8 to convert +// Windows wchar_t filenames to utf8. +// +// =========================================================================== +// +// Philosophy +// +// stb libraries are designed with the following priorities: +// +// 1. easy to use +// 2. easy to maintain +// 3. good performance +// +// Sometimes I let "good performance" creep up in priority over "easy to maintain", +// and for best performance I may provide less-easy-to-use APIs that give higher +// performance, in addition to the easy-to-use ones. Nevertheless, it's important +// to keep in mind that from the standpoint of you, a client of this library, +// all you care about is #1 and #3, and stb libraries DO NOT emphasize #3 above all. +// +// Some secondary priorities arise directly from the first two, some of which +// provide more explicit reasons why performance can't be emphasized. +// +// - Portable ("ease of use") +// - Small source code footprint ("easy to maintain") +// - No dependencies ("ease of use") +// +// =========================================================================== +// +// I/O callbacks +// +// I/O callbacks allow you to read from arbitrary sources, like packaged +// files or some other source. Data read from callbacks are processed +// through a small internal buffer (currently 128 bytes) to try to reduce +// overhead. +// +// The three functions you must define are "read" (reads some bytes of data), +// "skip" (skips some bytes of data), "eof" (reports if the stream is at the end). +// +// =========================================================================== +// +// SIMD support +// +// The JPEG decoder will try to automatically use SIMD kernels on x86 when +// supported by the compiler. For ARM Neon support, you must explicitly +// request it. +// +// (The old do-it-yourself SIMD API is no longer supported in the current +// code.) +// +// On x86, SSE2 will automatically be used when available based on a run-time +// test; if not, the generic C versions are used as a fall-back. On ARM targets, +// the typical path is to have separate builds for NEON and non-NEON devices +// (at least this is true for iOS and Android). Therefore, the NEON support is +// toggled by a build flag: define STBI_NEON to get NEON loops. +// +// If for some reason you do not want to use any of SIMD code, or if +// you have issues compiling it, you can disable it entirely by +// defining STBI_NO_SIMD. +// +// =========================================================================== +// +// HDR image support (disable by defining STBI_NO_HDR) +// +// stb_image supports loading HDR images in general, and currently the Radiance +// .HDR file format specifically. You can still load any file through the existing +// interface; if you attempt to load an HDR file, it will be automatically remapped +// to LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1; +// both of these constants can be reconfigured through this interface: +// +// stbi_hdr_to_ldr_gamma(2.2f); +// stbi_hdr_to_ldr_scale(1.0f); +// +// (note, do not use _inverse_ constants; stbi_image will invert them +// appropriately). +// +// Additionally, there is a new, parallel interface for loading files as +// (linear) floats to preserve the full dynamic range: +// +// float *data = stbi_loadf(filename, &x, &y, &n, 0); +// +// If you load LDR images through this interface, those images will +// be promoted to floating point values, run through the inverse of +// constants corresponding to the above: +// +// stbi_ldr_to_hdr_scale(1.0f); +// stbi_ldr_to_hdr_gamma(2.2f); +// +// Finally, given a filename (or an open file or memory block--see header +// file for details) containing image data, you can query for the "most +// appropriate" interface to use (that is, whether the image is HDR or +// not), using: +// +// stbi_is_hdr(char *filename); +// +// =========================================================================== +// +// iPhone PNG support: +// +// We optionally support converting iPhone-formatted PNGs (which store +// premultiplied BGRA) back to RGB, even though they're internally encoded +// differently. To enable this conversion, call +// stbi_convert_iphone_png_to_rgb(1). +// +// Call stbi_set_unpremultiply_on_load(1) as well to force a divide per +// pixel to remove any premultiplied alpha *only* if the image file explicitly +// says there's premultiplied data (currently only happens in iPhone images, +// and only if iPhone convert-to-rgb processing is on). +// +// =========================================================================== +// +// ADDITIONAL CONFIGURATION +// +// - You can suppress implementation of any of the decoders to reduce +// your code footprint by #defining one or more of the following +// symbols before creating the implementation. +// +// STBI_NO_JPEG +// STBI_NO_PNG +// STBI_NO_BMP +// STBI_NO_PSD +// STBI_NO_TGA +// STBI_NO_GIF +// STBI_NO_HDR +// STBI_NO_PIC +// STBI_NO_PNM (.ppm and .pgm) +// +// - You can request *only* certain decoders and suppress all other ones +// (this will be more forward-compatible, as addition of new decoders +// doesn't require you to disable them explicitly): +// +// STBI_ONLY_JPEG +// STBI_ONLY_PNG +// STBI_ONLY_BMP +// STBI_ONLY_PSD +// STBI_ONLY_TGA +// STBI_ONLY_GIF +// STBI_ONLY_HDR +// STBI_ONLY_PIC +// STBI_ONLY_PNM (.ppm and .pgm) +// +// - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still +// want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB +// +// - If you define STBI_MAX_DIMENSIONS, stb_image will reject images greater +// than that size (in either width or height) without further processing. +// This is to let programs in the wild set an upper bound to prevent +// denial-of-service attacks on untrusted data, as one could generate a +// valid image of gigantic dimensions and force stb_image to allocate a +// huge block of memory and spend disproportionate time decoding it. By +// default this is set to (1 << 24), which is 16777216, but that's still +// very big. + + #ifndef STBI_NO_STDIO + #include + #endif // STBI_NO_STDIO + + #define STBI_VERSION 1 + +enum { + STBI_default = 0, // only used for desired_channels + + STBI_grey = 1, + STBI_grey_alpha = 2, + STBI_rgb = 3, + STBI_rgb_alpha = 4 +}; + + #include +typedef unsigned char stbi_uc; +typedef unsigned short stbi_us; + + #ifdef __cplusplus +extern "C" { + #endif + + #ifndef STBIDEF + #ifdef STB_IMAGE_STATIC + #define STBIDEF static + #else + #define STBIDEF extern + #endif + #endif + +////////////////////////////////////////////////////////////////////////////// +// +// PRIMARY API - works on images of any type +// + +// +// load image by filename, open file, or memory buffer +// + +typedef struct { + int (*read)(void* user, char* data, int size); // fill 'data' with 'size' bytes. return number of bytes actually read + void (*skip)(void* user, int n); // skip the next 'n' bytes, or 'unget' the last -n bytes if negative + int (*eof)(void* user); // returns nonzero if we are at end of file/data +} stbi_io_callbacks; + +//////////////////////////////////// +// +// 8-bits-per-channel interface +// + +STBIDEF stbi_uc* stbi_load_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels); +STBIDEF stbi_uc* stbi_load_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels); + + #ifndef STBI_NO_STDIO +STBIDEF stbi_uc* stbi_load(const char* filename, int* x, int* y, int* channels_in_file, int desired_channels); +STBIDEF stbi_uc* stbi_load_from_file(FILE* f, int* x, int* y, int* channels_in_file, int desired_channels); + // for stbi_load_from_file, file pointer is left pointing immediately after image + #endif + + #ifndef STBI_NO_GIF +STBIDEF stbi_uc* stbi_load_gif_from_memory(const stbi_uc* buffer, int len, int** delays, int* x, int* y, int* z, int* comp, int req_comp); + #endif + + #ifdef STBI_WINDOWS_UTF8 +STBIDEF int stbi_convert_wchar_to_utf8(char* buffer, size_t bufferlen, const wchar_t* input); + #endif + +//////////////////////////////////// +// +// 16-bits-per-channel interface +// + +STBIDEF stbi_us* stbi_load_16_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels); +STBIDEF stbi_us* stbi_load_16_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels); + + #ifndef STBI_NO_STDIO +STBIDEF stbi_us* stbi_load_16(const char* filename, int* x, int* y, int* channels_in_file, int desired_channels); +STBIDEF stbi_us* stbi_load_from_file_16(FILE* f, int* x, int* y, int* channels_in_file, int desired_channels); + #endif + + //////////////////////////////////// + // + // float-per-channel interface + // + #ifndef STBI_NO_LINEAR +STBIDEF float* stbi_loadf_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels); +STBIDEF float* stbi_loadf_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels); + + #ifndef STBI_NO_STDIO +STBIDEF float* stbi_loadf(const char* filename, int* x, int* y, int* channels_in_file, int desired_channels); +STBIDEF float* stbi_loadf_from_file(FILE* f, int* x, int* y, int* channels_in_file, int desired_channels); + #endif + #endif + + #ifndef STBI_NO_HDR +STBIDEF void stbi_hdr_to_ldr_gamma(float gamma); +STBIDEF void stbi_hdr_to_ldr_scale(float scale); + #endif // STBI_NO_HDR + + #ifndef STBI_NO_LINEAR +STBIDEF void stbi_ldr_to_hdr_gamma(float gamma); +STBIDEF void stbi_ldr_to_hdr_scale(float scale); + #endif // STBI_NO_LINEAR + +// stbi_is_hdr is always defined, but always returns false if STBI_NO_HDR +STBIDEF int stbi_is_hdr_from_callbacks(const stbi_io_callbacks* clbk, void* user); +STBIDEF int stbi_is_hdr_from_memory(const stbi_uc* buffer, int len); + #ifndef STBI_NO_STDIO +STBIDEF int stbi_is_hdr(const char* filename); +STBIDEF int stbi_is_hdr_from_file(FILE* f); + #endif // STBI_NO_STDIO + +// get a VERY brief reason for failure +// on most compilers (and ALL modern mainstream compilers) this is threadsafe +STBIDEF const char* stbi_failure_reason(void); + +// free the loaded image -- this is just free() +STBIDEF void stbi_image_free(void* retval_from_stbi_load); + +// get image dimensions & components without fully decoding +STBIDEF int stbi_info_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp); +STBIDEF int stbi_info_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* comp); +STBIDEF int stbi_is_16_bit_from_memory(const stbi_uc* buffer, int len); +STBIDEF int stbi_is_16_bit_from_callbacks(const stbi_io_callbacks* clbk, void* user); + + #ifndef STBI_NO_STDIO +STBIDEF int stbi_info(const char* filename, int* x, int* y, int* comp); +STBIDEF int stbi_info_from_file(FILE* f, int* x, int* y, int* comp); +STBIDEF int stbi_is_16_bit(const char* filename); +STBIDEF int stbi_is_16_bit_from_file(FILE* f); + #endif + +// for image formats that explicitly notate that they have premultiplied alpha, +// we just return the colors as stored in the file. set this flag to force +// unpremultiplication. results are undefined if the unpremultiply overflow. +STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply); + +// indicate whether we should process iphone images back to canonical format, +// or just pass them through "as-is" +STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert); + +// flip the image vertically, so the first pixel in the output array is the bottom left +STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip); + +// as above, but only applies to images loaded on the thread that calls the function +// this function is only available if your compiler supports thread-local variables; +// calling it will fail to link if your compiler doesn't +STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply); +STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert); +STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip); + +// ZLIB client - used by PNG, available for other purposes + +STBIDEF char* stbi_zlib_decode_malloc_guesssize(const char* buffer, int len, int initial_size, int* outlen); +STBIDEF char* stbi_zlib_decode_malloc_guesssize_headerflag(const char* buffer, int len, int initial_size, int* outlen, int parse_header); +STBIDEF char* stbi_zlib_decode_malloc(const char* buffer, int len, int* outlen); +STBIDEF int stbi_zlib_decode_buffer(char* obuffer, int olen, const char* ibuffer, int ilen); + +STBIDEF char* stbi_zlib_decode_noheader_malloc(const char* buffer, int len, int* outlen); +STBIDEF int stbi_zlib_decode_noheader_buffer(char* obuffer, int olen, const char* ibuffer, int ilen); + + #ifdef __cplusplus +} + #endif + +// +// +//// end header file ///////////////////////////////////////////////////// +#endif // STBI_INCLUDE_STB_IMAGE_H + +#ifdef STB_IMAGE_IMPLEMENTATION + + #if defined(STBI_ONLY_JPEG) || defined(STBI_ONLY_PNG) || defined(STBI_ONLY_BMP) || defined(STBI_ONLY_TGA) || defined(STBI_ONLY_GIF) || \ + defined(STBI_ONLY_PSD) || defined(STBI_ONLY_HDR) || defined(STBI_ONLY_PIC) || defined(STBI_ONLY_PNM) || defined(STBI_ONLY_ZLIB) + #ifndef STBI_ONLY_JPEG + #define STBI_NO_JPEG + #endif + #ifndef STBI_ONLY_PNG + #define STBI_NO_PNG + #endif + #ifndef STBI_ONLY_BMP + #define STBI_NO_BMP + #endif + #ifndef STBI_ONLY_PSD + #define STBI_NO_PSD + #endif + #ifndef STBI_ONLY_TGA + #define STBI_NO_TGA + #endif + #ifndef STBI_ONLY_GIF + #define STBI_NO_GIF + #endif + #ifndef STBI_ONLY_HDR + #define STBI_NO_HDR + #endif + #ifndef STBI_ONLY_PIC + #define STBI_NO_PIC + #endif + #ifndef STBI_ONLY_PNM + #define STBI_NO_PNM + #endif + #endif + + #if defined(STBI_NO_PNG) && !defined(STBI_SUPPORT_ZLIB) && !defined(STBI_NO_ZLIB) + #define STBI_NO_ZLIB + #endif + + #include + #include + #include // ptrdiff_t on osx + #include + #include + + #if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) + #include // ldexp, pow + #endif + + #ifndef STBI_NO_STDIO + #include + #endif + + #ifndef STBI_ASSERT + #include + #define STBI_ASSERT(x) assert(x) + #endif + + #ifdef __cplusplus + #define STBI_EXTERN extern "C" + #else + #define STBI_EXTERN extern + #endif + + #ifndef _MSC_VER + #ifdef __cplusplus + #define stbi_inline inline + #else + #define stbi_inline + #endif + #else + #define stbi_inline __forceinline + #endif + + #ifndef STBI_NO_THREAD_LOCALS + #if defined(__cplusplus) && __cplusplus >= 201103L + #define STBI_THREAD_LOCAL thread_local + #elif defined(__GNUC__) && __GNUC__ < 5 + #define STBI_THREAD_LOCAL __thread + #elif defined(_MSC_VER) + #define STBI_THREAD_LOCAL __declspec(thread) + #elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L && !defined(__STDC_NO_THREADS__) + #define STBI_THREAD_LOCAL _Thread_local + #endif + + #ifndef STBI_THREAD_LOCAL + #if defined(__GNUC__) + #define STBI_THREAD_LOCAL __thread + #endif + #endif + #endif + + #if defined(_MSC_VER) || defined(__SYMBIAN32__) +typedef unsigned short stbi__uint16; +typedef signed short stbi__int16; +typedef unsigned int stbi__uint32; +typedef signed int stbi__int32; + #else + #include +typedef uint16_t stbi__uint16; +typedef int16_t stbi__int16; +typedef uint32_t stbi__uint32; +typedef int32_t stbi__int32; + #endif + +// should produce compiler error if size is wrong +typedef unsigned char validate_uint32[sizeof(stbi__uint32) == 4 ? 1 : -1]; + + #ifdef _MSC_VER + #define STBI_NOTUSED(v) (void)(v) + #else + #define STBI_NOTUSED(v) (void)sizeof(v) + #endif + + #ifdef _MSC_VER + #define STBI_HAS_LROTL + #endif + + #ifdef STBI_HAS_LROTL + #define stbi_lrot(x, y) _lrotl(x, y) + #else + #define stbi_lrot(x, y) (((x) << (y)) | ((x) >> (-(y) & 31))) + #endif + + #if defined(STBI_MALLOC) && defined(STBI_FREE) && (defined(STBI_REALLOC) || defined(STBI_REALLOC_SIZED)) + // ok + #elif !defined(STBI_MALLOC) && !defined(STBI_FREE) && !defined(STBI_REALLOC) && !defined(STBI_REALLOC_SIZED) + // ok + #else + #error "Must define all or none of STBI_MALLOC, STBI_FREE, and STBI_REALLOC (or STBI_REALLOC_SIZED)." + #endif + + #ifndef STBI_MALLOC + #define STBI_MALLOC(sz) malloc(sz) + #define STBI_REALLOC(p, newsz) realloc(p, newsz) + #define STBI_FREE(p) free(p) + #endif + + #ifndef STBI_REALLOC_SIZED + #define STBI_REALLOC_SIZED(p, oldsz, newsz) STBI_REALLOC(p, newsz) + #endif + + // x86/x64 detection + #if defined(__x86_64__) || defined(_M_X64) + #define STBI__X64_TARGET + #elif defined(__i386) || defined(_M_IX86) + #define STBI__X86_TARGET + #endif + + #if defined(__GNUC__) && defined(STBI__X86_TARGET) && !defined(__SSE2__) && !defined(STBI_NO_SIMD) + // gcc doesn't support sse2 intrinsics unless you compile with -msse2, + // which in turn means it gets to use SSE2 everywhere. This is unfortunate, + // but previous attempts to provide the SSE2 functions with runtime + // detection caused numerous issues. The way architecture extensions are + // exposed in GCC/Clang is, sadly, not really suited for one-file libs. + // New behavior: if compiled with -msse2, we use SSE2 without any + // detection; if not, we don't use it at all. + #define STBI_NO_SIMD + #endif + + #if defined(__MINGW32__) && defined(STBI__X86_TARGET) && !defined(STBI_MINGW_ENABLE_SSE2) && !defined(STBI_NO_SIMD) + // Note that __MINGW32__ doesn't actually mean 32-bit, so we have to avoid STBI__X64_TARGET + // + // 32-bit MinGW wants ESP to be 16-byte aligned, but this is not in the + // Windows ABI and VC++ as well as Windows DLLs don't maintain that invariant. + // As a result, enabling SSE2 on 32-bit MinGW is dangerous when not + // simultaneously enabling "-mstackrealign". + // + // See https://github.com/nothings/stb/issues/81 for more information. + // + // So default to no SSE2 on 32-bit MinGW. If you've read this far and added + // -mstackrealign to your build settings, feel free to #define STBI_MINGW_ENABLE_SSE2. + #define STBI_NO_SIMD + #endif + + #if !defined(STBI_NO_SIMD) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET)) + #define STBI_SSE2 + #include + + #ifdef _MSC_VER + + #if _MSC_VER >= 1400 // not VC6 + #include // __cpuid +static int stbi__cpuid3(void) +{ + int info[4]; + __cpuid(info, 1); + return info[3]; +} + #else +static int stbi__cpuid3(void) +{ + int res; + __asm { + mov eax,1 + cpuid + mov res,edx + } + return res; +} + #endif + + #define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name + + #if !defined(STBI_NO_JPEG) && defined(STBI_SSE2) +static int stbi__sse2_available(void) +{ + int info3 = stbi__cpuid3(); + return ((info3 >> 26) & 1) != 0; +} + #endif + + #else // assume GCC-style if not VC++ + #define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16))) + + #if !defined(STBI_NO_JPEG) && defined(STBI_SSE2) +static int stbi__sse2_available(void) +{ + // If we're even attempting to compile this on GCC/Clang, that means + // -msse2 is on, which means the compiler is allowed to use SSE2 + // instructions at will, and so are we. + return 1; +} + #endif + + #endif + #endif + + // ARM NEON + #if defined(STBI_NO_SIMD) && defined(STBI_NEON) + #undef STBI_NEON + #endif + + #ifdef STBI_NEON + #include + #ifdef _MSC_VER + #define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name + #else + #define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16))) + #endif + #endif + + #ifndef STBI_SIMD_ALIGN + #define STBI_SIMD_ALIGN(type, name) type name + #endif + + #ifndef STBI_MAX_DIMENSIONS + #define STBI_MAX_DIMENSIONS (1 << 24) + #endif + +/////////////////////////////////////////////// +// +// stbi__context struct and start_xxx functions + +// stbi__context structure is our basic context used by all images, so it +// contains all the IO context, plus some basic image information +typedef struct { + stbi__uint32 img_x, img_y; + int img_n, img_out_n; + + stbi_io_callbacks io; + void* io_user_data; + + int read_from_callbacks; + int buflen; + stbi_uc buffer_start[128]; + int callback_already_read; + + stbi_uc *img_buffer, *img_buffer_end; + stbi_uc *img_buffer_original, *img_buffer_original_end; +} stbi__context; + +static void stbi__refill_buffer(stbi__context* s); + +// initialize a memory-decode context +static void stbi__start_mem(stbi__context* s, const stbi_uc* buffer, int len) +{ + s->io.read = NULL; + s->read_from_callbacks = 0; + s->callback_already_read = 0; + s->img_buffer = s->img_buffer_original = (stbi_uc*)buffer; + s->img_buffer_end = s->img_buffer_original_end = (stbi_uc*)buffer + len; +} + +// initialize a callback-based context +static void stbi__start_callbacks(stbi__context* s, stbi_io_callbacks* c, void* user) +{ + s->io = *c; + s->io_user_data = user; + s->buflen = sizeof(s->buffer_start); + s->read_from_callbacks = 1; + s->callback_already_read = 0; + s->img_buffer = s->img_buffer_original = s->buffer_start; + stbi__refill_buffer(s); + s->img_buffer_original_end = s->img_buffer_end; +} + + #ifndef STBI_NO_STDIO + +static int stbi__stdio_read(void* user, char* data, int size) +{ + return (int)fread(data, 1, size, (FILE*)user); +} + +static void stbi__stdio_skip(void* user, int n) +{ + int ch; + fseek((FILE*)user, n, SEEK_CUR); + ch = fgetc((FILE*)user); /* have to read a byte to reset feof()'s flag */ + if (ch != EOF) + { + ungetc(ch, (FILE*)user); /* push byte back onto stream if valid. */ + } +} + +static int stbi__stdio_eof(void* user) +{ + return feof((FILE*)user) || ferror((FILE*)user); +} + +static stbi_io_callbacks stbi__stdio_callbacks = { + stbi__stdio_read, + stbi__stdio_skip, + stbi__stdio_eof, +}; + +static void stbi__start_file(stbi__context* s, FILE* f) +{ + stbi__start_callbacks(s, &stbi__stdio_callbacks, (void*)f); +} + + // static void stop_file(stbi__context *s) { } + + #endif // !STBI_NO_STDIO + +static void stbi__rewind(stbi__context* s) +{ + // conceptually rewind SHOULD rewind to the beginning of the stream, + // but we just rewind to the beginning of the initial buffer, because + // we only use it after doing 'test', which only ever looks at at most 92 bytes + s->img_buffer = s->img_buffer_original; + s->img_buffer_end = s->img_buffer_original_end; +} + +enum { STBI_ORDER_RGB, STBI_ORDER_BGR }; + +typedef struct { + int bits_per_channel; + int num_channels; + int channel_order; +} stbi__result_info; + + #ifndef STBI_NO_JPEG +static int stbi__jpeg_test(stbi__context* s); +static void* stbi__jpeg_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static int stbi__jpeg_info(stbi__context* s, int* x, int* y, int* comp); + #endif + + #ifndef STBI_NO_PNG +static int stbi__png_test(stbi__context* s); +static void* stbi__png_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static int stbi__png_info(stbi__context* s, int* x, int* y, int* comp); +static int stbi__png_is16(stbi__context* s); + #endif + + #ifndef STBI_NO_BMP +static int stbi__bmp_test(stbi__context* s); +static void* stbi__bmp_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static int stbi__bmp_info(stbi__context* s, int* x, int* y, int* comp); + #endif + + #ifndef STBI_NO_TGA +static int stbi__tga_test(stbi__context* s); +static void* stbi__tga_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static int stbi__tga_info(stbi__context* s, int* x, int* y, int* comp); + #endif + + #ifndef STBI_NO_PSD +static int stbi__psd_test(stbi__context* s); +static void* stbi__psd_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri, int bpc); +static int stbi__psd_info(stbi__context* s, int* x, int* y, int* comp); +static int stbi__psd_is16(stbi__context* s); + #endif + + #ifndef STBI_NO_HDR +static int stbi__hdr_test(stbi__context* s); +static float* stbi__hdr_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static int stbi__hdr_info(stbi__context* s, int* x, int* y, int* comp); + #endif + + #ifndef STBI_NO_PIC +static int stbi__pic_test(stbi__context* s); +static void* stbi__pic_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static int stbi__pic_info(stbi__context* s, int* x, int* y, int* comp); + #endif + + #ifndef STBI_NO_GIF +static int stbi__gif_test(stbi__context* s); +static void* stbi__gif_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static void* stbi__load_gif_main(stbi__context* s, int** delays, int* x, int* y, int* z, int* comp, int req_comp); +static int stbi__gif_info(stbi__context* s, int* x, int* y, int* comp); + #endif + + #ifndef STBI_NO_PNM +static int stbi__pnm_test(stbi__context* s); +static void* stbi__pnm_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); +static int stbi__pnm_info(stbi__context* s, int* x, int* y, int* comp); +static int stbi__pnm_is16(stbi__context* s); + #endif + +static + #ifdef STBI_THREAD_LOCAL + STBI_THREAD_LOCAL + #endif + const char* stbi__g_failure_reason; + +STBIDEF const char* stbi_failure_reason(void) +{ + return stbi__g_failure_reason; +} + + #ifndef STBI_NO_FAILURE_STRINGS +static int stbi__err(const char* str) +{ + stbi__g_failure_reason = str; + return 0; +} + #endif + +static void* stbi__malloc(size_t size) +{ + return STBI_MALLOC(size); +} + +// stb_image uses ints pervasively, including for offset calculations. +// therefore the largest decoded image size we can support with the +// current code, even on 64-bit targets, is INT_MAX. this is not a +// significant limitation for the intended use case. +// +// we do, however, need to make sure our size calculations don't +// overflow. hence a few helper functions for size calculations that +// multiply integers together, making sure that they're non-negative +// and no overflow occurs. + +// return 1 if the sum is valid, 0 on overflow. +// negative terms are considered invalid. +static int stbi__addsizes_valid(int a, int b) +{ + if (b < 0) + return 0; + // now 0 <= b <= INT_MAX, hence also + // 0 <= INT_MAX - b <= INTMAX. + // And "a + b <= INT_MAX" (which might overflow) is the + // same as a <= INT_MAX - b (no overflow) + return a <= INT_MAX - b; +} + +// returns 1 if the product is valid, 0 on overflow. +// negative factors are considered invalid. +static int stbi__mul2sizes_valid(int a, int b) +{ + if (a < 0 || b < 0) + return 0; + if (b == 0) + return 1; // mul-by-0 is always safe + // portable way to check for no overflows in a*b + return a <= INT_MAX / b; +} + + #if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR) +// returns 1 if "a*b + add" has no negative terms/factors and doesn't overflow +static int stbi__mad2sizes_valid(int a, int b, int add) +{ + return stbi__mul2sizes_valid(a, b) && stbi__addsizes_valid(a * b, add); +} + #endif + +// returns 1 if "a*b*c + add" has no negative terms/factors and doesn't overflow +static int stbi__mad3sizes_valid(int a, int b, int c, int add) +{ + return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a * b, c) && stbi__addsizes_valid(a * b * c, add); +} + + // returns 1 if "a*b*c*d + add" has no negative terms/factors and doesn't overflow + #if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM) +static int stbi__mad4sizes_valid(int a, int b, int c, int d, int add) +{ + return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a * b, c) && stbi__mul2sizes_valid(a * b * c, d) && + stbi__addsizes_valid(a * b * c * d, add); +} + #endif + + #if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR) +// mallocs with size overflow checking +static void* stbi__malloc_mad2(int a, int b, int add) +{ + if (!stbi__mad2sizes_valid(a, b, add)) + return NULL; + return stbi__malloc(a * b + add); +} + #endif + +static void* stbi__malloc_mad3(int a, int b, int c, int add) +{ + if (!stbi__mad3sizes_valid(a, b, c, add)) + return NULL; + return stbi__malloc(a * b * c + add); +} + + #if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM) +static void* stbi__malloc_mad4(int a, int b, int c, int d, int add) +{ + if (!stbi__mad4sizes_valid(a, b, c, d, add)) + return NULL; + return stbi__malloc(a * b * c * d + add); +} + #endif + +// returns 1 if the sum of two signed ints is valid (between -2^31 and 2^31-1 inclusive), 0 on overflow. +static int stbi__addints_valid(int a, int b) +{ + if ((a >= 0) != (b >= 0)) + return 1; // a and b have different signs, so no overflow + if (a < 0 && b < 0) + return a >= INT_MIN - b; // same as a + b >= INT_MIN; INT_MIN - b cannot overflow since b < 0. + return a <= INT_MAX - b; +} + +// returns 1 if the product of two ints fits in a signed short, 0 on overflow. +static int stbi__mul2shorts_valid(int a, int b) +{ + if (b == 0 || b == -1) + return 1; // multiplication by 0 is always 0; check for -1 so SHRT_MIN/b doesn't overflow + if ((a >= 0) == (b >= 0)) + return a <= SHRT_MAX / b; // product is positive, so similar to mul2sizes_valid + if (b < 0) + return a <= SHRT_MIN / b; // same as a * b >= SHRT_MIN + return a >= SHRT_MIN / b; +} + +// stbi__err - error +// stbi__errpf - error returning pointer to float +// stbi__errpuc - error returning pointer to unsigned char + + #ifdef STBI_NO_FAILURE_STRINGS + #define stbi__err(x, y) 0 + #elif defined(STBI_FAILURE_USERMSG) + #define stbi__err(x, y) stbi__err(y) + #else + #define stbi__err(x, y) stbi__err(x) + #endif + + #define stbi__errpf(x, y) ((float*)(size_t)(stbi__err(x, y) ? NULL : NULL)) + #define stbi__errpuc(x, y) ((unsigned char*)(size_t)(stbi__err(x, y) ? NULL : NULL)) + +STBIDEF void stbi_image_free(void* retval_from_stbi_load) +{ + STBI_FREE(retval_from_stbi_load); +} + + #ifndef STBI_NO_LINEAR +static float* stbi__ldr_to_hdr(stbi_uc* data, int x, int y, int comp); + #endif + + #ifndef STBI_NO_HDR +static stbi_uc* stbi__hdr_to_ldr(float* data, int x, int y, int comp); + #endif + +static int stbi__vertically_flip_on_load_global = 0; + +STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip) +{ + stbi__vertically_flip_on_load_global = flag_true_if_should_flip; +} + + #ifndef STBI_THREAD_LOCAL + #define stbi__vertically_flip_on_load stbi__vertically_flip_on_load_global + #else +static STBI_THREAD_LOCAL int stbi__vertically_flip_on_load_local, stbi__vertically_flip_on_load_set; + +STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip) +{ + stbi__vertically_flip_on_load_local = flag_true_if_should_flip; + stbi__vertically_flip_on_load_set = 1; +} + + #define stbi__vertically_flip_on_load \ + (stbi__vertically_flip_on_load_set ? stbi__vertically_flip_on_load_local : stbi__vertically_flip_on_load_global) + #endif // STBI_THREAD_LOCAL + +static void* stbi__load_main(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri, int bpc) +{ + memset(ri, 0, sizeof(*ri)); // make sure it's initialized if we add new fields + ri->bits_per_channel = 8; // default is 8 so most paths don't have to be changed + ri->channel_order = STBI_ORDER_RGB; // all current input & output are this, but this is here so we can add BGR order + ri->num_channels = 0; + + // test the formats with a very explicit header first (at least a FOURCC + // or distinctive magic number first) + #ifndef STBI_NO_PNG + if (stbi__png_test(s)) + return stbi__png_load(s, x, y, comp, req_comp, ri); + #endif + #ifndef STBI_NO_BMP + if (stbi__bmp_test(s)) + return stbi__bmp_load(s, x, y, comp, req_comp, ri); + #endif + #ifndef STBI_NO_GIF + if (stbi__gif_test(s)) + return stbi__gif_load(s, x, y, comp, req_comp, ri); + #endif + #ifndef STBI_NO_PSD + if (stbi__psd_test(s)) + return stbi__psd_load(s, x, y, comp, req_comp, ri, bpc); + #else + STBI_NOTUSED(bpc); + #endif + #ifndef STBI_NO_PIC + if (stbi__pic_test(s)) + return stbi__pic_load(s, x, y, comp, req_comp, ri); + #endif + + // then the formats that can end up attempting to load with just 1 or 2 + // bytes matching expectations; these are prone to false positives, so + // try them later + #ifndef STBI_NO_JPEG + if (stbi__jpeg_test(s)) + return stbi__jpeg_load(s, x, y, comp, req_comp, ri); + #endif + #ifndef STBI_NO_PNM + if (stbi__pnm_test(s)) + return stbi__pnm_load(s, x, y, comp, req_comp, ri); + #endif + + #ifndef STBI_NO_HDR + if (stbi__hdr_test(s)) + { + float* hdr = stbi__hdr_load(s, x, y, comp, req_comp, ri); + return stbi__hdr_to_ldr(hdr, *x, *y, req_comp ? req_comp : *comp); + } + #endif + + #ifndef STBI_NO_TGA + // test tga last because it's a crappy test! + if (stbi__tga_test(s)) + return stbi__tga_load(s, x, y, comp, req_comp, ri); + #endif + + return stbi__errpuc("unknown image type", "Image not of any known type, or corrupt"); +} + +static stbi_uc* stbi__convert_16_to_8(stbi__uint16* orig, int w, int h, int channels) +{ + int i; + int img_len = w * h * channels; + stbi_uc* reduced; + + reduced = (stbi_uc*)stbi__malloc(img_len); + if (reduced == NULL) + return stbi__errpuc("outofmem", "Out of memory"); + + for (i = 0; i < img_len; ++i) + reduced[i] = (stbi_uc)((orig[i] >> 8) & 0xFF); // top half of each byte is sufficient approx of 16->8 bit scaling + + STBI_FREE(orig); + return reduced; +} + +static stbi__uint16* stbi__convert_8_to_16(stbi_uc* orig, int w, int h, int channels) +{ + int i; + int img_len = w * h * channels; + stbi__uint16* enlarged; + + enlarged = (stbi__uint16*)stbi__malloc(img_len * 2); + if (enlarged == NULL) + return (stbi__uint16*)stbi__errpuc("outofmem", "Out of memory"); + + for (i = 0; i < img_len; ++i) + enlarged[i] = (stbi__uint16)((orig[i] << 8) + orig[i]); // replicate to high and low byte, maps 0->0, 255->0xffff + + STBI_FREE(orig); + return enlarged; +} + +static void stbi__vertical_flip(void* image, int w, int h, int bytes_per_pixel) +{ + int row; + size_t bytes_per_row = (size_t)w * bytes_per_pixel; + stbi_uc temp[2048]; + stbi_uc* bytes = (stbi_uc*)image; + + for (row = 0; row < (h >> 1); row++) + { + stbi_uc* row0 = bytes + row * bytes_per_row; + stbi_uc* row1 = bytes + (h - row - 1) * bytes_per_row; + // swap row0 with row1 + size_t bytes_left = bytes_per_row; + while (bytes_left) + { + size_t bytes_copy = (bytes_left < sizeof(temp)) ? bytes_left : sizeof(temp); + memcpy(temp, row0, bytes_copy); + memcpy(row0, row1, bytes_copy); + memcpy(row1, temp, bytes_copy); + row0 += bytes_copy; + row1 += bytes_copy; + bytes_left -= bytes_copy; + } + } +} + + #ifndef STBI_NO_GIF +static void stbi__vertical_flip_slices(void* image, int w, int h, int z, int bytes_per_pixel) +{ + int slice; + int slice_size = w * h * bytes_per_pixel; + + stbi_uc* bytes = (stbi_uc*)image; + for (slice = 0; slice < z; ++slice) + { + stbi__vertical_flip(bytes, w, h, bytes_per_pixel); + bytes += slice_size; + } +} + #endif + +static unsigned char* stbi__load_and_postprocess_8bit(stbi__context* s, int* x, int* y, int* comp, int req_comp) +{ + stbi__result_info ri; + void* result = stbi__load_main(s, x, y, comp, req_comp, &ri, 8); + + if (result == NULL) + return NULL; + + // it is the responsibility of the loaders to make sure we get either 8 or 16 bit. + STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16); + + if (ri.bits_per_channel != 8) + { + result = stbi__convert_16_to_8((stbi__uint16*)result, *x, *y, req_comp == 0 ? *comp : req_comp); + ri.bits_per_channel = 8; + } + + // @TODO: move stbi__convert_format to here + + if (stbi__vertically_flip_on_load) + { + int channels = req_comp ? req_comp : *comp; + stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi_uc)); + } + + return (unsigned char*)result; +} + +static stbi__uint16* stbi__load_and_postprocess_16bit(stbi__context* s, int* x, int* y, int* comp, int req_comp) +{ + stbi__result_info ri; + void* result = stbi__load_main(s, x, y, comp, req_comp, &ri, 16); + + if (result == NULL) + return NULL; + + // it is the responsibility of the loaders to make sure we get either 8 or 16 bit. + STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16); + + if (ri.bits_per_channel != 16) + { + result = stbi__convert_8_to_16((stbi_uc*)result, *x, *y, req_comp == 0 ? *comp : req_comp); + ri.bits_per_channel = 16; + } + + // @TODO: move stbi__convert_format16 to here + // @TODO: special case RGB-to-Y (and RGBA-to-YA) for 8-bit-to-16-bit case to keep more precision + + if (stbi__vertically_flip_on_load) + { + int channels = req_comp ? req_comp : *comp; + stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi__uint16)); + } + + return (stbi__uint16*)result; +} + + #if !defined(STBI_NO_HDR) && !defined(STBI_NO_LINEAR) +static void stbi__float_postprocess(float* result, int* x, int* y, int* comp, int req_comp) +{ + if (stbi__vertically_flip_on_load && result != NULL) + { + int channels = req_comp ? req_comp : *comp; + stbi__vertical_flip(result, *x, *y, channels * sizeof(float)); + } +} + #endif + + #ifndef STBI_NO_STDIO + + #if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) +STBI_EXTERN __declspec( + dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char* str, int cbmb, wchar_t* widestr, int cchwide); +STBI_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t* widestr, int cchwide, + char* str, int cbmb, const char* defchar, int* used_default); + #endif + + #if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) +STBIDEF int stbi_convert_wchar_to_utf8(char* buffer, size_t bufferlen, const wchar_t* input) +{ + return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int)bufferlen, NULL, NULL); +} + #endif + +static FILE* stbi__fopen(const char* filename, const char* mode) +{ + FILE* f; + #if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) + wchar_t wMode[64]; + wchar_t wFilename[1024]; + if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename) / sizeof(*wFilename))) + return 0; + + if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode) / sizeof(*wMode))) + return 0; + + #if defined(_MSC_VER) && _MSC_VER >= 1400 + if (0 != _wfopen_s(&f, wFilename, wMode)) + f = 0; + #else + f = _wfopen(wFilename, wMode); + #endif + + #elif defined(_MSC_VER) && _MSC_VER >= 1400 + if (0 != fopen_s(&f, filename, mode)) + f = 0; + #else + f = fopen(filename, mode); + #endif + return f; +} + +STBIDEF stbi_uc* stbi_load(const char* filename, int* x, int* y, int* comp, int req_comp) +{ + FILE* f = stbi__fopen(filename, "rb"); + unsigned char* result; + if (!f) + return stbi__errpuc("can't fopen", "Unable to open file"); + result = stbi_load_from_file(f, x, y, comp, req_comp); + fclose(f); + return result; +} + +STBIDEF stbi_uc* stbi_load_from_file(FILE* f, int* x, int* y, int* comp, int req_comp) +{ + unsigned char* result; + stbi__context s; + stbi__start_file(&s, f); + result = stbi__load_and_postprocess_8bit(&s, x, y, comp, req_comp); + if (result) + { + // need to 'unget' all the characters in the IO buffer + fseek(f, -(int)(s.img_buffer_end - s.img_buffer), SEEK_CUR); + } + return result; +} + +STBIDEF stbi__uint16* stbi_load_from_file_16(FILE* f, int* x, int* y, int* comp, int req_comp) +{ + stbi__uint16* result; + stbi__context s; + stbi__start_file(&s, f); + result = stbi__load_and_postprocess_16bit(&s, x, y, comp, req_comp); + if (result) + { + // need to 'unget' all the characters in the IO buffer + fseek(f, -(int)(s.img_buffer_end - s.img_buffer), SEEK_CUR); + } + return result; +} + +STBIDEF stbi_us* stbi_load_16(const char* filename, int* x, int* y, int* comp, int req_comp) +{ + FILE* f = stbi__fopen(filename, "rb"); + stbi__uint16* result; + if (!f) + return (stbi_us*)stbi__errpuc("can't fopen", "Unable to open file"); + result = stbi_load_from_file_16(f, x, y, comp, req_comp); + fclose(f); + return result; +} + + #endif //! STBI_NO_STDIO + +STBIDEF stbi_us* stbi_load_16_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels) +{ + stbi__context s; + stbi__start_mem(&s, buffer, len); + return stbi__load_and_postprocess_16bit(&s, x, y, channels_in_file, desired_channels); +} + +STBIDEF stbi_us* stbi_load_16_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); + return stbi__load_and_postprocess_16bit(&s, x, y, channels_in_file, desired_channels); +} + +STBIDEF stbi_uc* stbi_load_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp, int req_comp) +{ + stbi__context s; + stbi__start_mem(&s, buffer, len); + return stbi__load_and_postprocess_8bit(&s, x, y, comp, req_comp); +} + +STBIDEF stbi_uc* stbi_load_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* comp, int req_comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); + return stbi__load_and_postprocess_8bit(&s, x, y, comp, req_comp); +} + + #ifndef STBI_NO_GIF +STBIDEF stbi_uc* stbi_load_gif_from_memory(const stbi_uc* buffer, int len, int** delays, int* x, int* y, int* z, int* comp, int req_comp) +{ + unsigned char* result; + stbi__context s; + stbi__start_mem(&s, buffer, len); + + result = (unsigned char*)stbi__load_gif_main(&s, delays, x, y, z, comp, req_comp); + if (stbi__vertically_flip_on_load) + { + stbi__vertical_flip_slices(result, *x, *y, *z, *comp); + } + + return result; +} + #endif + + #ifndef STBI_NO_LINEAR +static float* stbi__loadf_main(stbi__context* s, int* x, int* y, int* comp, int req_comp) +{ + unsigned char* data; + #ifndef STBI_NO_HDR + if (stbi__hdr_test(s)) + { + stbi__result_info ri; + float* hdr_data = stbi__hdr_load(s, x, y, comp, req_comp, &ri); + if (hdr_data) + stbi__float_postprocess(hdr_data, x, y, comp, req_comp); + return hdr_data; + } + #endif + data = stbi__load_and_postprocess_8bit(s, x, y, comp, req_comp); + if (data) + return stbi__ldr_to_hdr(data, *x, *y, req_comp ? req_comp : *comp); + return stbi__errpf("unknown image type", "Image not of any known type, or corrupt"); +} + +STBIDEF float* stbi_loadf_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp, int req_comp) +{ + stbi__context s; + stbi__start_mem(&s, buffer, len); + return stbi__loadf_main(&s, x, y, comp, req_comp); +} + +STBIDEF float* stbi_loadf_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* comp, int req_comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); + return stbi__loadf_main(&s, x, y, comp, req_comp); +} + + #ifndef STBI_NO_STDIO +STBIDEF float* stbi_loadf(const char* filename, int* x, int* y, int* comp, int req_comp) +{ + float* result; + FILE* f = stbi__fopen(filename, "rb"); + if (!f) + return stbi__errpf("can't fopen", "Unable to open file"); + result = stbi_loadf_from_file(f, x, y, comp, req_comp); + fclose(f); + return result; +} + +STBIDEF float* stbi_loadf_from_file(FILE* f, int* x, int* y, int* comp, int req_comp) +{ + stbi__context s; + stbi__start_file(&s, f); + return stbi__loadf_main(&s, x, y, comp, req_comp); +} + #endif // !STBI_NO_STDIO + + #endif // !STBI_NO_LINEAR + +// these is-hdr-or-not is defined independent of whether STBI_NO_LINEAR is +// defined, for API simplicity; if STBI_NO_LINEAR is defined, it always +// reports false! + +STBIDEF int stbi_is_hdr_from_memory(const stbi_uc* buffer, int len) +{ + #ifndef STBI_NO_HDR + stbi__context s; + stbi__start_mem(&s, buffer, len); + return stbi__hdr_test(&s); + #else + STBI_NOTUSED(buffer); + STBI_NOTUSED(len); + return 0; + #endif +} + + #ifndef STBI_NO_STDIO +STBIDEF int stbi_is_hdr(const char* filename) +{ + FILE* f = stbi__fopen(filename, "rb"); + int result = 0; + if (f) + { + result = stbi_is_hdr_from_file(f); + fclose(f); + } + return result; +} + +STBIDEF int stbi_is_hdr_from_file(FILE* f) +{ + #ifndef STBI_NO_HDR + long pos = ftell(f); + int res; + stbi__context s; + stbi__start_file(&s, f); + res = stbi__hdr_test(&s); + fseek(f, pos, SEEK_SET); + return res; + #else + STBI_NOTUSED(f); + return 0; + #endif +} + #endif // !STBI_NO_STDIO + +STBIDEF int stbi_is_hdr_from_callbacks(const stbi_io_callbacks* clbk, void* user) +{ + #ifndef STBI_NO_HDR + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); + return stbi__hdr_test(&s); + #else + STBI_NOTUSED(clbk); + STBI_NOTUSED(user); + return 0; + #endif +} + + #ifndef STBI_NO_LINEAR +static float stbi__l2h_gamma = 2.2f, stbi__l2h_scale = 1.0f; + +STBIDEF void stbi_ldr_to_hdr_gamma(float gamma) +{ + stbi__l2h_gamma = gamma; +} +STBIDEF void stbi_ldr_to_hdr_scale(float scale) +{ + stbi__l2h_scale = scale; +} + #endif + +static float stbi__h2l_gamma_i = 1.0f / 2.2f, stbi__h2l_scale_i = 1.0f; + +STBIDEF void stbi_hdr_to_ldr_gamma(float gamma) +{ + stbi__h2l_gamma_i = 1 / gamma; +} +STBIDEF void stbi_hdr_to_ldr_scale(float scale) +{ + stbi__h2l_scale_i = 1 / scale; +} + +////////////////////////////////////////////////////////////////////////////// +// +// Common code used by all image loaders +// + +enum { STBI__SCAN_load = 0, STBI__SCAN_type, STBI__SCAN_header }; + +static void stbi__refill_buffer(stbi__context* s) +{ + int n = (s->io.read)(s->io_user_data, (char*)s->buffer_start, s->buflen); + s->callback_already_read += (int)(s->img_buffer - s->img_buffer_original); + if (n == 0) + { + // at end of file, treat same as if from memory, but need to handle case + // where s->img_buffer isn't pointing to safe memory, e.g. 0-byte file + s->read_from_callbacks = 0; + s->img_buffer = s->buffer_start; + s->img_buffer_end = s->buffer_start + 1; + *s->img_buffer = 0; + } + else + { + s->img_buffer = s->buffer_start; + s->img_buffer_end = s->buffer_start + n; + } +} + +stbi_inline static stbi_uc stbi__get8(stbi__context* s) +{ + if (s->img_buffer < s->img_buffer_end) + return *s->img_buffer++; + if (s->read_from_callbacks) + { + stbi__refill_buffer(s); + return *s->img_buffer++; + } + return 0; +} + + #if defined(STBI_NO_JPEG) && defined(STBI_NO_HDR) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM) + // nothing + #else +stbi_inline static int stbi__at_eof(stbi__context* s) +{ + if (s->io.read) + { + if (!(s->io.eof)(s->io_user_data)) + return 0; + // if feof() is true, check if buffer = end + // special case: we've only got the special 0 character at the end + if (s->read_from_callbacks == 0) + return 1; + } + + return s->img_buffer >= s->img_buffer_end; +} + #endif + + #if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && \ + defined(STBI_NO_GIF) && defined(STBI_NO_PIC) + // nothing + #else +static void stbi__skip(stbi__context* s, int n) +{ + if (n == 0) + return; // already there! + if (n < 0) + { + s->img_buffer = s->img_buffer_end; + return; + } + if (s->io.read) + { + int blen = (int)(s->img_buffer_end - s->img_buffer); + if (blen < n) + { + s->img_buffer = s->img_buffer_end; + (s->io.skip)(s->io_user_data, n - blen); + return; + } + } + s->img_buffer += n; +} + #endif + + #if defined(STBI_NO_PNG) && defined(STBI_NO_TGA) && defined(STBI_NO_HDR) && defined(STBI_NO_PNM) + // nothing + #else +static int stbi__getn(stbi__context* s, stbi_uc* buffer, int n) +{ + if (s->io.read) + { + int blen = (int)(s->img_buffer_end - s->img_buffer); + if (blen < n) + { + int res, count; + + memcpy(buffer, s->img_buffer, blen); + + count = (s->io.read)(s->io_user_data, (char*)buffer + blen, n - blen); + res = (count == (n - blen)); + s->img_buffer = s->img_buffer_end; + return res; + } + } + + if (s->img_buffer + n <= s->img_buffer_end) + { + memcpy(buffer, s->img_buffer, n); + s->img_buffer += n; + return 1; + } + else + return 0; +} + #endif + + #if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC) + // nothing + #else +static int stbi__get16be(stbi__context* s) +{ + int z = stbi__get8(s); + return (z << 8) + stbi__get8(s); +} + #endif + + #if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC) + // nothing + #else +static stbi__uint32 stbi__get32be(stbi__context* s) +{ + stbi__uint32 z = stbi__get16be(s); + return (z << 16) + stbi__get16be(s); +} + #endif + + #if defined(STBI_NO_BMP) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) + // nothing + #else +static int stbi__get16le(stbi__context* s) +{ + int z = stbi__get8(s); + return z + (stbi__get8(s) << 8); +} + #endif + + #ifndef STBI_NO_BMP +static stbi__uint32 stbi__get32le(stbi__context* s) +{ + stbi__uint32 z = stbi__get16le(s); + z += (stbi__uint32)stbi__get16le(s) << 16; + return z; +} + #endif + + #define STBI__BYTECAST(x) ((stbi_uc)((x) & 255)) // truncate int to byte without warnings + + #if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && \ + defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM) + // nothing + #else +////////////////////////////////////////////////////////////////////////////// +// +// generic converter from built-in img_n to req_comp +// individual types do this automatically as much as possible (e.g. jpeg +// does all cases internally since it needs to colorspace convert anyway, +// and it never has alpha, so very few cases ). png can automatically +// interleave an alpha=255 channel, but falls back to this for other cases +// +// assume data buffer is malloced, so malloc a new one and free that one +// only failure mode is malloc failing + +static stbi_uc stbi__compute_y(int r, int g, int b) +{ + return (stbi_uc)(((r * 77) + (g * 150) + (29 * b)) >> 8); +} + #endif + + #if defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && \ + defined(STBI_NO_PIC) && defined(STBI_NO_PNM) + // nothing + #else +static unsigned char* stbi__convert_format(unsigned char* data, int img_n, int req_comp, unsigned int x, unsigned int y) +{ + int i, j; + unsigned char* good; + + if (req_comp == img_n) + return data; + STBI_ASSERT(req_comp >= 1 && req_comp <= 4); + + good = (unsigned char*)stbi__malloc_mad3(req_comp, x, y, 0); + if (good == NULL) + { + STBI_FREE(data); + return stbi__errpuc("outofmem", "Out of memory"); + } + + for (j = 0; j < (int)y; ++j) + { + unsigned char* src = data + j * x * img_n; + unsigned char* dest = good + j * x * req_comp; + + #define STBI__COMBO(a, b) ((a) * 8 + (b)) + #define STBI__CASE(a, b) \ + case STBI__COMBO(a, b): \ + for (i = x - 1; i >= 0; --i, src += a, dest += b) + // convert source image with img_n components to one with req_comp components; + // avoid switch per pixel, so use switch per scanline and massive macros + switch (STBI__COMBO(img_n, req_comp)) + { + STBI__CASE(1, 2) + { + dest[0] = src[0]; + dest[1] = 255; + } + break; + STBI__CASE(1, 3) + { + dest[0] = dest[1] = dest[2] = src[0]; + } + break; + STBI__CASE(1, 4) + { + dest[0] = dest[1] = dest[2] = src[0]; + dest[3] = 255; + } + break; + STBI__CASE(2, 1) + { + dest[0] = src[0]; + } + break; + STBI__CASE(2, 3) + { + dest[0] = dest[1] = dest[2] = src[0]; + } + break; + STBI__CASE(2, 4) + { + dest[0] = dest[1] = dest[2] = src[0]; + dest[3] = src[1]; + } + break; + STBI__CASE(3, 4) + { + dest[0] = src[0]; + dest[1] = src[1]; + dest[2] = src[2]; + dest[3] = 255; + } + break; + STBI__CASE(3, 1) + { + dest[0] = stbi__compute_y(src[0], src[1], src[2]); + } + break; + STBI__CASE(3, 2) + { + dest[0] = stbi__compute_y(src[0], src[1], src[2]); + dest[1] = 255; + } + break; + STBI__CASE(4, 1) + { + dest[0] = stbi__compute_y(src[0], src[1], src[2]); + } + break; + STBI__CASE(4, 2) + { + dest[0] = stbi__compute_y(src[0], src[1], src[2]); + dest[1] = src[3]; + } + break; + STBI__CASE(4, 3) + { + dest[0] = src[0]; + dest[1] = src[1]; + dest[2] = src[2]; + } + break; + default: + STBI_ASSERT(0); + STBI_FREE(data); + STBI_FREE(good); + return stbi__errpuc("unsupported", "Unsupported format conversion"); + } + #undef STBI__CASE + } + + STBI_FREE(data); + return good; +} + #endif + + #if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) + // nothing + #else +static stbi__uint16 stbi__compute_y_16(int r, int g, int b) +{ + return (stbi__uint16)(((r * 77) + (g * 150) + (29 * b)) >> 8); +} + #endif + + #if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) + // nothing + #else +static stbi__uint16* stbi__convert_format16(stbi__uint16* data, int img_n, int req_comp, unsigned int x, unsigned int y) +{ + int i, j; + stbi__uint16* good; + + if (req_comp == img_n) + return data; + STBI_ASSERT(req_comp >= 1 && req_comp <= 4); + + good = (stbi__uint16*)stbi__malloc(req_comp * x * y * 2); + if (good == NULL) + { + STBI_FREE(data); + return (stbi__uint16*)stbi__errpuc("outofmem", "Out of memory"); + } + + for (j = 0; j < (int)y; ++j) + { + stbi__uint16* src = data + j * x * img_n; + stbi__uint16* dest = good + j * x * req_comp; + + #define STBI__COMBO(a, b) ((a) * 8 + (b)) + #define STBI__CASE(a, b) \ + case STBI__COMBO(a, b): \ + for (i = x - 1; i >= 0; --i, src += a, dest += b) + // convert source image with img_n components to one with req_comp components; + // avoid switch per pixel, so use switch per scanline and massive macros + switch (STBI__COMBO(img_n, req_comp)) + { + STBI__CASE(1, 2) + { + dest[0] = src[0]; + dest[1] = 0xffff; + } + break; + STBI__CASE(1, 3) + { + dest[0] = dest[1] = dest[2] = src[0]; + } + break; + STBI__CASE(1, 4) + { + dest[0] = dest[1] = dest[2] = src[0]; + dest[3] = 0xffff; + } + break; + STBI__CASE(2, 1) + { + dest[0] = src[0]; + } + break; + STBI__CASE(2, 3) + { + dest[0] = dest[1] = dest[2] = src[0]; + } + break; + STBI__CASE(2, 4) + { + dest[0] = dest[1] = dest[2] = src[0]; + dest[3] = src[1]; + } + break; + STBI__CASE(3, 4) + { + dest[0] = src[0]; + dest[1] = src[1]; + dest[2] = src[2]; + dest[3] = 0xffff; + } + break; + STBI__CASE(3, 1) + { + dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); + } + break; + STBI__CASE(3, 2) + { + dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); + dest[1] = 0xffff; + } + break; + STBI__CASE(4, 1) + { + dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); + } + break; + STBI__CASE(4, 2) + { + dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); + dest[1] = src[3]; + } + break; + STBI__CASE(4, 3) + { + dest[0] = src[0]; + dest[1] = src[1]; + dest[2] = src[2]; + } + break; + default: + STBI_ASSERT(0); + STBI_FREE(data); + STBI_FREE(good); + return (stbi__uint16*)stbi__errpuc("unsupported", "Unsupported format conversion"); + } + #undef STBI__CASE + } + + STBI_FREE(data); + return good; +} + #endif + + #ifndef STBI_NO_LINEAR +static float* stbi__ldr_to_hdr(stbi_uc* data, int x, int y, int comp) +{ + int i, k, n; + float* output; + if (!data) + return NULL; + output = (float*)stbi__malloc_mad4(x, y, comp, sizeof(float), 0); + if (output == NULL) + { + STBI_FREE(data); + return stbi__errpf("outofmem", "Out of memory"); + } + // compute number of non-alpha components + if (comp & 1) + n = comp; + else + n = comp - 1; + for (i = 0; i < x * y; ++i) + { + for (k = 0; k < n; ++k) + { + output[i * comp + k] = (float)(pow(data[i * comp + k] / 255.0f, stbi__l2h_gamma) * stbi__l2h_scale); + } + } + if (n < comp) + { + for (i = 0; i < x * y; ++i) + { + output[i * comp + n] = data[i * comp + n] / 255.0f; + } + } + STBI_FREE(data); + return output; +} + #endif + + #ifndef STBI_NO_HDR + #define stbi__float2int(x) ((int)(x)) +static stbi_uc* stbi__hdr_to_ldr(float* data, int x, int y, int comp) +{ + int i, k, n; + stbi_uc* output; + if (!data) + return NULL; + output = (stbi_uc*)stbi__malloc_mad3(x, y, comp, 0); + if (output == NULL) + { + STBI_FREE(data); + return stbi__errpuc("outofmem", "Out of memory"); + } + // compute number of non-alpha components + if (comp & 1) + n = comp; + else + n = comp - 1; + for (i = 0; i < x * y; ++i) + { + for (k = 0; k < n; ++k) + { + float z = (float)pow(data[i * comp + k] * stbi__h2l_scale_i, stbi__h2l_gamma_i) * 255 + 0.5f; + if (z < 0) + z = 0; + if (z > 255) + z = 255; + output[i * comp + k] = (stbi_uc)stbi__float2int(z); + } + if (k < comp) + { + float z = data[i * comp + k] * 255 + 0.5f; + if (z < 0) + z = 0; + if (z > 255) + z = 255; + output[i * comp + k] = (stbi_uc)stbi__float2int(z); + } + } + STBI_FREE(data); + return output; +} + #endif + +////////////////////////////////////////////////////////////////////////////// +// +// "baseline" JPEG/JFIF decoder +// +// simple implementation +// - doesn't support delayed output of y-dimension +// - simple interface (only one output format: 8-bit interleaved RGB) +// - doesn't try to recover corrupt jpegs +// - doesn't allow partial loading, loading multiple at once +// - still fast on x86 (copying globals into locals doesn't help x86) +// - allocates lots of intermediate memory (full size of all components) +// - non-interleaved case requires this anyway +// - allows good upsampling (see next) +// high-quality +// - upsampled channels are bilinearly interpolated, even across blocks +// - quality integer IDCT derived from IJG's 'slow' +// performance +// - fast huffman; reasonable integer IDCT +// - some SIMD kernels for common paths on targets with SSE2/NEON +// - uses a lot of intermediate memory, could cache poorly + + #ifndef STBI_NO_JPEG + + // huffman decoding acceleration + #define FAST_BITS 9 // larger handles more cases; smaller stomps less cache + +typedef struct { + stbi_uc fast[1 << FAST_BITS]; + // weirdly, repacking this into AoS is a 10% speed loss, instead of a win + stbi__uint16 code[256]; + stbi_uc values[256]; + stbi_uc size[257]; + unsigned int maxcode[18]; + int delta[17]; // old 'firstsymbol' - old 'firstcode' +} stbi__huffman; + +typedef struct { + stbi__context* s; + stbi__huffman huff_dc[4]; + stbi__huffman huff_ac[4]; + stbi__uint16 dequant[4][64]; + stbi__int16 fast_ac[4][1 << FAST_BITS]; + + // sizes for components, interleaved MCUs + int img_h_max, img_v_max; + int img_mcu_x, img_mcu_y; + int img_mcu_w, img_mcu_h; + + // definition of jpeg image component + struct { + int id; + int h, v; + int tq; + int hd, ha; + int dc_pred; + + int x, y, w2, h2; + stbi_uc* data; + void *raw_data, *raw_coeff; + stbi_uc* linebuf; + short* coeff; // progressive only + int coeff_w, coeff_h; // number of 8x8 coefficient blocks + } img_comp[4]; + + stbi__uint32 code_buffer; // jpeg entropy-coded buffer + int code_bits; // number of valid bits + unsigned char marker; // marker seen while filling entropy buffer + int nomore; // flag if we saw a marker so must stop + + int progressive; + int spec_start; + int spec_end; + int succ_high; + int succ_low; + int eob_run; + int jfif; + int app14_color_transform; // Adobe APP14 tag + int rgb; + + int scan_n, order[4]; + int restart_interval, todo; + + // kernels + void (*idct_block_kernel)(stbi_uc* out, int out_stride, short data[64]); + void (*YCbCr_to_RGB_kernel)(stbi_uc* out, const stbi_uc* y, const stbi_uc* pcb, const stbi_uc* pcr, int count, int step); + stbi_uc* (*resample_row_hv_2_kernel)(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs); +} stbi__jpeg; + +static int stbi__build_huffman(stbi__huffman* h, int* count) +{ + int i, j, k = 0; + unsigned int code; + // build size list for each symbol (from JPEG spec) + for (i = 0; i < 16; ++i) + { + for (j = 0; j < count[i]; ++j) + { + h->size[k++] = (stbi_uc)(i + 1); + if (k >= 257) + return stbi__err("bad size list", "Corrupt JPEG"); + } + } + h->size[k] = 0; + + // compute actual symbols (from jpeg spec) + code = 0; + k = 0; + for (j = 1; j <= 16; ++j) + { + // compute delta to add to code to compute symbol id + h->delta[j] = k - code; + if (h->size[k] == j) + { + while (h->size[k] == j) + h->code[k++] = (stbi__uint16)(code++); + if (code - 1 >= (1u << j)) + return stbi__err("bad code lengths", "Corrupt JPEG"); + } + // compute largest code + 1 for this size, preshifted as needed later + h->maxcode[j] = code << (16 - j); + code <<= 1; + } + h->maxcode[j] = 0xffffffff; + + // build non-spec acceleration table; 255 is flag for not-accelerated + memset(h->fast, 255, 1 << FAST_BITS); + for (i = 0; i < k; ++i) + { + int s = h->size[i]; + if (s <= FAST_BITS) + { + int c = h->code[i] << (FAST_BITS - s); + int m = 1 << (FAST_BITS - s); + for (j = 0; j < m; ++j) + { + h->fast[c + j] = (stbi_uc)i; + } + } + } + return 1; +} + +// build a table that decodes both magnitude and value of small ACs in +// one go. +static void stbi__build_fast_ac(stbi__int16* fast_ac, stbi__huffman* h) +{ + int i; + for (i = 0; i < (1 << FAST_BITS); ++i) + { + stbi_uc fast = h->fast[i]; + fast_ac[i] = 0; + if (fast < 255) + { + int rs = h->values[fast]; + int run = (rs >> 4) & 15; + int magbits = rs & 15; + int len = h->size[fast]; + + if (magbits && len + magbits <= FAST_BITS) + { + // magnitude code followed by receive_extend code + int k = ((i << len) & ((1 << FAST_BITS) - 1)) >> (FAST_BITS - magbits); + int m = 1 << (magbits - 1); + if (k < m) + k += (~0U << magbits) + 1; + // if the result is small enough, we can fit it in fast_ac table + if (k >= -128 && k <= 127) + fast_ac[i] = (stbi__int16)((k * 256) + (run * 16) + (len + magbits)); + } + } + } +} + +static void stbi__grow_buffer_unsafe(stbi__jpeg* j) +{ + do + { + unsigned int b = j->nomore ? 0 : stbi__get8(j->s); + if (b == 0xff) + { + int c = stbi__get8(j->s); + while (c == 0xff) + c = stbi__get8(j->s); // consume fill bytes + if (c != 0) + { + j->marker = (unsigned char)c; + j->nomore = 1; + return; + } + } + j->code_buffer |= b << (24 - j->code_bits); + j->code_bits += 8; + } while (j->code_bits <= 24); +} + +// (1 << n) - 1 +static const stbi__uint32 stbi__bmask[17] = {0, 1, 3, 7, 15, 31, 63, 127, 255, 511, 1023, 2047, 4095, 8191, 16383, 32767, 65535}; + +// decode a jpeg huffman value from the bitstream +stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg* j, stbi__huffman* h) +{ + unsigned int temp; + int c, k; + + if (j->code_bits < 16) + stbi__grow_buffer_unsafe(j); + + // look at the top FAST_BITS and determine what symbol ID it is, + // if the code is <= FAST_BITS + c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS) - 1); + k = h->fast[c]; + if (k < 255) + { + int s = h->size[k]; + if (s > j->code_bits) + return -1; + j->code_buffer <<= s; + j->code_bits -= s; + return h->values[k]; + } + + // naive test is to shift the code_buffer down so k bits are + // valid, then test against maxcode. To speed this up, we've + // preshifted maxcode left so that it has (16-k) 0s at the + // end; in other words, regardless of the number of bits, it + // wants to be compared against something shifted to have 16; + // that way we don't need to shift inside the loop. + temp = j->code_buffer >> 16; + for (k = FAST_BITS + 1;; ++k) + if (temp < h->maxcode[k]) + break; + if (k == 17) + { + // error! code not found + j->code_bits -= 16; + return -1; + } + + if (k > j->code_bits) + return -1; + + // convert the huffman code to the symbol id + c = ((j->code_buffer >> (32 - k)) & stbi__bmask[k]) + h->delta[k]; + if (c < 0 || c >= 256) // symbol id out of bounds! + return -1; + STBI_ASSERT((((j->code_buffer) >> (32 - h->size[c])) & stbi__bmask[h->size[c]]) == h->code[c]); + + // convert the id to a symbol + j->code_bits -= k; + j->code_buffer <<= k; + return h->values[c]; +} + +// bias[n] = (-1<code_bits < n) + stbi__grow_buffer_unsafe(j); + if (j->code_bits < n) + return 0; // ran out of bits from stream, return 0s intead of continuing + + sgn = j->code_buffer >> 31; // sign bit always in MSB; 0 if MSB clear (positive), 1 if MSB set (negative) + k = stbi_lrot(j->code_buffer, n); + j->code_buffer = k & ~stbi__bmask[n]; + k &= stbi__bmask[n]; + j->code_bits -= n; + return k + (stbi__jbias[n] & (sgn - 1)); +} + +// get some unsigned bits +stbi_inline static int stbi__jpeg_get_bits(stbi__jpeg* j, int n) +{ + unsigned int k; + if (j->code_bits < n) + stbi__grow_buffer_unsafe(j); + if (j->code_bits < n) + return 0; // ran out of bits from stream, return 0s intead of continuing + k = stbi_lrot(j->code_buffer, n); + j->code_buffer = k & ~stbi__bmask[n]; + k &= stbi__bmask[n]; + j->code_bits -= n; + return k; +} + +stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg* j) +{ + unsigned int k; + if (j->code_bits < 1) + stbi__grow_buffer_unsafe(j); + if (j->code_bits < 1) + return 0; // ran out of bits from stream, return 0s intead of continuing + k = j->code_buffer; + j->code_buffer <<= 1; + --j->code_bits; + return k & 0x80000000; +} + +// given a value that's at position X in the zigzag stream, +// where does it appear in the 8x8 matrix coded as row-major? +static const stbi_uc stbi__jpeg_dezigzag[64 + 15] = {0, 1, 8, 16, 9, 2, 3, 10, 17, 24, 32, 25, 18, 11, 4, 5, 12, 19, 26, 33, 40, 48, 41, 34, 27, 20, + 13, 6, 7, 14, 21, 28, 35, 42, 49, 56, 57, 50, 43, 36, 29, 22, 15, 23, 30, 37, 44, 51, 58, 59, 52, 45, 38, 31, 39, 46, 53, 60, 61, 54, 47, 55, 62, + 63, + // let corrupt input sample past end + 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63}; + +// decode one 64-entry block-- +static int stbi__jpeg_decode_block(stbi__jpeg* j, short data[64], stbi__huffman* hdc, stbi__huffman* hac, stbi__int16* fac, int b, + stbi__uint16* dequant) +{ + int diff, dc, k; + int t; + + if (j->code_bits < 16) + stbi__grow_buffer_unsafe(j); + t = stbi__jpeg_huff_decode(j, hdc); + if (t < 0 || t > 15) + return stbi__err("bad huffman code", "Corrupt JPEG"); + + // 0 all the ac values now so we can do it 32-bits at a time + memset(data, 0, 64 * sizeof(data[0])); + + diff = t ? stbi__extend_receive(j, t) : 0; + if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) + return stbi__err("bad delta", "Corrupt JPEG"); + dc = j->img_comp[b].dc_pred + diff; + j->img_comp[b].dc_pred = dc; + if (!stbi__mul2shorts_valid(dc, dequant[0])) + return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + data[0] = (short)(dc * dequant[0]); + + // decode AC components, see JPEG spec + k = 1; + do + { + unsigned int zig; + int c, r, s; + if (j->code_bits < 16) + stbi__grow_buffer_unsafe(j); + c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS) - 1); + r = fac[c]; + if (r) + { // fast-AC path + k += (r >> 4) & 15; // run + s = r & 15; // combined length + if (s > j->code_bits) + return stbi__err("bad huffman code", "Combined length longer than code bits available"); + j->code_buffer <<= s; + j->code_bits -= s; + // decode into unzigzag'd location + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short)((r >> 8) * dequant[zig]); + } + else + { + int rs = stbi__jpeg_huff_decode(j, hac); + if (rs < 0) + return stbi__err("bad huffman code", "Corrupt JPEG"); + s = rs & 15; + r = rs >> 4; + if (s == 0) + { + if (rs != 0xf0) + break; // end block + k += 16; + } + else + { + k += r; + // decode into unzigzag'd location + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short)(stbi__extend_receive(j, s) * dequant[zig]); + } + } + } while (k < 64); + return 1; +} + +static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg* j, short data[64], stbi__huffman* hdc, int b) +{ + int diff, dc; + int t; + if (j->spec_end != 0) + return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + + if (j->code_bits < 16) + stbi__grow_buffer_unsafe(j); + + if (j->succ_high == 0) + { + // first scan for DC coefficient, must be first + memset(data, 0, 64 * sizeof(data[0])); // 0 all the ac values now + t = stbi__jpeg_huff_decode(j, hdc); + if (t < 0 || t > 15) + return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + diff = t ? stbi__extend_receive(j, t) : 0; + + if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) + return stbi__err("bad delta", "Corrupt JPEG"); + dc = j->img_comp[b].dc_pred + diff; + j->img_comp[b].dc_pred = dc; + if (!stbi__mul2shorts_valid(dc, 1 << j->succ_low)) + return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + data[0] = (short)(dc * (1 << j->succ_low)); + } + else + { + // refinement scan for DC coefficient + if (stbi__jpeg_get_bit(j)) + data[0] += (short)(1 << j->succ_low); + } + return 1; +} + +// @OPTIMIZE: store non-zigzagged during the decode passes, +// and only de-zigzag when dequantizing +static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg* j, short data[64], stbi__huffman* hac, stbi__int16* fac) +{ + int k; + if (j->spec_start == 0) + return stbi__err("can't merge dc and ac", "Corrupt JPEG"); + + if (j->succ_high == 0) + { + int shift = j->succ_low; + + if (j->eob_run) + { + --j->eob_run; + return 1; + } + + k = j->spec_start; + do + { + unsigned int zig; + int c, r, s; + if (j->code_bits < 16) + stbi__grow_buffer_unsafe(j); + c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS) - 1); + r = fac[c]; + if (r) + { // fast-AC path + k += (r >> 4) & 15; // run + s = r & 15; // combined length + if (s > j->code_bits) + return stbi__err("bad huffman code", "Combined length longer than code bits available"); + j->code_buffer <<= s; + j->code_bits -= s; + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short)((r >> 8) * (1 << shift)); + } + else + { + int rs = stbi__jpeg_huff_decode(j, hac); + if (rs < 0) + return stbi__err("bad huffman code", "Corrupt JPEG"); + s = rs & 15; + r = rs >> 4; + if (s == 0) + { + if (r < 15) + { + j->eob_run = (1 << r); + if (r) + j->eob_run += stbi__jpeg_get_bits(j, r); + --j->eob_run; + break; + } + k += 16; + } + else + { + k += r; + zig = stbi__jpeg_dezigzag[k++]; + data[zig] = (short)(stbi__extend_receive(j, s) * (1 << shift)); + } + } + } while (k <= j->spec_end); + } + else + { + // refinement scan for these AC coefficients + + short bit = (short)(1 << j->succ_low); + + if (j->eob_run) + { + --j->eob_run; + for (k = j->spec_start; k <= j->spec_end; ++k) + { + short* p = &data[stbi__jpeg_dezigzag[k]]; + if (*p != 0) + if (stbi__jpeg_get_bit(j)) + if ((*p & bit) == 0) + { + if (*p > 0) + *p += bit; + else + *p -= bit; + } + } + } + else + { + k = j->spec_start; + do + { + int r, s; + int rs = stbi__jpeg_huff_decode(j, hac); // @OPTIMIZE see if we can use the fast path here, advance-by-r is so slow, eh + if (rs < 0) + return stbi__err("bad huffman code", "Corrupt JPEG"); + s = rs & 15; + r = rs >> 4; + if (s == 0) + { + if (r < 15) + { + j->eob_run = (1 << r) - 1; + if (r) + j->eob_run += stbi__jpeg_get_bits(j, r); + r = 64; // force end of block + } + else + { + // r=15 s=0 should write 16 0s, so we just do + // a run of 15 0s and then write s (which is 0), + // so we don't have to do anything special here + } + } + else + { + if (s != 1) + return stbi__err("bad huffman code", "Corrupt JPEG"); + // sign bit + if (stbi__jpeg_get_bit(j)) + s = bit; + else + s = -bit; + } + + // advance by r + while (k <= j->spec_end) + { + short* p = &data[stbi__jpeg_dezigzag[k++]]; + if (*p != 0) + { + if (stbi__jpeg_get_bit(j)) + if ((*p & bit) == 0) + { + if (*p > 0) + *p += bit; + else + *p -= bit; + } + } + else + { + if (r == 0) + { + *p = (short)s; + break; + } + --r; + } + } + } while (k <= j->spec_end); + } + } + return 1; +} + +// take a -128..127 value and stbi__clamp it and convert to 0..255 +stbi_inline static stbi_uc stbi__clamp(int x) +{ + // trick to use a single test to catch both cases + if ((unsigned int)x > 255) + { + if (x < 0) + return 0; + if (x > 255) + return 255; + } + return (stbi_uc)x; +} + + #define stbi__f2f(x) ((int)(((x) * 4096 + 0.5))) + #define stbi__fsh(x) ((x) * 4096) + + // derived from jidctint -- DCT_ISLOW + #define STBI__IDCT_1D(s0, s1, s2, s3, s4, s5, s6, s7) \ + int t0, t1, t2, t3, p1, p2, p3, p4, p5, x0, x1, x2, x3; \ + p2 = s2; \ + p3 = s6; \ + p1 = (p2 + p3) * stbi__f2f(0.5411961f); \ + t2 = p1 + p3 * stbi__f2f(-1.847759065f); \ + t3 = p1 + p2 * stbi__f2f(0.765366865f); \ + p2 = s0; \ + p3 = s4; \ + t0 = stbi__fsh(p2 + p3); \ + t1 = stbi__fsh(p2 - p3); \ + x0 = t0 + t3; \ + x3 = t0 - t3; \ + x1 = t1 + t2; \ + x2 = t1 - t2; \ + t0 = s7; \ + t1 = s5; \ + t2 = s3; \ + t3 = s1; \ + p3 = t0 + t2; \ + p4 = t1 + t3; \ + p1 = t0 + t3; \ + p2 = t1 + t2; \ + p5 = (p3 + p4) * stbi__f2f(1.175875602f); \ + t0 = t0 * stbi__f2f(0.298631336f); \ + t1 = t1 * stbi__f2f(2.053119869f); \ + t2 = t2 * stbi__f2f(3.072711026f); \ + t3 = t3 * stbi__f2f(1.501321110f); \ + p1 = p5 + p1 * stbi__f2f(-0.899976223f); \ + p2 = p5 + p2 * stbi__f2f(-2.562915447f); \ + p3 = p3 * stbi__f2f(-1.961570560f); \ + p4 = p4 * stbi__f2f(-0.390180644f); \ + t3 += p1 + p4; \ + t2 += p2 + p3; \ + t1 += p2 + p4; \ + t0 += p1 + p3; + +static void stbi__idct_block(stbi_uc* out, int out_stride, short data[64]) +{ + int i, val[64], *v = val; + stbi_uc* o; + short* d = data; + + // columns + for (i = 0; i < 8; ++i, ++d, ++v) + { + // if all zeroes, shortcut -- this avoids dequantizing 0s and IDCTing + if (d[8] == 0 && d[16] == 0 && d[24] == 0 && d[32] == 0 && d[40] == 0 && d[48] == 0 && d[56] == 0) + { + // no shortcut 0 seconds + // (1|2|3|4|5|6|7)==0 0 seconds + // all separate -0.047 seconds + // 1 && 2|3 && 4|5 && 6|7: -0.047 seconds + int dcterm = d[0] * 4; + v[0] = v[8] = v[16] = v[24] = v[32] = v[40] = v[48] = v[56] = dcterm; + } + else + { + STBI__IDCT_1D(d[0], d[8], d[16], d[24], d[32], d[40], d[48], d[56]) + // constants scaled things up by 1<<12; let's bring them back + // down, but keep 2 extra bits of precision + x0 += 512; + x1 += 512; + x2 += 512; + x3 += 512; + v[0] = (x0 + t3) >> 10; + v[56] = (x0 - t3) >> 10; + v[8] = (x1 + t2) >> 10; + v[48] = (x1 - t2) >> 10; + v[16] = (x2 + t1) >> 10; + v[40] = (x2 - t1) >> 10; + v[24] = (x3 + t0) >> 10; + v[32] = (x3 - t0) >> 10; + } + } + + for (i = 0, v = val, o = out; i < 8; ++i, v += 8, o += out_stride) + { + // no fast case since the first 1D IDCT spread components out + STBI__IDCT_1D(v[0], v[1], v[2], v[3], v[4], v[5], v[6], v[7]) + // constants scaled things up by 1<<12, plus we had 1<<2 from first + // loop, plus horizontal and vertical each scale by sqrt(8) so together + // we've got an extra 1<<3, so 1<<17 total we need to remove. + // so we want to round that, which means adding 0.5 * 1<<17, + // aka 65536. Also, we'll end up with -128 to 127 that we want + // to encode as 0..255 by adding 128, so we'll add that before the shift + x0 += 65536 + (128 << 17); + x1 += 65536 + (128 << 17); + x2 += 65536 + (128 << 17); + x3 += 65536 + (128 << 17); + // tried computing the shifts into temps, or'ing the temps to see + // if any were out of range, but that was slower + o[0] = stbi__clamp((x0 + t3) >> 17); + o[7] = stbi__clamp((x0 - t3) >> 17); + o[1] = stbi__clamp((x1 + t2) >> 17); + o[6] = stbi__clamp((x1 - t2) >> 17); + o[2] = stbi__clamp((x2 + t1) >> 17); + o[5] = stbi__clamp((x2 - t1) >> 17); + o[3] = stbi__clamp((x3 + t0) >> 17); + o[4] = stbi__clamp((x3 - t0) >> 17); + } +} + + #ifdef STBI_SSE2 +// sse2 integer IDCT. not the fastest possible implementation but it +// produces bit-identical results to the generic C version so it's +// fully "transparent". +static void stbi__idct_simd(stbi_uc* out, int out_stride, short data[64]) +{ + // This is constructed to match our regular (generic) integer IDCT exactly. + __m128i row0, row1, row2, row3, row4, row5, row6, row7; + __m128i tmp; + + // dot product constant: even elems=x, odd elems=y + #define dct_const(x, y) _mm_setr_epi16((x), (y), (x), (y), (x), (y), (x), (y)) + + // out(0) = c0[even]*x + c0[odd]*y (c0, x, y 16-bit, out 32-bit) + // out(1) = c1[even]*x + c1[odd]*y + #define dct_rot(out0, out1, x, y, c0, c1) \ + __m128i c0##lo = _mm_unpacklo_epi16((x), (y)); \ + __m128i c0##hi = _mm_unpackhi_epi16((x), (y)); \ + __m128i out0##_l = _mm_madd_epi16(c0##lo, c0); \ + __m128i out0##_h = _mm_madd_epi16(c0##hi, c0); \ + __m128i out1##_l = _mm_madd_epi16(c0##lo, c1); \ + __m128i out1##_h = _mm_madd_epi16(c0##hi, c1) + + // out = in << 12 (in 16-bit, out 32-bit) + #define dct_widen(out, in) \ + __m128i out##_l = _mm_srai_epi32(_mm_unpacklo_epi16(_mm_setzero_si128(), (in)), 4); \ + __m128i out##_h = _mm_srai_epi32(_mm_unpackhi_epi16(_mm_setzero_si128(), (in)), 4) + + // wide add + #define dct_wadd(out, a, b) \ + __m128i out##_l = _mm_add_epi32(a##_l, b##_l); \ + __m128i out##_h = _mm_add_epi32(a##_h, b##_h) + + // wide sub + #define dct_wsub(out, a, b) \ + __m128i out##_l = _mm_sub_epi32(a##_l, b##_l); \ + __m128i out##_h = _mm_sub_epi32(a##_h, b##_h) + + // butterfly a/b, add bias, then shift by "s" and pack + #define dct_bfly32o(out0, out1, a, b, bias, s) \ + { \ + __m128i abiased_l = _mm_add_epi32(a##_l, bias); \ + __m128i abiased_h = _mm_add_epi32(a##_h, bias); \ + dct_wadd(sum, abiased, b); \ + dct_wsub(dif, abiased, b); \ + out0 = _mm_packs_epi32(_mm_srai_epi32(sum_l, s), _mm_srai_epi32(sum_h, s)); \ + out1 = _mm_packs_epi32(_mm_srai_epi32(dif_l, s), _mm_srai_epi32(dif_h, s)); \ + } + + // 8-bit interleave step (for transposes) + #define dct_interleave8(a, b) \ + tmp = a; \ + a = _mm_unpacklo_epi8(a, b); \ + b = _mm_unpackhi_epi8(tmp, b) + + // 16-bit interleave step (for transposes) + #define dct_interleave16(a, b) \ + tmp = a; \ + a = _mm_unpacklo_epi16(a, b); \ + b = _mm_unpackhi_epi16(tmp, b) + + #define dct_pass(bias, shift) \ + { \ + /* even part */ \ + dct_rot(t2e, t3e, row2, row6, rot0_0, rot0_1); \ + __m128i sum04 = _mm_add_epi16(row0, row4); \ + __m128i dif04 = _mm_sub_epi16(row0, row4); \ + dct_widen(t0e, sum04); \ + dct_widen(t1e, dif04); \ + dct_wadd(x0, t0e, t3e); \ + dct_wsub(x3, t0e, t3e); \ + dct_wadd(x1, t1e, t2e); \ + dct_wsub(x2, t1e, t2e); \ + /* odd part */ \ + dct_rot(y0o, y2o, row7, row3, rot2_0, rot2_1); \ + dct_rot(y1o, y3o, row5, row1, rot3_0, rot3_1); \ + __m128i sum17 = _mm_add_epi16(row1, row7); \ + __m128i sum35 = _mm_add_epi16(row3, row5); \ + dct_rot(y4o, y5o, sum17, sum35, rot1_0, rot1_1); \ + dct_wadd(x4, y0o, y4o); \ + dct_wadd(x5, y1o, y5o); \ + dct_wadd(x6, y2o, y5o); \ + dct_wadd(x7, y3o, y4o); \ + dct_bfly32o(row0, row7, x0, x7, bias, shift); \ + dct_bfly32o(row1, row6, x1, x6, bias, shift); \ + dct_bfly32o(row2, row5, x2, x5, bias, shift); \ + dct_bfly32o(row3, row4, x3, x4, bias, shift); \ + } + + __m128i rot0_0 = dct_const(stbi__f2f(0.5411961f), stbi__f2f(0.5411961f) + stbi__f2f(-1.847759065f)); + __m128i rot0_1 = dct_const(stbi__f2f(0.5411961f) + stbi__f2f(0.765366865f), stbi__f2f(0.5411961f)); + __m128i rot1_0 = dct_const(stbi__f2f(1.175875602f) + stbi__f2f(-0.899976223f), stbi__f2f(1.175875602f)); + __m128i rot1_1 = dct_const(stbi__f2f(1.175875602f), stbi__f2f(1.175875602f) + stbi__f2f(-2.562915447f)); + __m128i rot2_0 = dct_const(stbi__f2f(-1.961570560f) + stbi__f2f(0.298631336f), stbi__f2f(-1.961570560f)); + __m128i rot2_1 = dct_const(stbi__f2f(-1.961570560f), stbi__f2f(-1.961570560f) + stbi__f2f(3.072711026f)); + __m128i rot3_0 = dct_const(stbi__f2f(-0.390180644f) + stbi__f2f(2.053119869f), stbi__f2f(-0.390180644f)); + __m128i rot3_1 = dct_const(stbi__f2f(-0.390180644f), stbi__f2f(-0.390180644f) + stbi__f2f(1.501321110f)); + + // rounding biases in column/row passes, see stbi__idct_block for explanation. + __m128i bias_0 = _mm_set1_epi32(512); + __m128i bias_1 = _mm_set1_epi32(65536 + (128 << 17)); + + // load + row0 = _mm_load_si128((const __m128i*)(data + 0 * 8)); + row1 = _mm_load_si128((const __m128i*)(data + 1 * 8)); + row2 = _mm_load_si128((const __m128i*)(data + 2 * 8)); + row3 = _mm_load_si128((const __m128i*)(data + 3 * 8)); + row4 = _mm_load_si128((const __m128i*)(data + 4 * 8)); + row5 = _mm_load_si128((const __m128i*)(data + 5 * 8)); + row6 = _mm_load_si128((const __m128i*)(data + 6 * 8)); + row7 = _mm_load_si128((const __m128i*)(data + 7 * 8)); + + // column pass + dct_pass(bias_0, 10); + + { + // 16bit 8x8 transpose pass 1 + dct_interleave16(row0, row4); + dct_interleave16(row1, row5); + dct_interleave16(row2, row6); + dct_interleave16(row3, row7); + + // transpose pass 2 + dct_interleave16(row0, row2); + dct_interleave16(row1, row3); + dct_interleave16(row4, row6); + dct_interleave16(row5, row7); + + // transpose pass 3 + dct_interleave16(row0, row1); + dct_interleave16(row2, row3); + dct_interleave16(row4, row5); + dct_interleave16(row6, row7); + } + + // row pass + dct_pass(bias_1, 17); + + { + // pack + __m128i p0 = _mm_packus_epi16(row0, row1); // a0a1a2a3...a7b0b1b2b3...b7 + __m128i p1 = _mm_packus_epi16(row2, row3); + __m128i p2 = _mm_packus_epi16(row4, row5); + __m128i p3 = _mm_packus_epi16(row6, row7); + + // 8bit 8x8 transpose pass 1 + dct_interleave8(p0, p2); // a0e0a1e1... + dct_interleave8(p1, p3); // c0g0c1g1... + + // transpose pass 2 + dct_interleave8(p0, p1); // a0c0e0g0... + dct_interleave8(p2, p3); // b0d0f0h0... + + // transpose pass 3 + dct_interleave8(p0, p2); // a0b0c0d0... + dct_interleave8(p1, p3); // a4b4c4d4... + + // store + _mm_storel_epi64((__m128i*)out, p0); + out += out_stride; + _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p0, 0x4e)); + out += out_stride; + _mm_storel_epi64((__m128i*)out, p2); + out += out_stride; + _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p2, 0x4e)); + out += out_stride; + _mm_storel_epi64((__m128i*)out, p1); + out += out_stride; + _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p1, 0x4e)); + out += out_stride; + _mm_storel_epi64((__m128i*)out, p3); + out += out_stride; + _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p3, 0x4e)); + } + + #undef dct_const + #undef dct_rot + #undef dct_widen + #undef dct_wadd + #undef dct_wsub + #undef dct_bfly32o + #undef dct_interleave8 + #undef dct_interleave16 + #undef dct_pass +} + + #endif // STBI_SSE2 + + #ifdef STBI_NEON + +// NEON integer IDCT. should produce bit-identical +// results to the generic C version. +static void stbi__idct_simd(stbi_uc* out, int out_stride, short data[64]) +{ + int16x8_t row0, row1, row2, row3, row4, row5, row6, row7; + + int16x4_t rot0_0 = vdup_n_s16(stbi__f2f(0.5411961f)); + int16x4_t rot0_1 = vdup_n_s16(stbi__f2f(-1.847759065f)); + int16x4_t rot0_2 = vdup_n_s16(stbi__f2f(0.765366865f)); + int16x4_t rot1_0 = vdup_n_s16(stbi__f2f(1.175875602f)); + int16x4_t rot1_1 = vdup_n_s16(stbi__f2f(-0.899976223f)); + int16x4_t rot1_2 = vdup_n_s16(stbi__f2f(-2.562915447f)); + int16x4_t rot2_0 = vdup_n_s16(stbi__f2f(-1.961570560f)); + int16x4_t rot2_1 = vdup_n_s16(stbi__f2f(-0.390180644f)); + int16x4_t rot3_0 = vdup_n_s16(stbi__f2f(0.298631336f)); + int16x4_t rot3_1 = vdup_n_s16(stbi__f2f(2.053119869f)); + int16x4_t rot3_2 = vdup_n_s16(stbi__f2f(3.072711026f)); + int16x4_t rot3_3 = vdup_n_s16(stbi__f2f(1.501321110f)); + + #define dct_long_mul(out, inq, coeff) \ + int32x4_t out##_l = vmull_s16(vget_low_s16(inq), coeff); \ + int32x4_t out##_h = vmull_s16(vget_high_s16(inq), coeff) + + #define dct_long_mac(out, acc, inq, coeff) \ + int32x4_t out##_l = vmlal_s16(acc##_l, vget_low_s16(inq), coeff); \ + int32x4_t out##_h = vmlal_s16(acc##_h, vget_high_s16(inq), coeff) + + #define dct_widen(out, inq) \ + int32x4_t out##_l = vshll_n_s16(vget_low_s16(inq), 12); \ + int32x4_t out##_h = vshll_n_s16(vget_high_s16(inq), 12) + + // wide add + #define dct_wadd(out, a, b) \ + int32x4_t out##_l = vaddq_s32(a##_l, b##_l); \ + int32x4_t out##_h = vaddq_s32(a##_h, b##_h) + + // wide sub + #define dct_wsub(out, a, b) \ + int32x4_t out##_l = vsubq_s32(a##_l, b##_l); \ + int32x4_t out##_h = vsubq_s32(a##_h, b##_h) + + // butterfly a/b, then shift using "shiftop" by "s" and pack + #define dct_bfly32o(out0, out1, a, b, shiftop, s) \ + { \ + dct_wadd(sum, a, b); \ + dct_wsub(dif, a, b); \ + out0 = vcombine_s16(shiftop(sum_l, s), shiftop(sum_h, s)); \ + out1 = vcombine_s16(shiftop(dif_l, s), shiftop(dif_h, s)); \ + } + + #define dct_pass(shiftop, shift) \ + { \ + /* even part */ \ + int16x8_t sum26 = vaddq_s16(row2, row6); \ + dct_long_mul(p1e, sum26, rot0_0); \ + dct_long_mac(t2e, p1e, row6, rot0_1); \ + dct_long_mac(t3e, p1e, row2, rot0_2); \ + int16x8_t sum04 = vaddq_s16(row0, row4); \ + int16x8_t dif04 = vsubq_s16(row0, row4); \ + dct_widen(t0e, sum04); \ + dct_widen(t1e, dif04); \ + dct_wadd(x0, t0e, t3e); \ + dct_wsub(x3, t0e, t3e); \ + dct_wadd(x1, t1e, t2e); \ + dct_wsub(x2, t1e, t2e); \ + /* odd part */ \ + int16x8_t sum15 = vaddq_s16(row1, row5); \ + int16x8_t sum17 = vaddq_s16(row1, row7); \ + int16x8_t sum35 = vaddq_s16(row3, row5); \ + int16x8_t sum37 = vaddq_s16(row3, row7); \ + int16x8_t sumodd = vaddq_s16(sum17, sum35); \ + dct_long_mul(p5o, sumodd, rot1_0); \ + dct_long_mac(p1o, p5o, sum17, rot1_1); \ + dct_long_mac(p2o, p5o, sum35, rot1_2); \ + dct_long_mul(p3o, sum37, rot2_0); \ + dct_long_mul(p4o, sum15, rot2_1); \ + dct_wadd(sump13o, p1o, p3o); \ + dct_wadd(sump24o, p2o, p4o); \ + dct_wadd(sump23o, p2o, p3o); \ + dct_wadd(sump14o, p1o, p4o); \ + dct_long_mac(x4, sump13o, row7, rot3_0); \ + dct_long_mac(x5, sump24o, row5, rot3_1); \ + dct_long_mac(x6, sump23o, row3, rot3_2); \ + dct_long_mac(x7, sump14o, row1, rot3_3); \ + dct_bfly32o(row0, row7, x0, x7, shiftop, shift); \ + dct_bfly32o(row1, row6, x1, x6, shiftop, shift); \ + dct_bfly32o(row2, row5, x2, x5, shiftop, shift); \ + dct_bfly32o(row3, row4, x3, x4, shiftop, shift); \ + } + + // load + row0 = vld1q_s16(data + 0 * 8); + row1 = vld1q_s16(data + 1 * 8); + row2 = vld1q_s16(data + 2 * 8); + row3 = vld1q_s16(data + 3 * 8); + row4 = vld1q_s16(data + 4 * 8); + row5 = vld1q_s16(data + 5 * 8); + row6 = vld1q_s16(data + 6 * 8); + row7 = vld1q_s16(data + 7 * 8); + + // add DC bias + row0 = vaddq_s16(row0, vsetq_lane_s16(1024, vdupq_n_s16(0), 0)); + + // column pass + dct_pass(vrshrn_n_s32, 10); + + // 16bit 8x8 transpose + { + // these three map to a single VTRN.16, VTRN.32, and VSWP, respectively. + // whether compilers actually get this is another story, sadly. + #define dct_trn16(x, y) \ + { \ + int16x8x2_t t = vtrnq_s16(x, y); \ + x = t.val[0]; \ + y = t.val[1]; \ + } + #define dct_trn32(x, y) \ + { \ + int32x4x2_t t = vtrnq_s32(vreinterpretq_s32_s16(x), vreinterpretq_s32_s16(y)); \ + x = vreinterpretq_s16_s32(t.val[0]); \ + y = vreinterpretq_s16_s32(t.val[1]); \ + } + #define dct_trn64(x, y) \ + { \ + int16x8_t x0 = x; \ + int16x8_t y0 = y; \ + x = vcombine_s16(vget_low_s16(x0), vget_low_s16(y0)); \ + y = vcombine_s16(vget_high_s16(x0), vget_high_s16(y0)); \ + } + + // pass 1 + dct_trn16(row0, row1); // a0b0a2b2a4b4a6b6 + dct_trn16(row2, row3); + dct_trn16(row4, row5); + dct_trn16(row6, row7); + + // pass 2 + dct_trn32(row0, row2); // a0b0c0d0a4b4c4d4 + dct_trn32(row1, row3); + dct_trn32(row4, row6); + dct_trn32(row5, row7); + + // pass 3 + dct_trn64(row0, row4); // a0b0c0d0e0f0g0h0 + dct_trn64(row1, row5); + dct_trn64(row2, row6); + dct_trn64(row3, row7); + + #undef dct_trn16 + #undef dct_trn32 + #undef dct_trn64 + } + + // row pass + // vrshrn_n_s32 only supports shifts up to 16, we need + // 17. so do a non-rounding shift of 16 first then follow + // up with a rounding shift by 1. + dct_pass(vshrn_n_s32, 16); + + { + // pack and round + uint8x8_t p0 = vqrshrun_n_s16(row0, 1); + uint8x8_t p1 = vqrshrun_n_s16(row1, 1); + uint8x8_t p2 = vqrshrun_n_s16(row2, 1); + uint8x8_t p3 = vqrshrun_n_s16(row3, 1); + uint8x8_t p4 = vqrshrun_n_s16(row4, 1); + uint8x8_t p5 = vqrshrun_n_s16(row5, 1); + uint8x8_t p6 = vqrshrun_n_s16(row6, 1); + uint8x8_t p7 = vqrshrun_n_s16(row7, 1); + + // again, these can translate into one instruction, but often don't. + #define dct_trn8_8(x, y) \ + { \ + uint8x8x2_t t = vtrn_u8(x, y); \ + x = t.val[0]; \ + y = t.val[1]; \ + } + #define dct_trn8_16(x, y) \ + { \ + uint16x4x2_t t = vtrn_u16(vreinterpret_u16_u8(x), vreinterpret_u16_u8(y)); \ + x = vreinterpret_u8_u16(t.val[0]); \ + y = vreinterpret_u8_u16(t.val[1]); \ + } + #define dct_trn8_32(x, y) \ + { \ + uint32x2x2_t t = vtrn_u32(vreinterpret_u32_u8(x), vreinterpret_u32_u8(y)); \ + x = vreinterpret_u8_u32(t.val[0]); \ + y = vreinterpret_u8_u32(t.val[1]); \ + } + + // sadly can't use interleaved stores here since we only write + // 8 bytes to each scan line! + + // 8x8 8-bit transpose pass 1 + dct_trn8_8(p0, p1); + dct_trn8_8(p2, p3); + dct_trn8_8(p4, p5); + dct_trn8_8(p6, p7); + + // pass 2 + dct_trn8_16(p0, p2); + dct_trn8_16(p1, p3); + dct_trn8_16(p4, p6); + dct_trn8_16(p5, p7); + + // pass 3 + dct_trn8_32(p0, p4); + dct_trn8_32(p1, p5); + dct_trn8_32(p2, p6); + dct_trn8_32(p3, p7); + + // store + vst1_u8(out, p0); + out += out_stride; + vst1_u8(out, p1); + out += out_stride; + vst1_u8(out, p2); + out += out_stride; + vst1_u8(out, p3); + out += out_stride; + vst1_u8(out, p4); + out += out_stride; + vst1_u8(out, p5); + out += out_stride; + vst1_u8(out, p6); + out += out_stride; + vst1_u8(out, p7); + + #undef dct_trn8_8 + #undef dct_trn8_16 + #undef dct_trn8_32 + } + + #undef dct_long_mul + #undef dct_long_mac + #undef dct_widen + #undef dct_wadd + #undef dct_wsub + #undef dct_bfly32o + #undef dct_pass +} + + #endif // STBI_NEON + + #define STBI__MARKER_none 0xff +// if there's a pending marker from the entropy stream, return that +// otherwise, fetch from the stream and get a marker. if there's no +// marker, return 0xff, which is never a valid marker value +static stbi_uc stbi__get_marker(stbi__jpeg* j) +{ + stbi_uc x; + if (j->marker != STBI__MARKER_none) + { + x = j->marker; + j->marker = STBI__MARKER_none; + return x; + } + x = stbi__get8(j->s); + if (x != 0xff) + return STBI__MARKER_none; + while (x == 0xff) + x = stbi__get8(j->s); // consume repeated 0xff fill bytes + return x; +} + + // in each scan, we'll have scan_n components, and the order + // of the components is specified by order[] + #define STBI__RESTART(x) ((x) >= 0xd0 && (x) <= 0xd7) + +// after a restart interval, stbi__jpeg_reset the entropy decoder and +// the dc prediction +static void stbi__jpeg_reset(stbi__jpeg* j) +{ + j->code_bits = 0; + j->code_buffer = 0; + j->nomore = 0; + j->img_comp[0].dc_pred = j->img_comp[1].dc_pred = j->img_comp[2].dc_pred = j->img_comp[3].dc_pred = 0; + j->marker = STBI__MARKER_none; + j->todo = j->restart_interval ? j->restart_interval : 0x7fffffff; + j->eob_run = 0; + // no more than 1<<31 MCUs if no restart_interal? that's plenty safe, + // since we don't even allow 1<<30 pixels +} + +static int stbi__parse_entropy_coded_data(stbi__jpeg* z) +{ + stbi__jpeg_reset(z); + if (!z->progressive) + { + if (z->scan_n == 1) + { + int i, j; + STBI_SIMD_ALIGN(short, data[64]); + int n = z->order[0]; + // non-interleaved data, we just need to process one block at a time, + // in trivial scanline order + // number of blocks to do just depends on how many actual "pixels" this + // component has, independent of interleaved MCU blocking and such + int w = (z->img_comp[n].x + 7) >> 3; + int h = (z->img_comp[n].y + 7) >> 3; + for (j = 0; j < h; ++j) + { + for (i = 0; i < w; ++i) + { + int ha = z->img_comp[n].ha; + if (!stbi__jpeg_decode_block(z, data, z->huff_dc + z->img_comp[n].hd, z->huff_ac + ha, z->fast_ac[ha], n, + z->dequant[z->img_comp[n].tq])) + return 0; + z->idct_block_kernel(z->img_comp[n].data + z->img_comp[n].w2 * j * 8 + i * 8, z->img_comp[n].w2, data); + // every data block is an MCU, so countdown the restart interval + if (--z->todo <= 0) + { + if (z->code_bits < 24) + stbi__grow_buffer_unsafe(z); + // if it's NOT a restart, then just bail, so we get corrupt data + // rather than no data + if (!STBI__RESTART(z->marker)) + return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } + else + { // interleaved + int i, j, k, x, y; + STBI_SIMD_ALIGN(short, data[64]); + for (j = 0; j < z->img_mcu_y; ++j) + { + for (i = 0; i < z->img_mcu_x; ++i) + { + // scan an interleaved mcu... process scan_n components in order + for (k = 0; k < z->scan_n; ++k) + { + int n = z->order[k]; + // scan out an mcu's worth of this component; that's just determined + // by the basic H and V specified for the component + for (y = 0; y < z->img_comp[n].v; ++y) + { + for (x = 0; x < z->img_comp[n].h; ++x) + { + int x2 = (i * z->img_comp[n].h + x) * 8; + int y2 = (j * z->img_comp[n].v + y) * 8; + int ha = z->img_comp[n].ha; + if (!stbi__jpeg_decode_block(z, data, z->huff_dc + z->img_comp[n].hd, z->huff_ac + ha, z->fast_ac[ha], n, + z->dequant[z->img_comp[n].tq])) + return 0; + z->idct_block_kernel(z->img_comp[n].data + z->img_comp[n].w2 * y2 + x2, z->img_comp[n].w2, data); + } + } + } + // after all interleaved components, that's an interleaved MCU, + // so now count down the restart interval + if (--z->todo <= 0) + { + if (z->code_bits < 24) + stbi__grow_buffer_unsafe(z); + if (!STBI__RESTART(z->marker)) + return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } + } + else + { + if (z->scan_n == 1) + { + int i, j; + int n = z->order[0]; + // non-interleaved data, we just need to process one block at a time, + // in trivial scanline order + // number of blocks to do just depends on how many actual "pixels" this + // component has, independent of interleaved MCU blocking and such + int w = (z->img_comp[n].x + 7) >> 3; + int h = (z->img_comp[n].y + 7) >> 3; + for (j = 0; j < h; ++j) + { + for (i = 0; i < w; ++i) + { + short* data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w); + if (z->spec_start == 0) + { + if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n)) + return 0; + } + else + { + int ha = z->img_comp[n].ha; + if (!stbi__jpeg_decode_block_prog_ac(z, data, &z->huff_ac[ha], z->fast_ac[ha])) + return 0; + } + // every data block is an MCU, so countdown the restart interval + if (--z->todo <= 0) + { + if (z->code_bits < 24) + stbi__grow_buffer_unsafe(z); + if (!STBI__RESTART(z->marker)) + return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } + else + { // interleaved + int i, j, k, x, y; + for (j = 0; j < z->img_mcu_y; ++j) + { + for (i = 0; i < z->img_mcu_x; ++i) + { + // scan an interleaved mcu... process scan_n components in order + for (k = 0; k < z->scan_n; ++k) + { + int n = z->order[k]; + // scan out an mcu's worth of this component; that's just determined + // by the basic H and V specified for the component + for (y = 0; y < z->img_comp[n].v; ++y) + { + for (x = 0; x < z->img_comp[n].h; ++x) + { + int x2 = (i * z->img_comp[n].h + x); + int y2 = (j * z->img_comp[n].v + y); + short* data = z->img_comp[n].coeff + 64 * (x2 + y2 * z->img_comp[n].coeff_w); + if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n)) + return 0; + } + } + } + // after all interleaved components, that's an interleaved MCU, + // so now count down the restart interval + if (--z->todo <= 0) + { + if (z->code_bits < 24) + stbi__grow_buffer_unsafe(z); + if (!STBI__RESTART(z->marker)) + return 1; + stbi__jpeg_reset(z); + } + } + } + return 1; + } + } +} + +static void stbi__jpeg_dequantize(short* data, stbi__uint16* dequant) +{ + int i; + for (i = 0; i < 64; ++i) + data[i] *= dequant[i]; +} + +static void stbi__jpeg_finish(stbi__jpeg* z) +{ + if (z->progressive) + { + // dequantize and idct the data + int i, j, n; + for (n = 0; n < z->s->img_n; ++n) + { + int w = (z->img_comp[n].x + 7) >> 3; + int h = (z->img_comp[n].y + 7) >> 3; + for (j = 0; j < h; ++j) + { + for (i = 0; i < w; ++i) + { + short* data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w); + stbi__jpeg_dequantize(data, z->dequant[z->img_comp[n].tq]); + z->idct_block_kernel(z->img_comp[n].data + z->img_comp[n].w2 * j * 8 + i * 8, z->img_comp[n].w2, data); + } + } + } + } +} + +static int stbi__process_marker(stbi__jpeg* z, int m) +{ + int L; + switch (m) + { + case STBI__MARKER_none: // no marker found + return stbi__err("expected marker", "Corrupt JPEG"); + + case 0xDD: // DRI - specify restart interval + if (stbi__get16be(z->s) != 4) + return stbi__err("bad DRI len", "Corrupt JPEG"); + z->restart_interval = stbi__get16be(z->s); + return 1; + + case 0xDB: // DQT - define quantization table + L = stbi__get16be(z->s) - 2; + while (L > 0) + { + int q = stbi__get8(z->s); + int p = q >> 4, sixteen = (p != 0); + int t = q & 15, i; + if (p != 0 && p != 1) + return stbi__err("bad DQT type", "Corrupt JPEG"); + if (t > 3) + return stbi__err("bad DQT table", "Corrupt JPEG"); + + for (i = 0; i < 64; ++i) + z->dequant[t][stbi__jpeg_dezigzag[i]] = (stbi__uint16)(sixteen ? stbi__get16be(z->s) : stbi__get8(z->s)); + L -= (sixteen ? 129 : 65); + } + return L == 0; + + case 0xC4: // DHT - define huffman table + L = stbi__get16be(z->s) - 2; + while (L > 0) + { + stbi_uc* v; + int sizes[16], i, n = 0; + int q = stbi__get8(z->s); + int tc = q >> 4; + int th = q & 15; + if (tc > 1 || th > 3) + return stbi__err("bad DHT header", "Corrupt JPEG"); + for (i = 0; i < 16; ++i) + { + sizes[i] = stbi__get8(z->s); + n += sizes[i]; + } + if (n > 256) + return stbi__err("bad DHT header", "Corrupt JPEG"); // Loop over i < n would write past end of values! + L -= 17; + if (tc == 0) + { + if (!stbi__build_huffman(z->huff_dc + th, sizes)) + return 0; + v = z->huff_dc[th].values; + } + else + { + if (!stbi__build_huffman(z->huff_ac + th, sizes)) + return 0; + v = z->huff_ac[th].values; + } + for (i = 0; i < n; ++i) + v[i] = stbi__get8(z->s); + if (tc != 0) + stbi__build_fast_ac(z->fast_ac[th], z->huff_ac + th); + L -= n; + } + return L == 0; + } + + // check for comment block or APP blocks + if ((m >= 0xE0 && m <= 0xEF) || m == 0xFE) + { + L = stbi__get16be(z->s); + if (L < 2) + { + if (m == 0xFE) + return stbi__err("bad COM len", "Corrupt JPEG"); + else + return stbi__err("bad APP len", "Corrupt JPEG"); + } + L -= 2; + + if (m == 0xE0 && L >= 5) + { // JFIF APP0 segment + static const unsigned char tag[5] = {'J', 'F', 'I', 'F', '\0'}; + int ok = 1; + int i; + for (i = 0; i < 5; ++i) + if (stbi__get8(z->s) != tag[i]) + ok = 0; + L -= 5; + if (ok) + z->jfif = 1; + } + else if (m == 0xEE && L >= 12) + { // Adobe APP14 segment + static const unsigned char tag[6] = {'A', 'd', 'o', 'b', 'e', '\0'}; + int ok = 1; + int i; + for (i = 0; i < 6; ++i) + if (stbi__get8(z->s) != tag[i]) + ok = 0; + L -= 6; + if (ok) + { + stbi__get8(z->s); // version + stbi__get16be(z->s); // flags0 + stbi__get16be(z->s); // flags1 + z->app14_color_transform = stbi__get8(z->s); // color transform + L -= 6; + } + } + + stbi__skip(z->s, L); + return 1; + } + + return stbi__err("unknown marker", "Corrupt JPEG"); +} + +// after we see SOS +static int stbi__process_scan_header(stbi__jpeg* z) +{ + int i; + int Ls = stbi__get16be(z->s); + z->scan_n = stbi__get8(z->s); + if (z->scan_n < 1 || z->scan_n > 4 || z->scan_n > (int)z->s->img_n) + return stbi__err("bad SOS component count", "Corrupt JPEG"); + if (Ls != 6 + 2 * z->scan_n) + return stbi__err("bad SOS len", "Corrupt JPEG"); + for (i = 0; i < z->scan_n; ++i) + { + int id = stbi__get8(z->s), which; + int q = stbi__get8(z->s); + for (which = 0; which < z->s->img_n; ++which) + if (z->img_comp[which].id == id) + break; + if (which == z->s->img_n) + return 0; // no match + z->img_comp[which].hd = q >> 4; + if (z->img_comp[which].hd > 3) + return stbi__err("bad DC huff", "Corrupt JPEG"); + z->img_comp[which].ha = q & 15; + if (z->img_comp[which].ha > 3) + return stbi__err("bad AC huff", "Corrupt JPEG"); + z->order[i] = which; + } + + { + int aa; + z->spec_start = stbi__get8(z->s); + z->spec_end = stbi__get8(z->s); // should be 63, but might be 0 + aa = stbi__get8(z->s); + z->succ_high = (aa >> 4); + z->succ_low = (aa & 15); + if (z->progressive) + { + if (z->spec_start > 63 || z->spec_end > 63 || z->spec_start > z->spec_end || z->succ_high > 13 || z->succ_low > 13) + return stbi__err("bad SOS", "Corrupt JPEG"); + } + else + { + if (z->spec_start != 0) + return stbi__err("bad SOS", "Corrupt JPEG"); + if (z->succ_high != 0 || z->succ_low != 0) + return stbi__err("bad SOS", "Corrupt JPEG"); + z->spec_end = 63; + } + } + + return 1; +} + +static int stbi__free_jpeg_components(stbi__jpeg* z, int ncomp, int why) +{ + int i; + for (i = 0; i < ncomp; ++i) + { + if (z->img_comp[i].raw_data) + { + STBI_FREE(z->img_comp[i].raw_data); + z->img_comp[i].raw_data = NULL; + z->img_comp[i].data = NULL; + } + if (z->img_comp[i].raw_coeff) + { + STBI_FREE(z->img_comp[i].raw_coeff); + z->img_comp[i].raw_coeff = 0; + z->img_comp[i].coeff = 0; + } + if (z->img_comp[i].linebuf) + { + STBI_FREE(z->img_comp[i].linebuf); + z->img_comp[i].linebuf = NULL; + } + } + return why; +} + +static int stbi__process_frame_header(stbi__jpeg* z, int scan) +{ + stbi__context* s = z->s; + int Lf, p, i, q, h_max = 1, v_max = 1, c; + Lf = stbi__get16be(s); + if (Lf < 11) + return stbi__err("bad SOF len", "Corrupt JPEG"); // JPEG + p = stbi__get8(s); + if (p != 8) + return stbi__err("only 8-bit", "JPEG format not supported: 8-bit only"); // JPEG baseline + s->img_y = stbi__get16be(s); + if (s->img_y == 0) + return stbi__err("no header height", "JPEG format not supported: delayed height"); // Legal, but we don't handle it--but neither does IJG + s->img_x = stbi__get16be(s); + if (s->img_x == 0) + return stbi__err("0 width", "Corrupt JPEG"); // JPEG requires + if (s->img_y > STBI_MAX_DIMENSIONS) + return stbi__err("too large", "Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) + return stbi__err("too large", "Very large image (corrupt?)"); + c = stbi__get8(s); + if (c != 3 && c != 1 && c != 4) + return stbi__err("bad component count", "Corrupt JPEG"); + s->img_n = c; + for (i = 0; i < c; ++i) + { + z->img_comp[i].data = NULL; + z->img_comp[i].linebuf = NULL; + } + + if (Lf != 8 + 3 * s->img_n) + return stbi__err("bad SOF len", "Corrupt JPEG"); + + z->rgb = 0; + for (i = 0; i < s->img_n; ++i) + { + static const unsigned char rgb[3] = {'R', 'G', 'B'}; + z->img_comp[i].id = stbi__get8(s); + if (s->img_n == 3 && z->img_comp[i].id == rgb[i]) + ++z->rgb; + q = stbi__get8(s); + z->img_comp[i].h = (q >> 4); + if (!z->img_comp[i].h || z->img_comp[i].h > 4) + return stbi__err("bad H", "Corrupt JPEG"); + z->img_comp[i].v = q & 15; + if (!z->img_comp[i].v || z->img_comp[i].v > 4) + return stbi__err("bad V", "Corrupt JPEG"); + z->img_comp[i].tq = stbi__get8(s); + if (z->img_comp[i].tq > 3) + return stbi__err("bad TQ", "Corrupt JPEG"); + } + + if (scan != STBI__SCAN_load) + return 1; + + if (!stbi__mad3sizes_valid(s->img_x, s->img_y, s->img_n, 0)) + return stbi__err("too large", "Image too large to decode"); + + for (i = 0; i < s->img_n; ++i) + { + if (z->img_comp[i].h > h_max) + h_max = z->img_comp[i].h; + if (z->img_comp[i].v > v_max) + v_max = z->img_comp[i].v; + } + + // check that plane subsampling factors are integer ratios; our resamplers can't deal with fractional ratios + // and I've never seen a non-corrupted JPEG file actually use them + for (i = 0; i < s->img_n; ++i) + { + if (h_max % z->img_comp[i].h != 0) + return stbi__err("bad H", "Corrupt JPEG"); + if (v_max % z->img_comp[i].v != 0) + return stbi__err("bad V", "Corrupt JPEG"); + } + + // compute interleaved mcu info + z->img_h_max = h_max; + z->img_v_max = v_max; + z->img_mcu_w = h_max * 8; + z->img_mcu_h = v_max * 8; + // these sizes can't be more than 17 bits + z->img_mcu_x = (s->img_x + z->img_mcu_w - 1) / z->img_mcu_w; + z->img_mcu_y = (s->img_y + z->img_mcu_h - 1) / z->img_mcu_h; + + for (i = 0; i < s->img_n; ++i) + { + // number of effective pixels (e.g. for non-interleaved MCU) + z->img_comp[i].x = (s->img_x * z->img_comp[i].h + h_max - 1) / h_max; + z->img_comp[i].y = (s->img_y * z->img_comp[i].v + v_max - 1) / v_max; + // to simplify generation, we'll allocate enough memory to decode + // the bogus oversized data from using interleaved MCUs and their + // big blocks (e.g. a 16x16 iMCU on an image of width 33); we won't + // discard the extra data until colorspace conversion + // + // img_mcu_x, img_mcu_y: <=17 bits; comp[i].h and .v are <=4 (checked earlier) + // so these muls can't overflow with 32-bit ints (which we require) + z->img_comp[i].w2 = z->img_mcu_x * z->img_comp[i].h * 8; + z->img_comp[i].h2 = z->img_mcu_y * z->img_comp[i].v * 8; + z->img_comp[i].coeff = 0; + z->img_comp[i].raw_coeff = 0; + z->img_comp[i].linebuf = NULL; + z->img_comp[i].raw_data = stbi__malloc_mad2(z->img_comp[i].w2, z->img_comp[i].h2, 15); + if (z->img_comp[i].raw_data == NULL) + return stbi__free_jpeg_components(z, i + 1, stbi__err("outofmem", "Out of memory")); + // align blocks for idct using mmx/sse + z->img_comp[i].data = (stbi_uc*)(((size_t)z->img_comp[i].raw_data + 15) & ~15); + if (z->progressive) + { + // w2, h2 are multiples of 8 (see above) + z->img_comp[i].coeff_w = z->img_comp[i].w2 / 8; + z->img_comp[i].coeff_h = z->img_comp[i].h2 / 8; + z->img_comp[i].raw_coeff = stbi__malloc_mad3(z->img_comp[i].w2, z->img_comp[i].h2, sizeof(short), 15); + if (z->img_comp[i].raw_coeff == NULL) + return stbi__free_jpeg_components(z, i + 1, stbi__err("outofmem", "Out of memory")); + z->img_comp[i].coeff = (short*)(((size_t)z->img_comp[i].raw_coeff + 15) & ~15); + } + } + + return 1; +} + + // use comparisons since in some cases we handle more than one case (e.g. SOF) + #define stbi__DNL(x) ((x) == 0xdc) + #define stbi__SOI(x) ((x) == 0xd8) + #define stbi__EOI(x) ((x) == 0xd9) + #define stbi__SOF(x) ((x) == 0xc0 || (x) == 0xc1 || (x) == 0xc2) + #define stbi__SOS(x) ((x) == 0xda) + + #define stbi__SOF_progressive(x) ((x) == 0xc2) + +static int stbi__decode_jpeg_header(stbi__jpeg* z, int scan) +{ + int m; + z->jfif = 0; + z->app14_color_transform = -1; // valid values are 0,1,2 + z->marker = STBI__MARKER_none; // initialize cached marker to empty + m = stbi__get_marker(z); + if (!stbi__SOI(m)) + return stbi__err("no SOI", "Corrupt JPEG"); + if (scan == STBI__SCAN_type) + return 1; + m = stbi__get_marker(z); + while (!stbi__SOF(m)) + { + if (!stbi__process_marker(z, m)) + return 0; + m = stbi__get_marker(z); + while (m == STBI__MARKER_none) + { + // some files have extra padding after their blocks, so ok, we'll scan + if (stbi__at_eof(z->s)) + return stbi__err("no SOF", "Corrupt JPEG"); + m = stbi__get_marker(z); + } + } + z->progressive = stbi__SOF_progressive(m); + if (!stbi__process_frame_header(z, scan)) + return 0; + return 1; +} + +static stbi_uc stbi__skip_jpeg_junk_at_end(stbi__jpeg* j) +{ + // some JPEGs have junk at end, skip over it but if we find what looks + // like a valid marker, resume there + while (!stbi__at_eof(j->s)) + { + stbi_uc x = stbi__get8(j->s); + while (x == 0xff) + { // might be a marker + if (stbi__at_eof(j->s)) + return STBI__MARKER_none; + x = stbi__get8(j->s); + if (x != 0x00 && x != 0xff) + { + // not a stuffed zero or lead-in to another marker, looks + // like an actual marker, return it + return x; + } + // stuffed zero has x=0 now which ends the loop, meaning we go + // back to regular scan loop. + // repeated 0xff keeps trying to read the next byte of the marker. + } + } + return STBI__MARKER_none; +} + +// decode image to YCbCr format +static int stbi__decode_jpeg_image(stbi__jpeg* j) +{ + int m; + for (m = 0; m < 4; m++) + { + j->img_comp[m].raw_data = NULL; + j->img_comp[m].raw_coeff = NULL; + } + j->restart_interval = 0; + if (!stbi__decode_jpeg_header(j, STBI__SCAN_load)) + return 0; + m = stbi__get_marker(j); + while (!stbi__EOI(m)) + { + if (stbi__SOS(m)) + { + if (!stbi__process_scan_header(j)) + return 0; + if (!stbi__parse_entropy_coded_data(j)) + return 0; + if (j->marker == STBI__MARKER_none) + { + j->marker = stbi__skip_jpeg_junk_at_end(j); + // if we reach eof without hitting a marker, stbi__get_marker() below will fail and we'll eventually return 0 + } + m = stbi__get_marker(j); + if (STBI__RESTART(m)) + m = stbi__get_marker(j); + } + else if (stbi__DNL(m)) + { + int Ld = stbi__get16be(j->s); + stbi__uint32 NL = stbi__get16be(j->s); + if (Ld != 4) + return stbi__err("bad DNL len", "Corrupt JPEG"); + if (NL != j->s->img_y) + return stbi__err("bad DNL height", "Corrupt JPEG"); + m = stbi__get_marker(j); + } + else + { + if (!stbi__process_marker(j, m)) + return 1; + m = stbi__get_marker(j); + } + } + if (j->progressive) + stbi__jpeg_finish(j); + return 1; +} + +// static jfif-centered resampling (across block boundaries) + +typedef stbi_uc* (*resample_row_func)(stbi_uc* out, stbi_uc* in0, stbi_uc* in1, int w, int hs); + + #define stbi__div4(x) ((stbi_uc)((x) >> 2)) + +static stbi_uc* resample_row_1(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) +{ + STBI_NOTUSED(out); + STBI_NOTUSED(in_far); + STBI_NOTUSED(w); + STBI_NOTUSED(hs); + return in_near; +} + +static stbi_uc* stbi__resample_row_v_2(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) +{ + // need to generate two samples vertically for every one in input + int i; + STBI_NOTUSED(hs); + for (i = 0; i < w; ++i) + out[i] = stbi__div4(3 * in_near[i] + in_far[i] + 2); + return out; +} + +static stbi_uc* stbi__resample_row_h_2(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) +{ + // need to generate two samples horizontally for every one in input + int i; + stbi_uc* input = in_near; + + if (w == 1) + { + // if only one sample, can't do any interpolation + out[0] = out[1] = input[0]; + return out; + } + + out[0] = input[0]; + out[1] = stbi__div4(input[0] * 3 + input[1] + 2); + for (i = 1; i < w - 1; ++i) + { + int n = 3 * input[i] + 2; + out[i * 2 + 0] = stbi__div4(n + input[i - 1]); + out[i * 2 + 1] = stbi__div4(n + input[i + 1]); + } + out[i * 2 + 0] = stbi__div4(input[w - 2] * 3 + input[w - 1] + 2); + out[i * 2 + 1] = input[w - 1]; + + STBI_NOTUSED(in_far); + STBI_NOTUSED(hs); + + return out; +} + + #define stbi__div16(x) ((stbi_uc)((x) >> 4)) + +static stbi_uc* stbi__resample_row_hv_2(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) +{ + // need to generate 2x2 samples for every one in input + int i, t0, t1; + if (w == 1) + { + out[0] = out[1] = stbi__div4(3 * in_near[0] + in_far[0] + 2); + return out; + } + + t1 = 3 * in_near[0] + in_far[0]; + out[0] = stbi__div4(t1 + 2); + for (i = 1; i < w; ++i) + { + t0 = t1; + t1 = 3 * in_near[i] + in_far[i]; + out[i * 2 - 1] = stbi__div16(3 * t0 + t1 + 8); + out[i * 2] = stbi__div16(3 * t1 + t0 + 8); + } + out[w * 2 - 1] = stbi__div4(t1 + 2); + + STBI_NOTUSED(hs); + + return out; +} + + #if defined(STBI_SSE2) || defined(STBI_NEON) +static stbi_uc* stbi__resample_row_hv_2_simd(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) +{ + // need to generate 2x2 samples for every one in input + int i = 0, t0, t1; + + if (w == 1) + { + out[0] = out[1] = stbi__div4(3 * in_near[0] + in_far[0] + 2); + return out; + } + + t1 = 3 * in_near[0] + in_far[0]; + // process groups of 8 pixels for as long as we can. + // note we can't handle the last pixel in a row in this loop + // because we need to handle the filter boundary conditions. + for (; i < ((w - 1) & ~7); i += 8) + { + #if defined(STBI_SSE2) + // load and perform the vertical filtering pass + // this uses 3*x + y = 4*x + (y - x) + __m128i zero = _mm_setzero_si128(); + __m128i farb = _mm_loadl_epi64((__m128i*)(in_far + i)); + __m128i nearb = _mm_loadl_epi64((__m128i*)(in_near + i)); + __m128i farw = _mm_unpacklo_epi8(farb, zero); + __m128i nearw = _mm_unpacklo_epi8(nearb, zero); + __m128i diff = _mm_sub_epi16(farw, nearw); + __m128i nears = _mm_slli_epi16(nearw, 2); + __m128i curr = _mm_add_epi16(nears, diff); // current row + + // horizontal filter works the same based on shifted vers of current + // row. "prev" is current row shifted right by 1 pixel; we need to + // insert the previous pixel value (from t1). + // "next" is current row shifted left by 1 pixel, with first pixel + // of next block of 8 pixels added in. + __m128i prv0 = _mm_slli_si128(curr, 2); + __m128i nxt0 = _mm_srli_si128(curr, 2); + __m128i prev = _mm_insert_epi16(prv0, t1, 0); + __m128i next = _mm_insert_epi16(nxt0, 3 * in_near[i + 8] + in_far[i + 8], 7); + + // horizontal filter, polyphase implementation since it's convenient: + // even pixels = 3*cur + prev = cur*4 + (prev - cur) + // odd pixels = 3*cur + next = cur*4 + (next - cur) + // note the shared term. + __m128i bias = _mm_set1_epi16(8); + __m128i curs = _mm_slli_epi16(curr, 2); + __m128i prvd = _mm_sub_epi16(prev, curr); + __m128i nxtd = _mm_sub_epi16(next, curr); + __m128i curb = _mm_add_epi16(curs, bias); + __m128i even = _mm_add_epi16(prvd, curb); + __m128i odd = _mm_add_epi16(nxtd, curb); + + // interleave even and odd pixels, then undo scaling. + __m128i int0 = _mm_unpacklo_epi16(even, odd); + __m128i int1 = _mm_unpackhi_epi16(even, odd); + __m128i de0 = _mm_srli_epi16(int0, 4); + __m128i de1 = _mm_srli_epi16(int1, 4); + + // pack and write output + __m128i outv = _mm_packus_epi16(de0, de1); + _mm_storeu_si128((__m128i*)(out + i * 2), outv); + #elif defined(STBI_NEON) + // load and perform the vertical filtering pass + // this uses 3*x + y = 4*x + (y - x) + uint8x8_t farb = vld1_u8(in_far + i); + uint8x8_t nearb = vld1_u8(in_near + i); + int16x8_t diff = vreinterpretq_s16_u16(vsubl_u8(farb, nearb)); + int16x8_t nears = vreinterpretq_s16_u16(vshll_n_u8(nearb, 2)); + int16x8_t curr = vaddq_s16(nears, diff); // current row + + // horizontal filter works the same based on shifted vers of current + // row. "prev" is current row shifted right by 1 pixel; we need to + // insert the previous pixel value (from t1). + // "next" is current row shifted left by 1 pixel, with first pixel + // of next block of 8 pixels added in. + int16x8_t prv0 = vextq_s16(curr, curr, 7); + int16x8_t nxt0 = vextq_s16(curr, curr, 1); + int16x8_t prev = vsetq_lane_s16(t1, prv0, 0); + int16x8_t next = vsetq_lane_s16(3 * in_near[i + 8] + in_far[i + 8], nxt0, 7); + + // horizontal filter, polyphase implementation since it's convenient: + // even pixels = 3*cur + prev = cur*4 + (prev - cur) + // odd pixels = 3*cur + next = cur*4 + (next - cur) + // note the shared term. + int16x8_t curs = vshlq_n_s16(curr, 2); + int16x8_t prvd = vsubq_s16(prev, curr); + int16x8_t nxtd = vsubq_s16(next, curr); + int16x8_t even = vaddq_s16(curs, prvd); + int16x8_t odd = vaddq_s16(curs, nxtd); + + // undo scaling and round, then store with even/odd phases interleaved + uint8x8x2_t o; + o.val[0] = vqrshrun_n_s16(even, 4); + o.val[1] = vqrshrun_n_s16(odd, 4); + vst2_u8(out + i * 2, o); + #endif + + // "previous" value for next iter + t1 = 3 * in_near[i + 7] + in_far[i + 7]; + } + + t0 = t1; + t1 = 3 * in_near[i] + in_far[i]; + out[i * 2] = stbi__div16(3 * t1 + t0 + 8); + + for (++i; i < w; ++i) + { + t0 = t1; + t1 = 3 * in_near[i] + in_far[i]; + out[i * 2 - 1] = stbi__div16(3 * t0 + t1 + 8); + out[i * 2] = stbi__div16(3 * t1 + t0 + 8); + } + out[w * 2 - 1] = stbi__div4(t1 + 2); + + STBI_NOTUSED(hs); + + return out; +} + #endif + +static stbi_uc* stbi__resample_row_generic(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) +{ + // resample with nearest-neighbor + int i, j; + STBI_NOTUSED(in_far); + for (i = 0; i < w; ++i) + for (j = 0; j < hs; ++j) + out[i * hs + j] = in_near[i]; + return out; +} + + // this is a reduced-precision calculation of YCbCr-to-RGB introduced + // to make sure the code produces the same results in both SIMD and scalar + #define stbi__float2fixed(x) (((int)((x) * 4096.0f + 0.5f)) << 8) +static void stbi__YCbCr_to_RGB_row(stbi_uc* out, const stbi_uc* y, const stbi_uc* pcb, const stbi_uc* pcr, int count, int step) +{ + int i; + for (i = 0; i < count; ++i) + { + int y_fixed = (y[i] << 20) + (1 << 19); // rounding + int r, g, b; + int cr = pcr[i] - 128; + int cb = pcb[i] - 128; + r = y_fixed + cr * stbi__float2fixed(1.40200f); + g = y_fixed + (cr * -stbi__float2fixed(0.71414f)) + ((cb * -stbi__float2fixed(0.34414f)) & 0xffff0000); + b = y_fixed + cb * stbi__float2fixed(1.77200f); + r >>= 20; + g >>= 20; + b >>= 20; + if ((unsigned)r > 255) + { + if (r < 0) + r = 0; + else + r = 255; + } + if ((unsigned)g > 255) + { + if (g < 0) + g = 0; + else + g = 255; + } + if ((unsigned)b > 255) + { + if (b < 0) + b = 0; + else + b = 255; + } + out[0] = (stbi_uc)r; + out[1] = (stbi_uc)g; + out[2] = (stbi_uc)b; + out[3] = 255; + out += step; + } +} + + #if defined(STBI_SSE2) || defined(STBI_NEON) +static void stbi__YCbCr_to_RGB_simd(stbi_uc* out, const stbi_uc* y, const stbi_uc* pcb, const stbi_uc* pcr, int count, int step) +{ + int i = 0; + + #ifdef STBI_SSE2 + // step == 3 is pretty ugly on the final interleave, and i'm not convinced + // it's useful in practice (you wouldn't use it for textures, for example). + // so just accelerate step == 4 case. + if (step == 4) + { + // this is a fairly straightforward implementation and not super-optimized. + __m128i signflip = _mm_set1_epi8(-0x80); + __m128i cr_const0 = _mm_set1_epi16((short)(1.40200f * 4096.0f + 0.5f)); + __m128i cr_const1 = _mm_set1_epi16(-(short)(0.71414f * 4096.0f + 0.5f)); + __m128i cb_const0 = _mm_set1_epi16(-(short)(0.34414f * 4096.0f + 0.5f)); + __m128i cb_const1 = _mm_set1_epi16((short)(1.77200f * 4096.0f + 0.5f)); + __m128i y_bias = _mm_set1_epi8((char)(unsigned char)128); + __m128i xw = _mm_set1_epi16(255); // alpha channel + + for (; i + 7 < count; i += 8) + { + // load + __m128i y_bytes = _mm_loadl_epi64((__m128i*)(y + i)); + __m128i cr_bytes = _mm_loadl_epi64((__m128i*)(pcr + i)); + __m128i cb_bytes = _mm_loadl_epi64((__m128i*)(pcb + i)); + __m128i cr_biased = _mm_xor_si128(cr_bytes, signflip); // -128 + __m128i cb_biased = _mm_xor_si128(cb_bytes, signflip); // -128 + + // unpack to short (and left-shift cr, cb by 8) + __m128i yw = _mm_unpacklo_epi8(y_bias, y_bytes); + __m128i crw = _mm_unpacklo_epi8(_mm_setzero_si128(), cr_biased); + __m128i cbw = _mm_unpacklo_epi8(_mm_setzero_si128(), cb_biased); + + // color transform + __m128i yws = _mm_srli_epi16(yw, 4); + __m128i cr0 = _mm_mulhi_epi16(cr_const0, crw); + __m128i cb0 = _mm_mulhi_epi16(cb_const0, cbw); + __m128i cb1 = _mm_mulhi_epi16(cbw, cb_const1); + __m128i cr1 = _mm_mulhi_epi16(crw, cr_const1); + __m128i rws = _mm_add_epi16(cr0, yws); + __m128i gwt = _mm_add_epi16(cb0, yws); + __m128i bws = _mm_add_epi16(yws, cb1); + __m128i gws = _mm_add_epi16(gwt, cr1); + + // descale + __m128i rw = _mm_srai_epi16(rws, 4); + __m128i bw = _mm_srai_epi16(bws, 4); + __m128i gw = _mm_srai_epi16(gws, 4); + + // back to byte, set up for transpose + __m128i brb = _mm_packus_epi16(rw, bw); + __m128i gxb = _mm_packus_epi16(gw, xw); + + // transpose to interleave channels + __m128i t0 = _mm_unpacklo_epi8(brb, gxb); + __m128i t1 = _mm_unpackhi_epi8(brb, gxb); + __m128i o0 = _mm_unpacklo_epi16(t0, t1); + __m128i o1 = _mm_unpackhi_epi16(t0, t1); + + // store + _mm_storeu_si128((__m128i*)(out + 0), o0); + _mm_storeu_si128((__m128i*)(out + 16), o1); + out += 32; + } + } + #endif + + #ifdef STBI_NEON + // in this version, step=3 support would be easy to add. but is there demand? + if (step == 4) + { + // this is a fairly straightforward implementation and not super-optimized. + uint8x8_t signflip = vdup_n_u8(0x80); + int16x8_t cr_const0 = vdupq_n_s16((short)(1.40200f * 4096.0f + 0.5f)); + int16x8_t cr_const1 = vdupq_n_s16(-(short)(0.71414f * 4096.0f + 0.5f)); + int16x8_t cb_const0 = vdupq_n_s16(-(short)(0.34414f * 4096.0f + 0.5f)); + int16x8_t cb_const1 = vdupq_n_s16((short)(1.77200f * 4096.0f + 0.5f)); + + for (; i + 7 < count; i += 8) + { + // load + uint8x8_t y_bytes = vld1_u8(y + i); + uint8x8_t cr_bytes = vld1_u8(pcr + i); + uint8x8_t cb_bytes = vld1_u8(pcb + i); + int8x8_t cr_biased = vreinterpret_s8_u8(vsub_u8(cr_bytes, signflip)); + int8x8_t cb_biased = vreinterpret_s8_u8(vsub_u8(cb_bytes, signflip)); + + // expand to s16 + int16x8_t yws = vreinterpretq_s16_u16(vshll_n_u8(y_bytes, 4)); + int16x8_t crw = vshll_n_s8(cr_biased, 7); + int16x8_t cbw = vshll_n_s8(cb_biased, 7); + + // color transform + int16x8_t cr0 = vqdmulhq_s16(crw, cr_const0); + int16x8_t cb0 = vqdmulhq_s16(cbw, cb_const0); + int16x8_t cr1 = vqdmulhq_s16(crw, cr_const1); + int16x8_t cb1 = vqdmulhq_s16(cbw, cb_const1); + int16x8_t rws = vaddq_s16(yws, cr0); + int16x8_t gws = vaddq_s16(vaddq_s16(yws, cb0), cr1); + int16x8_t bws = vaddq_s16(yws, cb1); + + // undo scaling, round, convert to byte + uint8x8x4_t o; + o.val[0] = vqrshrun_n_s16(rws, 4); + o.val[1] = vqrshrun_n_s16(gws, 4); + o.val[2] = vqrshrun_n_s16(bws, 4); + o.val[3] = vdup_n_u8(255); + + // store, interleaving r/g/b/a + vst4_u8(out, o); + out += 8 * 4; + } + } + #endif + + for (; i < count; ++i) + { + int y_fixed = (y[i] << 20) + (1 << 19); // rounding + int r, g, b; + int cr = pcr[i] - 128; + int cb = pcb[i] - 128; + r = y_fixed + cr * stbi__float2fixed(1.40200f); + g = y_fixed + cr * -stbi__float2fixed(0.71414f) + ((cb * -stbi__float2fixed(0.34414f)) & 0xffff0000); + b = y_fixed + cb * stbi__float2fixed(1.77200f); + r >>= 20; + g >>= 20; + b >>= 20; + if ((unsigned)r > 255) + { + if (r < 0) + r = 0; + else + r = 255; + } + if ((unsigned)g > 255) + { + if (g < 0) + g = 0; + else + g = 255; + } + if ((unsigned)b > 255) + { + if (b < 0) + b = 0; + else + b = 255; + } + out[0] = (stbi_uc)r; + out[1] = (stbi_uc)g; + out[2] = (stbi_uc)b; + out[3] = 255; + out += step; + } +} + #endif + +// set up the kernels +static void stbi__setup_jpeg(stbi__jpeg* j) +{ + j->idct_block_kernel = stbi__idct_block; + j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_row; + j->resample_row_hv_2_kernel = stbi__resample_row_hv_2; + + #ifdef STBI_SSE2 + if (stbi__sse2_available()) + { + j->idct_block_kernel = stbi__idct_simd; + j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; + j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; + } + #endif + + #ifdef STBI_NEON + j->idct_block_kernel = stbi__idct_simd; + j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; + j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; + #endif +} + +// clean up the temporary component buffers +static void stbi__cleanup_jpeg(stbi__jpeg* j) +{ + stbi__free_jpeg_components(j, j->s->img_n, 0); +} + +typedef struct { + resample_row_func resample; + stbi_uc *line0, *line1; + int hs, vs; // expansion factor in each axis + int w_lores; // horizontal pixels pre-expansion + int ystep; // how far through vertical expansion we are + int ypos; // which pre-expansion row we're on +} stbi__resample; + +// fast 0..255 * 0..255 => 0..255 rounded multiplication +static stbi_uc stbi__blinn_8x8(stbi_uc x, stbi_uc y) +{ + unsigned int t = x * y + 128; + return (stbi_uc)((t + (t >> 8)) >> 8); +} + +static stbi_uc* load_jpeg_image(stbi__jpeg* z, int* out_x, int* out_y, int* comp, int req_comp) +{ + int n, decode_n, is_rgb; + z->s->img_n = 0; // make stbi__cleanup_jpeg safe + + // validate req_comp + if (req_comp < 0 || req_comp > 4) + return stbi__errpuc("bad req_comp", "Internal error"); + + // load a jpeg image from whichever source, but leave in YCbCr format + if (!stbi__decode_jpeg_image(z)) + { + stbi__cleanup_jpeg(z); + return NULL; + } + + // determine actual number of components to generate + n = req_comp ? req_comp : z->s->img_n >= 3 ? 3 : 1; + + is_rgb = z->s->img_n == 3 && (z->rgb == 3 || (z->app14_color_transform == 0 && !z->jfif)); + + if (z->s->img_n == 3 && n < 3 && !is_rgb) + decode_n = 1; + else + decode_n = z->s->img_n; + + // nothing to do if no components requested; check this now to avoid + // accessing uninitialized coutput[0] later + if (decode_n <= 0) + { + stbi__cleanup_jpeg(z); + return NULL; + } + + // resample and color-convert + { + int k; + unsigned int i, j; + stbi_uc* output; + stbi_uc* coutput[4] = {NULL, NULL, NULL, NULL}; + + stbi__resample res_comp[4]; + + for (k = 0; k < decode_n; ++k) + { + stbi__resample* r = &res_comp[k]; + + // allocate line buffer big enough for upsampling off the edges + // with upsample factor of 4 + z->img_comp[k].linebuf = (stbi_uc*)stbi__malloc(z->s->img_x + 3); + if (!z->img_comp[k].linebuf) + { + stbi__cleanup_jpeg(z); + return stbi__errpuc("outofmem", "Out of memory"); + } + + r->hs = z->img_h_max / z->img_comp[k].h; + r->vs = z->img_v_max / z->img_comp[k].v; + r->ystep = r->vs >> 1; + r->w_lores = (z->s->img_x + r->hs - 1) / r->hs; + r->ypos = 0; + r->line0 = r->line1 = z->img_comp[k].data; + + if (r->hs == 1 && r->vs == 1) + r->resample = resample_row_1; + else if (r->hs == 1 && r->vs == 2) + r->resample = stbi__resample_row_v_2; + else if (r->hs == 2 && r->vs == 1) + r->resample = stbi__resample_row_h_2; + else if (r->hs == 2 && r->vs == 2) + r->resample = z->resample_row_hv_2_kernel; + else + r->resample = stbi__resample_row_generic; + } + + // can't error after this so, this is safe + output = (stbi_uc*)stbi__malloc_mad3(n, z->s->img_x, z->s->img_y, 1); + if (!output) + { + stbi__cleanup_jpeg(z); + return stbi__errpuc("outofmem", "Out of memory"); + } + + // now go ahead and resample + for (j = 0; j < z->s->img_y; ++j) + { + stbi_uc* out = output + n * z->s->img_x * j; + for (k = 0; k < decode_n; ++k) + { + stbi__resample* r = &res_comp[k]; + int y_bot = r->ystep >= (r->vs >> 1); + coutput[k] = r->resample(z->img_comp[k].linebuf, y_bot ? r->line1 : r->line0, y_bot ? r->line0 : r->line1, r->w_lores, r->hs); + if (++r->ystep >= r->vs) + { + r->ystep = 0; + r->line0 = r->line1; + if (++r->ypos < z->img_comp[k].y) + r->line1 += z->img_comp[k].w2; + } + } + if (n >= 3) + { + stbi_uc* y = coutput[0]; + if (z->s->img_n == 3) + { + if (is_rgb) + { + for (i = 0; i < z->s->img_x; ++i) + { + out[0] = y[i]; + out[1] = coutput[1][i]; + out[2] = coutput[2][i]; + out[3] = 255; + out += n; + } + } + else + { + z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); + } + } + else if (z->s->img_n == 4) + { + if (z->app14_color_transform == 0) + { // CMYK + for (i = 0; i < z->s->img_x; ++i) + { + stbi_uc m = coutput[3][i]; + out[0] = stbi__blinn_8x8(coutput[0][i], m); + out[1] = stbi__blinn_8x8(coutput[1][i], m); + out[2] = stbi__blinn_8x8(coutput[2][i], m); + out[3] = 255; + out += n; + } + } + else if (z->app14_color_transform == 2) + { // YCCK + z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); + for (i = 0; i < z->s->img_x; ++i) + { + stbi_uc m = coutput[3][i]; + out[0] = stbi__blinn_8x8(255 - out[0], m); + out[1] = stbi__blinn_8x8(255 - out[1], m); + out[2] = stbi__blinn_8x8(255 - out[2], m); + out += n; + } + } + else + { // YCbCr + alpha? Ignore the fourth channel for now + z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); + } + } + else + for (i = 0; i < z->s->img_x; ++i) + { + out[0] = out[1] = out[2] = y[i]; + out[3] = 255; // not used if n==3 + out += n; + } + } + else + { + if (is_rgb) + { + if (n == 1) + for (i = 0; i < z->s->img_x; ++i) + *out++ = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]); + else + { + for (i = 0; i < z->s->img_x; ++i, out += 2) + { + out[0] = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]); + out[1] = 255; + } + } + } + else if (z->s->img_n == 4 && z->app14_color_transform == 0) + { + for (i = 0; i < z->s->img_x; ++i) + { + stbi_uc m = coutput[3][i]; + stbi_uc r = stbi__blinn_8x8(coutput[0][i], m); + stbi_uc g = stbi__blinn_8x8(coutput[1][i], m); + stbi_uc b = stbi__blinn_8x8(coutput[2][i], m); + out[0] = stbi__compute_y(r, g, b); + out[1] = 255; + out += n; + } + } + else if (z->s->img_n == 4 && z->app14_color_transform == 2) + { + for (i = 0; i < z->s->img_x; ++i) + { + out[0] = stbi__blinn_8x8(255 - coutput[0][i], coutput[3][i]); + out[1] = 255; + out += n; + } + } + else + { + stbi_uc* y = coutput[0]; + if (n == 1) + for (i = 0; i < z->s->img_x; ++i) + out[i] = y[i]; + else + for (i = 0; i < z->s->img_x; ++i) + { + *out++ = y[i]; + *out++ = 255; + } + } + } + } + stbi__cleanup_jpeg(z); + *out_x = z->s->img_x; + *out_y = z->s->img_y; + if (comp) + *comp = z->s->img_n >= 3 ? 3 : 1; // report original components, not output + return output; + } +} + +static void* stbi__jpeg_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) +{ + unsigned char* result; + stbi__jpeg* j = (stbi__jpeg*)stbi__malloc(sizeof(stbi__jpeg)); + if (!j) + return stbi__errpuc("outofmem", "Out of memory"); + memset(j, 0, sizeof(stbi__jpeg)); + STBI_NOTUSED(ri); + j->s = s; + stbi__setup_jpeg(j); + result = load_jpeg_image(j, x, y, comp, req_comp); + STBI_FREE(j); + return result; +} + +static int stbi__jpeg_test(stbi__context* s) +{ + int r; + stbi__jpeg* j = (stbi__jpeg*)stbi__malloc(sizeof(stbi__jpeg)); + if (!j) + return stbi__err("outofmem", "Out of memory"); + memset(j, 0, sizeof(stbi__jpeg)); + j->s = s; + stbi__setup_jpeg(j); + r = stbi__decode_jpeg_header(j, STBI__SCAN_type); + stbi__rewind(s); + STBI_FREE(j); + return r; +} + +static int stbi__jpeg_info_raw(stbi__jpeg* j, int* x, int* y, int* comp) +{ + if (!stbi__decode_jpeg_header(j, STBI__SCAN_header)) + { + stbi__rewind(j->s); + return 0; + } + if (x) + *x = j->s->img_x; + if (y) + *y = j->s->img_y; + if (comp) + *comp = j->s->img_n >= 3 ? 3 : 1; + return 1; +} + +static int stbi__jpeg_info(stbi__context* s, int* x, int* y, int* comp) +{ + int result; + stbi__jpeg* j = (stbi__jpeg*)(stbi__malloc(sizeof(stbi__jpeg))); + if (!j) + return stbi__err("outofmem", "Out of memory"); + memset(j, 0, sizeof(stbi__jpeg)); + j->s = s; + result = stbi__jpeg_info_raw(j, x, y, comp); + STBI_FREE(j); + return result; +} + #endif + +// public domain zlib decode v0.2 Sean Barrett 2006-11-18 +// simple implementation +// - all input must be provided in an upfront buffer +// - all output is written to a single output buffer (can malloc/realloc) +// performance +// - fast huffman + + #ifndef STBI_NO_ZLIB + + // fast-way is faster to check than jpeg huffman, but slow way is slower + #define STBI__ZFAST_BITS 9 // accelerate all cases in default tables + #define STBI__ZFAST_MASK ((1 << STBI__ZFAST_BITS) - 1) + #define STBI__ZNSYMS 288 // number of symbols in literal/length alphabet + +// zlib-style huffman encoding +// (jpegs packs from left, zlib from right, so can't share code) +typedef struct { + stbi__uint16 fast[1 << STBI__ZFAST_BITS]; + stbi__uint16 firstcode[16]; + int maxcode[17]; + stbi__uint16 firstsymbol[16]; + stbi_uc size[STBI__ZNSYMS]; + stbi__uint16 value[STBI__ZNSYMS]; +} stbi__zhuffman; + +stbi_inline static int stbi__bitreverse16(int n) +{ + n = ((n & 0xAAAA) >> 1) | ((n & 0x5555) << 1); + n = ((n & 0xCCCC) >> 2) | ((n & 0x3333) << 2); + n = ((n & 0xF0F0) >> 4) | ((n & 0x0F0F) << 4); + n = ((n & 0xFF00) >> 8) | ((n & 0x00FF) << 8); + return n; +} + +stbi_inline static int stbi__bit_reverse(int v, int bits) +{ + STBI_ASSERT(bits <= 16); + // to bit reverse n bits, reverse 16 and shift + // e.g. 11 bits, bit reverse and shift away 5 + return stbi__bitreverse16(v) >> (16 - bits); +} + +static int stbi__zbuild_huffman(stbi__zhuffman* z, const stbi_uc* sizelist, int num) +{ + int i, k = 0; + int code, next_code[16], sizes[17]; + + // DEFLATE spec for generating codes + memset(sizes, 0, sizeof(sizes)); + memset(z->fast, 0, sizeof(z->fast)); + for (i = 0; i < num; ++i) + ++sizes[sizelist[i]]; + sizes[0] = 0; + for (i = 1; i < 16; ++i) + if (sizes[i] > (1 << i)) + return stbi__err("bad sizes", "Corrupt PNG"); + code = 0; + for (i = 1; i < 16; ++i) + { + next_code[i] = code; + z->firstcode[i] = (stbi__uint16)code; + z->firstsymbol[i] = (stbi__uint16)k; + code = (code + sizes[i]); + if (sizes[i]) + if (code - 1 >= (1 << i)) + return stbi__err("bad codelengths", "Corrupt PNG"); + z->maxcode[i] = code << (16 - i); // preshift for inner loop + code <<= 1; + k += sizes[i]; + } + z->maxcode[16] = 0x10000; // sentinel + for (i = 0; i < num; ++i) + { + int s = sizelist[i]; + if (s) + { + int c = next_code[s] - z->firstcode[s] + z->firstsymbol[s]; + stbi__uint16 fastv = (stbi__uint16)((s << 9) | i); + z->size[c] = (stbi_uc)s; + z->value[c] = (stbi__uint16)i; + if (s <= STBI__ZFAST_BITS) + { + int j = stbi__bit_reverse(next_code[s], s); + while (j < (1 << STBI__ZFAST_BITS)) + { + z->fast[j] = fastv; + j += (1 << s); + } + } + ++next_code[s]; + } + } + return 1; +} + +// zlib-from-memory implementation for PNG reading +// because PNG allows splitting the zlib stream arbitrarily, +// and it's annoying structurally to have PNG call ZLIB call PNG, +// we require PNG read all the IDATs and combine them into a single +// memory buffer + +typedef struct { + stbi_uc *zbuffer, *zbuffer_end; + int num_bits; + int hit_zeof_once; + stbi__uint32 code_buffer; + + char* zout; + char* zout_start; + char* zout_end; + int z_expandable; + + stbi__zhuffman z_length, z_distance; +} stbi__zbuf; + +stbi_inline static int stbi__zeof(stbi__zbuf* z) +{ + return (z->zbuffer >= z->zbuffer_end); +} + +stbi_inline static stbi_uc stbi__zget8(stbi__zbuf* z) +{ + return stbi__zeof(z) ? 0 : *z->zbuffer++; +} + +static void stbi__fill_bits(stbi__zbuf* z) +{ + do + { + if (z->code_buffer >= (1U << z->num_bits)) + { + z->zbuffer = z->zbuffer_end; /* treat this as EOF so we fail. */ + return; + } + z->code_buffer |= (unsigned int)stbi__zget8(z) << z->num_bits; + z->num_bits += 8; + } while (z->num_bits <= 24); +} + +stbi_inline static unsigned int stbi__zreceive(stbi__zbuf* z, int n) +{ + unsigned int k; + if (z->num_bits < n) + stbi__fill_bits(z); + k = z->code_buffer & ((1 << n) - 1); + z->code_buffer >>= n; + z->num_bits -= n; + return k; +} + +static int stbi__zhuffman_decode_slowpath(stbi__zbuf* a, stbi__zhuffman* z) +{ + int b, s, k; + // not resolved by fast table, so compute it the slow way + // use jpeg approach, which requires MSbits at top + k = stbi__bit_reverse(a->code_buffer, 16); + for (s = STBI__ZFAST_BITS + 1;; ++s) + if (k < z->maxcode[s]) + break; + if (s >= 16) + return -1; // invalid code! + // code size is s, so: + b = (k >> (16 - s)) - z->firstcode[s] + z->firstsymbol[s]; + if (b >= STBI__ZNSYMS) + return -1; // some data was corrupt somewhere! + if (z->size[b] != s) + return -1; // was originally an assert, but report failure instead. + a->code_buffer >>= s; + a->num_bits -= s; + return z->value[b]; +} + +stbi_inline static int stbi__zhuffman_decode(stbi__zbuf* a, stbi__zhuffman* z) +{ + int b, s; + if (a->num_bits < 16) + { + if (stbi__zeof(a)) + { + if (!a->hit_zeof_once) + { + // This is the first time we hit eof, insert 16 extra padding btis + // to allow us to keep going; if we actually consume any of them + // though, that is invalid data. This is caught later. + a->hit_zeof_once = 1; + a->num_bits += 16; // add 16 implicit zero bits + } + else + { + // We already inserted our extra 16 padding bits and are again + // out, this stream is actually prematurely terminated. + return -1; + } + } + else + { + stbi__fill_bits(a); + } + } + b = z->fast[a->code_buffer & STBI__ZFAST_MASK]; + if (b) + { + s = b >> 9; + a->code_buffer >>= s; + a->num_bits -= s; + return b & 511; + } + return stbi__zhuffman_decode_slowpath(a, z); +} + +static int stbi__zexpand(stbi__zbuf* z, char* zout, int n) // need to make room for n bytes +{ + char* q; + unsigned int cur, limit, old_limit; + z->zout = zout; + if (!z->z_expandable) + return stbi__err("output buffer limit", "Corrupt PNG"); + cur = (unsigned int)(z->zout - z->zout_start); + limit = old_limit = (unsigned)(z->zout_end - z->zout_start); + if (UINT_MAX - cur < (unsigned)n) + return stbi__err("outofmem", "Out of memory"); + while (cur + n > limit) + { + if (limit > UINT_MAX / 2) + return stbi__err("outofmem", "Out of memory"); + limit *= 2; + } + q = (char*)STBI_REALLOC_SIZED(z->zout_start, old_limit, limit); + STBI_NOTUSED(old_limit); + if (q == NULL) + return stbi__err("outofmem", "Out of memory"); + z->zout_start = q; + z->zout = q + cur; + z->zout_end = q + limit; + return 1; +} + +static const int stbi__zlength_base[31] = {3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31, 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, + 227, 258, 0, 0}; + +static const int stbi__zlength_extra[31] = {0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 0, 0}; + +static const int stbi__zdist_base[32] = {1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193, 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, + 6145, 8193, 12289, 16385, 24577, 0, 0}; + +static const int stbi__zdist_extra[32] = {0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13}; + +static int stbi__parse_huffman_block(stbi__zbuf* a) +{ + char* zout = a->zout; + for (;;) + { + int z = stbi__zhuffman_decode(a, &a->z_length); + if (z < 256) + { + if (z < 0) + return stbi__err("bad huffman code", "Corrupt PNG"); // error in huffman codes + if (zout >= a->zout_end) + { + if (!stbi__zexpand(a, zout, 1)) + return 0; + zout = a->zout; + } + *zout++ = (char)z; + } + else + { + stbi_uc* p; + int len, dist; + if (z == 256) + { + a->zout = zout; + if (a->hit_zeof_once && a->num_bits < 16) + { + // The first time we hit zeof, we inserted 16 extra zero bits into our bit + // buffer so the decoder can just do its speculative decoding. But if we + // actually consumed any of those bits (which is the case when num_bits < 16), + // the stream actually read past the end so it is malformed. + return stbi__err("unexpected end", "Corrupt PNG"); + } + return 1; + } + if (z >= 286) + return stbi__err("bad huffman code", "Corrupt PNG"); // per DEFLATE, length codes 286 and 287 must not appear in compressed data + z -= 257; + len = stbi__zlength_base[z]; + if (stbi__zlength_extra[z]) + len += stbi__zreceive(a, stbi__zlength_extra[z]); + z = stbi__zhuffman_decode(a, &a->z_distance); + if (z < 0 || z >= 30) + return stbi__err("bad huffman code", "Corrupt PNG"); // per DEFLATE, distance codes 30 and 31 must not appear in compressed data + dist = stbi__zdist_base[z]; + if (stbi__zdist_extra[z]) + dist += stbi__zreceive(a, stbi__zdist_extra[z]); + if (zout - a->zout_start < dist) + return stbi__err("bad dist", "Corrupt PNG"); + if (len > a->zout_end - zout) + { + if (!stbi__zexpand(a, zout, len)) + return 0; + zout = a->zout; + } + p = (stbi_uc*)(zout - dist); + if (dist == 1) + { // run of one byte; common in images. + stbi_uc v = *p; + if (len) + { + do + *zout++ = v; + while (--len); + } + } + else + { + if (len) + { + do + *zout++ = *p++; + while (--len); + } + } + } + } +} + +static int stbi__compute_huffman_codes(stbi__zbuf* a) +{ + static const stbi_uc length_dezigzag[19] = {16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15}; + stbi__zhuffman z_codelength; + stbi_uc lencodes[286 + 32 + 137]; // padding for maximum single op + stbi_uc codelength_sizes[19]; + int i, n; + + int hlit = stbi__zreceive(a, 5) + 257; + int hdist = stbi__zreceive(a, 5) + 1; + int hclen = stbi__zreceive(a, 4) + 4; + int ntot = hlit + hdist; + + memset(codelength_sizes, 0, sizeof(codelength_sizes)); + for (i = 0; i < hclen; ++i) + { + int s = stbi__zreceive(a, 3); + codelength_sizes[length_dezigzag[i]] = (stbi_uc)s; + } + if (!stbi__zbuild_huffman(&z_codelength, codelength_sizes, 19)) + return 0; + + n = 0; + while (n < ntot) + { + int c = stbi__zhuffman_decode(a, &z_codelength); + if (c < 0 || c >= 19) + return stbi__err("bad codelengths", "Corrupt PNG"); + if (c < 16) + lencodes[n++] = (stbi_uc)c; + else + { + stbi_uc fill = 0; + if (c == 16) + { + c = stbi__zreceive(a, 2) + 3; + if (n == 0) + return stbi__err("bad codelengths", "Corrupt PNG"); + fill = lencodes[n - 1]; + } + else if (c == 17) + { + c = stbi__zreceive(a, 3) + 3; + } + else if (c == 18) + { + c = stbi__zreceive(a, 7) + 11; + } + else + { + return stbi__err("bad codelengths", "Corrupt PNG"); + } + if (ntot - n < c) + return stbi__err("bad codelengths", "Corrupt PNG"); + memset(lencodes + n, fill, c); + n += c; + } + } + if (n != ntot) + return stbi__err("bad codelengths", "Corrupt PNG"); + if (!stbi__zbuild_huffman(&a->z_length, lencodes, hlit)) + return 0; + if (!stbi__zbuild_huffman(&a->z_distance, lencodes + hlit, hdist)) + return 0; + return 1; +} + +static int stbi__parse_uncompressed_block(stbi__zbuf* a) +{ + stbi_uc header[4]; + int len, nlen, k; + if (a->num_bits & 7) + stbi__zreceive(a, a->num_bits & 7); // discard + // drain the bit-packed data into header + k = 0; + while (a->num_bits > 0) + { + header[k++] = (stbi_uc)(a->code_buffer & 255); // suppress MSVC run-time check + a->code_buffer >>= 8; + a->num_bits -= 8; + } + if (a->num_bits < 0) + return stbi__err("zlib corrupt", "Corrupt PNG"); + // now fill header the normal way + while (k < 4) + header[k++] = stbi__zget8(a); + len = header[1] * 256 + header[0]; + nlen = header[3] * 256 + header[2]; + if (nlen != (len ^ 0xffff)) + return stbi__err("zlib corrupt", "Corrupt PNG"); + if (a->zbuffer + len > a->zbuffer_end) + return stbi__err("read past buffer", "Corrupt PNG"); + if (a->zout + len > a->zout_end) + if (!stbi__zexpand(a, a->zout, len)) + return 0; + memcpy(a->zout, a->zbuffer, len); + a->zbuffer += len; + a->zout += len; + return 1; +} + +static int stbi__parse_zlib_header(stbi__zbuf* a) +{ + int cmf = stbi__zget8(a); + int cm = cmf & 15; + /* int cinfo = cmf >> 4; */ + int flg = stbi__zget8(a); + if (stbi__zeof(a)) + return stbi__err("bad zlib header", "Corrupt PNG"); // zlib spec + if ((cmf * 256 + flg) % 31 != 0) + return stbi__err("bad zlib header", "Corrupt PNG"); // zlib spec + if (flg & 32) + return stbi__err("no preset dict", "Corrupt PNG"); // preset dictionary not allowed in png + if (cm != 8) + return stbi__err("bad compression", "Corrupt PNG"); // DEFLATE required for png + // window = 1 << (8 + cinfo)... but who cares, we fully buffer output + return 1; +} + +static const stbi_uc stbi__zdefault_length[STBI__ZNSYMS] = {8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8}; +static const stbi_uc stbi__zdefault_distance[32] = {5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5}; +/* +Init algorithm: +{ + int i; // use <= to match clearly with spec + for (i=0; i <= 143; ++i) stbi__zdefault_length[i] = 8; + for ( ; i <= 255; ++i) stbi__zdefault_length[i] = 9; + for ( ; i <= 279; ++i) stbi__zdefault_length[i] = 7; + for ( ; i <= 287; ++i) stbi__zdefault_length[i] = 8; + + for (i=0; i <= 31; ++i) stbi__zdefault_distance[i] = 5; +} +*/ + +static int stbi__parse_zlib(stbi__zbuf* a, int parse_header) +{ + int final, type; + if (parse_header) + if (!stbi__parse_zlib_header(a)) + return 0; + a->num_bits = 0; + a->code_buffer = 0; + a->hit_zeof_once = 0; + do + { + final = stbi__zreceive(a, 1); + type = stbi__zreceive(a, 2); + if (type == 0) + { + if (!stbi__parse_uncompressed_block(a)) + return 0; + } + else if (type == 3) + { + return 0; + } + else + { + if (type == 1) + { + // use fixed code lengths + if (!stbi__zbuild_huffman(&a->z_length, stbi__zdefault_length, STBI__ZNSYMS)) + return 0; + if (!stbi__zbuild_huffman(&a->z_distance, stbi__zdefault_distance, 32)) + return 0; + } + else + { + if (!stbi__compute_huffman_codes(a)) + return 0; + } + if (!stbi__parse_huffman_block(a)) + return 0; + } + } while (!final); + return 1; +} + +static int stbi__do_zlib(stbi__zbuf* a, char* obuf, int olen, int exp, int parse_header) +{ + a->zout_start = obuf; + a->zout = obuf; + a->zout_end = obuf + olen; + a->z_expandable = exp; + + return stbi__parse_zlib(a, parse_header); +} + +STBIDEF char* stbi_zlib_decode_malloc_guesssize(const char* buffer, int len, int initial_size, int* outlen) +{ + stbi__zbuf a; + char* p = (char*)stbi__malloc(initial_size); + if (p == NULL) + return NULL; + a.zbuffer = (stbi_uc*)buffer; + a.zbuffer_end = (stbi_uc*)buffer + len; + if (stbi__do_zlib(&a, p, initial_size, 1, 1)) + { + if (outlen) + *outlen = (int)(a.zout - a.zout_start); + return a.zout_start; + } + else + { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF char* stbi_zlib_decode_malloc(const char* buffer, int len, int* outlen) +{ + return stbi_zlib_decode_malloc_guesssize(buffer, len, 16384, outlen); +} + +STBIDEF char* stbi_zlib_decode_malloc_guesssize_headerflag(const char* buffer, int len, int initial_size, int* outlen, int parse_header) +{ + stbi__zbuf a; + char* p = (char*)stbi__malloc(initial_size); + if (p == NULL) + return NULL; + a.zbuffer = (stbi_uc*)buffer; + a.zbuffer_end = (stbi_uc*)buffer + len; + if (stbi__do_zlib(&a, p, initial_size, 1, parse_header)) + { + if (outlen) + *outlen = (int)(a.zout - a.zout_start); + return a.zout_start; + } + else + { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF int stbi_zlib_decode_buffer(char* obuffer, int olen, const char* ibuffer, int ilen) +{ + stbi__zbuf a; + a.zbuffer = (stbi_uc*)ibuffer; + a.zbuffer_end = (stbi_uc*)ibuffer + ilen; + if (stbi__do_zlib(&a, obuffer, olen, 0, 1)) + return (int)(a.zout - a.zout_start); + else + return -1; +} + +STBIDEF char* stbi_zlib_decode_noheader_malloc(const char* buffer, int len, int* outlen) +{ + stbi__zbuf a; + char* p = (char*)stbi__malloc(16384); + if (p == NULL) + return NULL; + a.zbuffer = (stbi_uc*)buffer; + a.zbuffer_end = (stbi_uc*)buffer + len; + if (stbi__do_zlib(&a, p, 16384, 1, 0)) + { + if (outlen) + *outlen = (int)(a.zout - a.zout_start); + return a.zout_start; + } + else + { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF int stbi_zlib_decode_noheader_buffer(char* obuffer, int olen, const char* ibuffer, int ilen) +{ + stbi__zbuf a; + a.zbuffer = (stbi_uc*)ibuffer; + a.zbuffer_end = (stbi_uc*)ibuffer + ilen; + if (stbi__do_zlib(&a, obuffer, olen, 0, 0)) + return (int)(a.zout - a.zout_start); + else + return -1; +} + #endif + +// public domain "baseline" PNG decoder v0.10 Sean Barrett 2006-11-18 +// simple implementation +// - only 8-bit samples +// - no CRC checking +// - allocates lots of intermediate memory +// - avoids problem of streaming data between subsystems +// - avoids explicit window management +// performance +// - uses stb_zlib, a PD zlib implementation with fast huffman decoding + + #ifndef STBI_NO_PNG +typedef struct { + stbi__uint32 length; + stbi__uint32 type; +} stbi__pngchunk; + +static stbi__pngchunk stbi__get_chunk_header(stbi__context* s) +{ + stbi__pngchunk c; + c.length = stbi__get32be(s); + c.type = stbi__get32be(s); + return c; +} + +static int stbi__check_png_header(stbi__context* s) +{ + static const stbi_uc png_sig[8] = {137, 80, 78, 71, 13, 10, 26, 10}; + int i; + for (i = 0; i < 8; ++i) + if (stbi__get8(s) != png_sig[i]) + return stbi__err("bad png sig", "Not a PNG"); + return 1; +} + +typedef struct { + stbi__context* s; + stbi_uc *idata, *expanded, *out; + int depth; +} stbi__png; + +enum { + STBI__F_none = 0, + STBI__F_sub = 1, + STBI__F_up = 2, + STBI__F_avg = 3, + STBI__F_paeth = 4, + // synthetic filter used for first scanline to avoid needing a dummy row of 0s + STBI__F_avg_first +}; + +static stbi_uc first_row_filter[5] = { + STBI__F_none, STBI__F_sub, STBI__F_none, STBI__F_avg_first, + STBI__F_sub // Paeth with b=c=0 turns out to be equivalent to sub +}; + +static int stbi__paeth(int a, int b, int c) +{ + // This formulation looks very different from the reference in the PNG spec, but is + // actually equivalent and has favorable data dependencies and admits straightforward + // generation of branch-free code, which helps performance significantly. + int thresh = c * 3 - (a + b); + int lo = a < b ? a : b; + int hi = a < b ? b : a; + int t0 = (hi <= thresh) ? lo : c; + int t1 = (thresh <= lo) ? hi : t0; + return t1; +} + +static const stbi_uc stbi__depth_scale_table[9] = {0, 0xff, 0x55, 0, 0x11, 0, 0, 0, 0x01}; + +// adds an extra all-255 alpha channel +// dest == src is legal +// img_n must be 1 or 3 +static void stbi__create_png_alpha_expand8(stbi_uc* dest, stbi_uc* src, stbi__uint32 x, int img_n) +{ + int i; + // must process data backwards since we allow dest==src + if (img_n == 1) + { + for (i = x - 1; i >= 0; --i) + { + dest[i * 2 + 1] = 255; + dest[i * 2 + 0] = src[i]; + } + } + else + { + STBI_ASSERT(img_n == 3); + for (i = x - 1; i >= 0; --i) + { + dest[i * 4 + 3] = 255; + dest[i * 4 + 2] = src[i * 3 + 2]; + dest[i * 4 + 1] = src[i * 3 + 1]; + dest[i * 4 + 0] = src[i * 3 + 0]; + } + } +} + +// create the png data from post-deflated data +static int stbi__create_png_image_raw(stbi__png* a, stbi_uc* raw, stbi__uint32 raw_len, int out_n, stbi__uint32 x, stbi__uint32 y, int depth, + int color) +{ + int bytes = (depth == 16 ? 2 : 1); + stbi__context* s = a->s; + stbi__uint32 i, j, stride = x * out_n * bytes; + stbi__uint32 img_len, img_width_bytes; + stbi_uc* filter_buf; + int all_ok = 1; + int k; + int img_n = s->img_n; // copy it into a local for later + + int output_bytes = out_n * bytes; + int filter_bytes = img_n * bytes; + int width = x; + + STBI_ASSERT(out_n == s->img_n || out_n == s->img_n + 1); + a->out = (stbi_uc*)stbi__malloc_mad3(x, y, output_bytes, 0); // extra bytes to write off the end into + if (!a->out) + return stbi__err("outofmem", "Out of memory"); + + // note: error exits here don't need to clean up a->out individually, + // stbi__do_png always does on error. + if (!stbi__mad3sizes_valid(img_n, x, depth, 7)) + return stbi__err("too large", "Corrupt PNG"); + img_width_bytes = (((img_n * x * depth) + 7) >> 3); + if (!stbi__mad2sizes_valid(img_width_bytes, y, img_width_bytes)) + return stbi__err("too large", "Corrupt PNG"); + img_len = (img_width_bytes + 1) * y; + + // we used to check for exact match between raw_len and img_len on non-interlaced PNGs, + // but issue #276 reported a PNG in the wild that had extra data at the end (all zeros), + // so just check for raw_len < img_len always. + if (raw_len < img_len) + return stbi__err("not enough pixels", "Corrupt PNG"); + + // Allocate two scan lines worth of filter workspace buffer. + filter_buf = (stbi_uc*)stbi__malloc_mad2(img_width_bytes, 2, 0); + if (!filter_buf) + return stbi__err("outofmem", "Out of memory"); + + // Filtering for low-bit-depth images + if (depth < 8) + { + filter_bytes = 1; + width = img_width_bytes; + } + + for (j = 0; j < y; ++j) + { + // cur/prior filter buffers alternate + stbi_uc* cur = filter_buf + (j & 1) * img_width_bytes; + stbi_uc* prior = filter_buf + (~j & 1) * img_width_bytes; + stbi_uc* dest = a->out + stride * j; + int nk = width * filter_bytes; + int filter = *raw++; + + // check filter type + if (filter > 4) + { + all_ok = stbi__err("invalid filter", "Corrupt PNG"); + break; + } + + // if first row, use special filter that doesn't sample previous row + if (j == 0) + filter = first_row_filter[filter]; + + // perform actual filtering + switch (filter) + { + case STBI__F_none: memcpy(cur, raw, nk); break; + case STBI__F_sub: + memcpy(cur, raw, filter_bytes); + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + cur[k - filter_bytes]); + break; + case STBI__F_up: + for (k = 0; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + prior[k]); + break; + case STBI__F_avg: + for (k = 0; k < filter_bytes; ++k) + cur[k] = STBI__BYTECAST(raw[k] + (prior[k] >> 1)); + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k - filter_bytes]) >> 1)); + break; + case STBI__F_paeth: + for (k = 0; k < filter_bytes; ++k) + cur[k] = STBI__BYTECAST(raw[k] + prior[k]); // prior[k] == stbi__paeth(0,prior[k],0) + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k - filter_bytes], prior[k], prior[k - filter_bytes])); + break; + case STBI__F_avg_first: + memcpy(cur, raw, filter_bytes); + for (k = filter_bytes; k < nk; ++k) + cur[k] = STBI__BYTECAST(raw[k] + (cur[k - filter_bytes] >> 1)); + break; + } + + raw += nk; + + // expand decoded bits in cur to dest, also adding an extra alpha channel if desired + if (depth < 8) + { + stbi_uc scale = (color == 0) ? stbi__depth_scale_table[depth] : 1; // scale grayscale values to 0..255 range + stbi_uc* in = cur; + stbi_uc* out = dest; + stbi_uc inb = 0; + stbi__uint32 nsmp = x * img_n; + + // expand bits to bytes first + if (depth == 4) + { + for (i = 0; i < nsmp; ++i) + { + if ((i & 1) == 0) + inb = *in++; + *out++ = scale * (inb >> 4); + inb <<= 4; + } + } + else if (depth == 2) + { + for (i = 0; i < nsmp; ++i) + { + if ((i & 3) == 0) + inb = *in++; + *out++ = scale * (inb >> 6); + inb <<= 2; + } + } + else + { + STBI_ASSERT(depth == 1); + for (i = 0; i < nsmp; ++i) + { + if ((i & 7) == 0) + inb = *in++; + *out++ = scale * (inb >> 7); + inb <<= 1; + } + } + + // insert alpha=255 values if desired + if (img_n != out_n) + stbi__create_png_alpha_expand8(dest, dest, x, img_n); + } + else if (depth == 8) + { + if (img_n == out_n) + memcpy(dest, cur, x * img_n); + else + stbi__create_png_alpha_expand8(dest, cur, x, img_n); + } + else if (depth == 16) + { + // convert the image data from big-endian to platform-native + stbi__uint16* dest16 = (stbi__uint16*)dest; + stbi__uint32 nsmp = x * img_n; + + if (img_n == out_n) + { + for (i = 0; i < nsmp; ++i, ++dest16, cur += 2) + *dest16 = (cur[0] << 8) | cur[1]; + } + else + { + STBI_ASSERT(img_n + 1 == out_n); + if (img_n == 1) + { + for (i = 0; i < x; ++i, dest16 += 2, cur += 2) + { + dest16[0] = (cur[0] << 8) | cur[1]; + dest16[1] = 0xffff; + } + } + else + { + STBI_ASSERT(img_n == 3); + for (i = 0; i < x; ++i, dest16 += 4, cur += 6) + { + dest16[0] = (cur[0] << 8) | cur[1]; + dest16[1] = (cur[2] << 8) | cur[3]; + dest16[2] = (cur[4] << 8) | cur[5]; + dest16[3] = 0xffff; + } + } + } + } + } + + STBI_FREE(filter_buf); + if (!all_ok) + return 0; + + return 1; +} + +static int stbi__create_png_image(stbi__png* a, stbi_uc* image_data, stbi__uint32 image_data_len, int out_n, int depth, int color, int interlaced) +{ + int bytes = (depth == 16 ? 2 : 1); + int out_bytes = out_n * bytes; + stbi_uc* final; + int p; + if (!interlaced) + return stbi__create_png_image_raw(a, image_data, image_data_len, out_n, a->s->img_x, a->s->img_y, depth, color); + + // de-interlacing + final = (stbi_uc*)stbi__malloc_mad3(a->s->img_x, a->s->img_y, out_bytes, 0); + if (!final) + return stbi__err("outofmem", "Out of memory"); + for (p = 0; p < 7; ++p) + { + int xorig[] = {0, 4, 0, 2, 0, 1, 0}; + int yorig[] = {0, 0, 4, 0, 2, 0, 1}; + int xspc[] = {8, 8, 4, 4, 2, 2, 1}; + int yspc[] = {8, 8, 8, 4, 4, 2, 2}; + int i, j, x, y; + // pass1_x[4] = 0, pass1_x[5] = 1, pass1_x[12] = 1 + x = (a->s->img_x - xorig[p] + xspc[p] - 1) / xspc[p]; + y = (a->s->img_y - yorig[p] + yspc[p] - 1) / yspc[p]; + if (x && y) + { + stbi__uint32 img_len = ((((a->s->img_n * x * depth) + 7) >> 3) + 1) * y; + if (!stbi__create_png_image_raw(a, image_data, image_data_len, out_n, x, y, depth, color)) + { + STBI_FREE(final); + return 0; + } + for (j = 0; j < y; ++j) + { + for (i = 0; i < x; ++i) + { + int out_y = j * yspc[p] + yorig[p]; + int out_x = i * xspc[p] + xorig[p]; + memcpy(final + out_y * a->s->img_x * out_bytes + out_x * out_bytes, a->out + (j * x + i) * out_bytes, out_bytes); + } + } + STBI_FREE(a->out); + image_data += img_len; + image_data_len -= img_len; + } + } + a->out = final; + + return 1; +} + +static int stbi__compute_transparency(stbi__png* z, stbi_uc tc[3], int out_n) +{ + stbi__context* s = z->s; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi_uc* p = z->out; + + // compute color-based transparency, assuming we've + // already got 255 as the alpha value in the output + STBI_ASSERT(out_n == 2 || out_n == 4); + + if (out_n == 2) + { + for (i = 0; i < pixel_count; ++i) + { + p[1] = (p[0] == tc[0] ? 0 : 255); + p += 2; + } + } + else + { + for (i = 0; i < pixel_count; ++i) + { + if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2]) + p[3] = 0; + p += 4; + } + } + return 1; +} + +static int stbi__compute_transparency16(stbi__png* z, stbi__uint16 tc[3], int out_n) +{ + stbi__context* s = z->s; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi__uint16* p = (stbi__uint16*)z->out; + + // compute color-based transparency, assuming we've + // already got 65535 as the alpha value in the output + STBI_ASSERT(out_n == 2 || out_n == 4); + + if (out_n == 2) + { + for (i = 0; i < pixel_count; ++i) + { + p[1] = (p[0] == tc[0] ? 0 : 65535); + p += 2; + } + } + else + { + for (i = 0; i < pixel_count; ++i) + { + if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2]) + p[3] = 0; + p += 4; + } + } + return 1; +} + +static int stbi__expand_png_palette(stbi__png* a, stbi_uc* palette, int len, int pal_img_n) +{ + stbi__uint32 i, pixel_count = a->s->img_x * a->s->img_y; + stbi_uc *p, *temp_out, *orig = a->out; + + p = (stbi_uc*)stbi__malloc_mad2(pixel_count, pal_img_n, 0); + if (p == NULL) + return stbi__err("outofmem", "Out of memory"); + + // between here and free(out) below, exitting would leak + temp_out = p; + + if (pal_img_n == 3) + { + for (i = 0; i < pixel_count; ++i) + { + int n = orig[i] * 4; + p[0] = palette[n]; + p[1] = palette[n + 1]; + p[2] = palette[n + 2]; + p += 3; + } + } + else + { + for (i = 0; i < pixel_count; ++i) + { + int n = orig[i] * 4; + p[0] = palette[n]; + p[1] = palette[n + 1]; + p[2] = palette[n + 2]; + p[3] = palette[n + 3]; + p += 4; + } + } + STBI_FREE(a->out); + a->out = temp_out; + + STBI_NOTUSED(len); + + return 1; +} + +static int stbi__unpremultiply_on_load_global = 0; +static int stbi__de_iphone_flag_global = 0; + +STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply) +{ + stbi__unpremultiply_on_load_global = flag_true_if_should_unpremultiply; +} + +STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert) +{ + stbi__de_iphone_flag_global = flag_true_if_should_convert; +} + + #ifndef STBI_THREAD_LOCAL + #define stbi__unpremultiply_on_load stbi__unpremultiply_on_load_global + #define stbi__de_iphone_flag stbi__de_iphone_flag_global + #else +static STBI_THREAD_LOCAL int stbi__unpremultiply_on_load_local, stbi__unpremultiply_on_load_set; +static STBI_THREAD_LOCAL int stbi__de_iphone_flag_local, stbi__de_iphone_flag_set; + +STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply) +{ + stbi__unpremultiply_on_load_local = flag_true_if_should_unpremultiply; + stbi__unpremultiply_on_load_set = 1; +} + +STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert) +{ + stbi__de_iphone_flag_local = flag_true_if_should_convert; + stbi__de_iphone_flag_set = 1; +} + + #define stbi__unpremultiply_on_load \ + (stbi__unpremultiply_on_load_set ? stbi__unpremultiply_on_load_local : stbi__unpremultiply_on_load_global) + #define stbi__de_iphone_flag (stbi__de_iphone_flag_set ? stbi__de_iphone_flag_local : stbi__de_iphone_flag_global) + #endif // STBI_THREAD_LOCAL + +static void stbi__de_iphone(stbi__png* z) +{ + stbi__context* s = z->s; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi_uc* p = z->out; + + if (s->img_out_n == 3) + { // convert bgr to rgb + for (i = 0; i < pixel_count; ++i) + { + stbi_uc t = p[0]; + p[0] = p[2]; + p[2] = t; + p += 3; + } + } + else + { + STBI_ASSERT(s->img_out_n == 4); + if (stbi__unpremultiply_on_load) + { + // convert bgr to rgb and unpremultiply + for (i = 0; i < pixel_count; ++i) + { + stbi_uc a = p[3]; + stbi_uc t = p[0]; + if (a) + { + stbi_uc half = a / 2; + p[0] = (p[2] * 255 + half) / a; + p[1] = (p[1] * 255 + half) / a; + p[2] = (t * 255 + half) / a; + } + else + { + p[0] = p[2]; + p[2] = t; + } + p += 4; + } + } + else + { + // convert bgr to rgb + for (i = 0; i < pixel_count; ++i) + { + stbi_uc t = p[0]; + p[0] = p[2]; + p[2] = t; + p += 4; + } + } + } +} + + #define STBI__PNG_TYPE(a, b, c, d) (((unsigned)(a) << 24) + ((unsigned)(b) << 16) + ((unsigned)(c) << 8) + (unsigned)(d)) + +static int stbi__parse_png_file(stbi__png* z, int scan, int req_comp) +{ + stbi_uc palette[1024], pal_img_n = 0; + stbi_uc has_trans = 0, tc[3] = {0}; + stbi__uint16 tc16[3]; + stbi__uint32 ioff = 0, idata_limit = 0, i, pal_len = 0; + int first = 1, k, interlace = 0, color = 0, is_iphone = 0; + stbi__context* s = z->s; + + z->expanded = NULL; + z->idata = NULL; + z->out = NULL; + + if (!stbi__check_png_header(s)) + return 0; + + if (scan == STBI__SCAN_type) + return 1; + + for (;;) + { + stbi__pngchunk c = stbi__get_chunk_header(s); + switch (c.type) + { + case STBI__PNG_TYPE('C', 'g', 'B', 'I'): + is_iphone = 1; + stbi__skip(s, c.length); + break; + case STBI__PNG_TYPE('I', 'H', 'D', 'R'): + { + int comp, filter; + if (!first) + return stbi__err("multiple IHDR", "Corrupt PNG"); + first = 0; + if (c.length != 13) + return stbi__err("bad IHDR len", "Corrupt PNG"); + s->img_x = stbi__get32be(s); + s->img_y = stbi__get32be(s); + if (s->img_y > STBI_MAX_DIMENSIONS) + return stbi__err("too large", "Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) + return stbi__err("too large", "Very large image (corrupt?)"); + z->depth = stbi__get8(s); + if (z->depth != 1 && z->depth != 2 && z->depth != 4 && z->depth != 8 && z->depth != 16) + return stbi__err("1/2/4/8/16-bit only", "PNG not supported: 1/2/4/8/16-bit only"); + color = stbi__get8(s); + if (color > 6) + return stbi__err("bad ctype", "Corrupt PNG"); + if (color == 3 && z->depth == 16) + return stbi__err("bad ctype", "Corrupt PNG"); + if (color == 3) + pal_img_n = 3; + else if (color & 1) + return stbi__err("bad ctype", "Corrupt PNG"); + comp = stbi__get8(s); + if (comp) + return stbi__err("bad comp method", "Corrupt PNG"); + filter = stbi__get8(s); + if (filter) + return stbi__err("bad filter method", "Corrupt PNG"); + interlace = stbi__get8(s); + if (interlace > 1) + return stbi__err("bad interlace method", "Corrupt PNG"); + if (!s->img_x || !s->img_y) + return stbi__err("0-pixel image", "Corrupt PNG"); + if (!pal_img_n) + { + s->img_n = (color & 2 ? 3 : 1) + (color & 4 ? 1 : 0); + if ((1 << 30) / s->img_x / s->img_n < s->img_y) + return stbi__err("too large", "Image too large to decode"); + } + else + { + // if paletted, then pal_n is our final components, and + // img_n is # components to decompress/filter. + s->img_n = 1; + if ((1 << 30) / s->img_x / 4 < s->img_y) + return stbi__err("too large", "Corrupt PNG"); + } + // even with SCAN_header, have to scan to see if we have a tRNS + break; + } + + case STBI__PNG_TYPE('P', 'L', 'T', 'E'): + { + if (first) + return stbi__err("first not IHDR", "Corrupt PNG"); + if (c.length > 256 * 3) + return stbi__err("invalid PLTE", "Corrupt PNG"); + pal_len = c.length / 3; + if (pal_len * 3 != c.length) + return stbi__err("invalid PLTE", "Corrupt PNG"); + for (i = 0; i < pal_len; ++i) + { + palette[i * 4 + 0] = stbi__get8(s); + palette[i * 4 + 1] = stbi__get8(s); + palette[i * 4 + 2] = stbi__get8(s); + palette[i * 4 + 3] = 255; + } + break; + } + + case STBI__PNG_TYPE('t', 'R', 'N', 'S'): + { + if (first) + return stbi__err("first not IHDR", "Corrupt PNG"); + if (z->idata) + return stbi__err("tRNS after IDAT", "Corrupt PNG"); + if (pal_img_n) + { + if (scan == STBI__SCAN_header) + { + s->img_n = 4; + return 1; + } + if (pal_len == 0) + return stbi__err("tRNS before PLTE", "Corrupt PNG"); + if (c.length > pal_len) + return stbi__err("bad tRNS len", "Corrupt PNG"); + pal_img_n = 4; + for (i = 0; i < c.length; ++i) + palette[i * 4 + 3] = stbi__get8(s); + } + else + { + if (!(s->img_n & 1)) + return stbi__err("tRNS with alpha", "Corrupt PNG"); + if (c.length != (stbi__uint32)s->img_n * 2) + return stbi__err("bad tRNS len", "Corrupt PNG"); + has_trans = 1; + // non-paletted with tRNS = constant alpha. if header-scanning, we can stop now. + if (scan == STBI__SCAN_header) + { + ++s->img_n; + return 1; + } + if (z->depth == 16) + { + for (k = 0; k < s->img_n && k < 3; ++k) // extra loop test to suppress false GCC warning + tc16[k] = (stbi__uint16)stbi__get16be(s); // copy the values as-is + } + else + { + for (k = 0; k < s->img_n && k < 3; ++k) + tc[k] = (stbi_uc)(stbi__get16be(s) & 255) * stbi__depth_scale_table[z->depth]; // non 8-bit images will be larger + } + } + break; + } + + case STBI__PNG_TYPE('I', 'D', 'A', 'T'): + { + if (first) + return stbi__err("first not IHDR", "Corrupt PNG"); + if (pal_img_n && !pal_len) + return stbi__err("no PLTE", "Corrupt PNG"); + if (scan == STBI__SCAN_header) + { + // header scan definitely stops at first IDAT + if (pal_img_n) + s->img_n = pal_img_n; + return 1; + } + if (c.length > (1u << 30)) + return stbi__err("IDAT size limit", "IDAT section larger than 2^30 bytes"); + if ((int)(ioff + c.length) < (int)ioff) + return 0; + if (ioff + c.length > idata_limit) + { + stbi__uint32 idata_limit_old = idata_limit; + stbi_uc* p; + if (idata_limit == 0) + idata_limit = c.length > 4096 ? c.length : 4096; + while (ioff + c.length > idata_limit) + idata_limit *= 2; + STBI_NOTUSED(idata_limit_old); + p = (stbi_uc*)STBI_REALLOC_SIZED(z->idata, idata_limit_old, idata_limit); + if (p == NULL) + return stbi__err("outofmem", "Out of memory"); + z->idata = p; + } + if (!stbi__getn(s, z->idata + ioff, c.length)) + return stbi__err("outofdata", "Corrupt PNG"); + ioff += c.length; + break; + } + + case STBI__PNG_TYPE('I', 'E', 'N', 'D'): + { + stbi__uint32 raw_len, bpl; + if (first) + return stbi__err("first not IHDR", "Corrupt PNG"); + if (scan != STBI__SCAN_load) + return 1; + if (z->idata == NULL) + return stbi__err("no IDAT", "Corrupt PNG"); + // initial guess for decoded data size to avoid unnecessary reallocs + bpl = (s->img_x * z->depth + 7) / 8; // bytes per line, per component + raw_len = bpl * s->img_y * s->img_n /* pixels */ + s->img_y /* filter mode per row */; + z->expanded = (stbi_uc*)stbi_zlib_decode_malloc_guesssize_headerflag((char*)z->idata, ioff, raw_len, (int*)&raw_len, !is_iphone); + if (z->expanded == NULL) + return 0; // zlib should set error + STBI_FREE(z->idata); + z->idata = NULL; + if ((req_comp == s->img_n + 1 && req_comp != 3 && !pal_img_n) || has_trans) + s->img_out_n = s->img_n + 1; + else + s->img_out_n = s->img_n; + if (!stbi__create_png_image(z, z->expanded, raw_len, s->img_out_n, z->depth, color, interlace)) + return 0; + if (has_trans) + { + if (z->depth == 16) + { + if (!stbi__compute_transparency16(z, tc16, s->img_out_n)) + return 0; + } + else + { + if (!stbi__compute_transparency(z, tc, s->img_out_n)) + return 0; + } + } + if (is_iphone && stbi__de_iphone_flag && s->img_out_n > 2) + stbi__de_iphone(z); + if (pal_img_n) + { + // pal_img_n == 3 or 4 + s->img_n = pal_img_n; // record the actual colors we had + s->img_out_n = pal_img_n; + if (req_comp >= 3) + s->img_out_n = req_comp; + if (!stbi__expand_png_palette(z, palette, pal_len, s->img_out_n)) + return 0; + } + else if (has_trans) + { + // non-paletted image with tRNS -> source image has (constant) alpha + ++s->img_n; + } + STBI_FREE(z->expanded); + z->expanded = NULL; + // end of PNG chunk, read and skip CRC + stbi__get32be(s); + return 1; + } + + default: + // if critical, fail + if (first) + return stbi__err("first not IHDR", "Corrupt PNG"); + if ((c.type & (1 << 29)) == 0) + { + #ifndef STBI_NO_FAILURE_STRINGS + // not threadsafe + static char invalid_chunk[] = "XXXX PNG chunk not known"; + invalid_chunk[0] = STBI__BYTECAST(c.type >> 24); + invalid_chunk[1] = STBI__BYTECAST(c.type >> 16); + invalid_chunk[2] = STBI__BYTECAST(c.type >> 8); + invalid_chunk[3] = STBI__BYTECAST(c.type >> 0); + #endif + return stbi__err(invalid_chunk, "PNG not supported: unknown PNG chunk type"); + } + stbi__skip(s, c.length); + break; + } + // end of PNG chunk, read and skip CRC + stbi__get32be(s); + } +} + +static void* stbi__do_png(stbi__png* p, int* x, int* y, int* n, int req_comp, stbi__result_info* ri) +{ + void* result = NULL; + if (req_comp < 0 || req_comp > 4) + return stbi__errpuc("bad req_comp", "Internal error"); + if (stbi__parse_png_file(p, STBI__SCAN_load, req_comp)) + { + if (p->depth <= 8) + ri->bits_per_channel = 8; + else if (p->depth == 16) + ri->bits_per_channel = 16; + else + return stbi__errpuc("bad bits_per_channel", "PNG not supported: unsupported color depth"); + result = p->out; + p->out = NULL; + if (req_comp && req_comp != p->s->img_out_n) + { + if (ri->bits_per_channel == 8) + result = stbi__convert_format((unsigned char*)result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y); + else + result = stbi__convert_format16((stbi__uint16*)result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y); + p->s->img_out_n = req_comp; + if (result == NULL) + return result; + } + *x = p->s->img_x; + *y = p->s->img_y; + if (n) + *n = p->s->img_n; + } + STBI_FREE(p->out); + p->out = NULL; + STBI_FREE(p->expanded); + p->expanded = NULL; + STBI_FREE(p->idata); + p->idata = NULL; + + return result; +} + +static void* stbi__png_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) +{ + stbi__png p; + p.s = s; + return stbi__do_png(&p, x, y, comp, req_comp, ri); +} + +static int stbi__png_test(stbi__context* s) +{ + int r; + r = stbi__check_png_header(s); + stbi__rewind(s); + return r; +} + +static int stbi__png_info_raw(stbi__png* p, int* x, int* y, int* comp) +{ + if (!stbi__parse_png_file(p, STBI__SCAN_header, 0)) + { + stbi__rewind(p->s); + return 0; + } + if (x) + *x = p->s->img_x; + if (y) + *y = p->s->img_y; + if (comp) + *comp = p->s->img_n; + return 1; +} + +static int stbi__png_info(stbi__context* s, int* x, int* y, int* comp) +{ + stbi__png p; + p.s = s; + return stbi__png_info_raw(&p, x, y, comp); +} + +static int stbi__png_is16(stbi__context* s) +{ + stbi__png p; + p.s = s; + if (!stbi__png_info_raw(&p, NULL, NULL, NULL)) + return 0; + if (p.depth != 16) + { + stbi__rewind(p.s); + return 0; + } + return 1; +} + #endif + +// Microsoft/Windows BMP image + + #ifndef STBI_NO_BMP +static int stbi__bmp_test_raw(stbi__context* s) +{ + int r; + int sz; + if (stbi__get8(s) != 'B') + return 0; + if (stbi__get8(s) != 'M') + return 0; + stbi__get32le(s); // discard filesize + stbi__get16le(s); // discard reserved + stbi__get16le(s); // discard reserved + stbi__get32le(s); // discard data offset + sz = stbi__get32le(s); + r = (sz == 12 || sz == 40 || sz == 56 || sz == 108 || sz == 124); + return r; +} + +static int stbi__bmp_test(stbi__context* s) +{ + int r = stbi__bmp_test_raw(s); + stbi__rewind(s); + return r; +} + +// returns 0..31 for the highest set bit +static int stbi__high_bit(unsigned int z) +{ + int n = 0; + if (z == 0) + return -1; + if (z >= 0x10000) + { + n += 16; + z >>= 16; + } + if (z >= 0x00100) + { + n += 8; + z >>= 8; + } + if (z >= 0x00010) + { + n += 4; + z >>= 4; + } + if (z >= 0x00004) + { + n += 2; + z >>= 2; + } + if (z >= 0x00002) + { + n += 1; /* >>= 1;*/ + } + return n; +} + +static int stbi__bitcount(unsigned int a) +{ + a = (a & 0x55555555) + ((a >> 1) & 0x55555555); // max 2 + a = (a & 0x33333333) + ((a >> 2) & 0x33333333); // max 4 + a = (a + (a >> 4)) & 0x0f0f0f0f; // max 8 per 4, now 8 bits + a = (a + (a >> 8)); // max 16 per 8 bits + a = (a + (a >> 16)); // max 32 per 8 bits + return a & 0xff; +} + +// extract an arbitrarily-aligned N-bit value (N=bits) +// from v, and then make it 8-bits long and fractionally +// extend it to full full range. +static int stbi__shiftsigned(unsigned int v, int shift, int bits) +{ + static unsigned int mul_table[9] = { + 0, + 0xff /*0b11111111*/, + 0x55 /*0b01010101*/, + 0x49 /*0b01001001*/, + 0x11 /*0b00010001*/, + 0x21 /*0b00100001*/, + 0x41 /*0b01000001*/, + 0x81 /*0b10000001*/, + 0x01 /*0b00000001*/, + }; + static unsigned int shift_table[9] = { + 0, + 0, + 0, + 1, + 0, + 2, + 4, + 6, + 0, + }; + if (shift < 0) + v <<= -shift; + else + v >>= shift; + STBI_ASSERT(v < 256); + v >>= (8 - bits); + STBI_ASSERT(bits >= 0 && bits <= 8); + return (int)((unsigned)v * mul_table[bits]) >> shift_table[bits]; +} + +typedef struct { + int bpp, offset, hsz; + unsigned int mr, mg, mb, ma, all_a; + int extra_read; +} stbi__bmp_data; + +static int stbi__bmp_set_mask_defaults(stbi__bmp_data* info, int compress) +{ + // BI_BITFIELDS specifies masks explicitly, don't override + if (compress == 3) + return 1; + + if (compress == 0) + { + if (info->bpp == 16) + { + info->mr = 31u << 10; + info->mg = 31u << 5; + info->mb = 31u << 0; + } + else if (info->bpp == 32) + { + info->mr = 0xffu << 16; + info->mg = 0xffu << 8; + info->mb = 0xffu << 0; + info->ma = 0xffu << 24; + info->all_a = 0; // if all_a is 0 at end, then we loaded alpha channel but it was all 0 + } + else + { + // otherwise, use defaults, which is all-0 + info->mr = info->mg = info->mb = info->ma = 0; + } + return 1; + } + return 0; // error +} + +static void* stbi__bmp_parse_header(stbi__context* s, stbi__bmp_data* info) +{ + int hsz; + if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') + return stbi__errpuc("not BMP", "Corrupt BMP"); + stbi__get32le(s); // discard filesize + stbi__get16le(s); // discard reserved + stbi__get16le(s); // discard reserved + info->offset = stbi__get32le(s); + info->hsz = hsz = stbi__get32le(s); + info->mr = info->mg = info->mb = info->ma = 0; + info->extra_read = 14; + + if (info->offset < 0) + return stbi__errpuc("bad BMP", "bad BMP"); + + if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) + return stbi__errpuc("unknown BMP", "BMP type not supported: unknown"); + if (hsz == 12) + { + s->img_x = stbi__get16le(s); + s->img_y = stbi__get16le(s); + } + else + { + s->img_x = stbi__get32le(s); + s->img_y = stbi__get32le(s); + } + if (stbi__get16le(s) != 1) + return stbi__errpuc("bad BMP", "bad BMP"); + info->bpp = stbi__get16le(s); + if (hsz != 12) + { + int compress = stbi__get32le(s); + if (compress == 1 || compress == 2) + return stbi__errpuc("BMP RLE", "BMP type not supported: RLE"); + if (compress >= 4) + return stbi__errpuc("BMP JPEG/PNG", "BMP type not supported: unsupported compression"); // this includes PNG/JPEG modes + if (compress == 3 && info->bpp != 16 && info->bpp != 32) + return stbi__errpuc("bad BMP", "bad BMP"); // bitfields requires 16 or 32 bits/pixel + stbi__get32le(s); // discard sizeof + stbi__get32le(s); // discard hres + stbi__get32le(s); // discard vres + stbi__get32le(s); // discard colorsused + stbi__get32le(s); // discard max important + if (hsz == 40 || hsz == 56) + { + if (hsz == 56) + { + stbi__get32le(s); + stbi__get32le(s); + stbi__get32le(s); + stbi__get32le(s); + } + if (info->bpp == 16 || info->bpp == 32) + { + if (compress == 0) + { + stbi__bmp_set_mask_defaults(info, compress); + } + else if (compress == 3) + { + info->mr = stbi__get32le(s); + info->mg = stbi__get32le(s); + info->mb = stbi__get32le(s); + info->extra_read += 12; + // not documented, but generated by photoshop and handled by mspaint + if (info->mr == info->mg && info->mg == info->mb) + { + // ?!?!? + return stbi__errpuc("bad BMP", "bad BMP"); + } + } + else + return stbi__errpuc("bad BMP", "bad BMP"); + } + } + else + { + // V4/V5 header + int i; + if (hsz != 108 && hsz != 124) + return stbi__errpuc("bad BMP", "bad BMP"); + info->mr = stbi__get32le(s); + info->mg = stbi__get32le(s); + info->mb = stbi__get32le(s); + info->ma = stbi__get32le(s); + if (compress != 3) // override mr/mg/mb unless in BI_BITFIELDS mode, as per docs + stbi__bmp_set_mask_defaults(info, compress); + stbi__get32le(s); // discard color space + for (i = 0; i < 12; ++i) + stbi__get32le(s); // discard color space parameters + if (hsz == 124) + { + stbi__get32le(s); // discard rendering intent + stbi__get32le(s); // discard offset of profile data + stbi__get32le(s); // discard size of profile data + stbi__get32le(s); // discard reserved + } + } + } + return (void*)1; +} + +static void* stbi__bmp_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) +{ + stbi_uc* out; + unsigned int mr = 0, mg = 0, mb = 0, ma = 0, all_a; + stbi_uc pal[256][4]; + int psize = 0, i, j, width; + int flip_vertically, pad, target; + stbi__bmp_data info; + STBI_NOTUSED(ri); + + info.all_a = 255; + if (stbi__bmp_parse_header(s, &info) == NULL) + return NULL; // error code already set + + flip_vertically = ((int)s->img_y) > 0; + s->img_y = abs((int)s->img_y); + + if (s->img_y > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + + mr = info.mr; + mg = info.mg; + mb = info.mb; + ma = info.ma; + all_a = info.all_a; + + if (info.hsz == 12) + { + if (info.bpp < 24) + psize = (info.offset - info.extra_read - 24) / 3; + } + else + { + if (info.bpp < 16) + psize = (info.offset - info.extra_read - info.hsz) >> 2; + } + if (psize == 0) + { + // accept some number of extra bytes after the header, but if the offset points either to before + // the header ends or implies a large amount of extra data, reject the file as malformed + int bytes_read_so_far = s->callback_already_read + (int)(s->img_buffer - s->img_buffer_original); + int header_limit = 1024; // max we actually read is below 256 bytes currently. + int extra_data_limit = 256 * 4; // what ordinarily goes here is a palette; 256 entries*4 bytes is its max size. + if (bytes_read_so_far <= 0 || bytes_read_so_far > header_limit) + { + return stbi__errpuc("bad header", "Corrupt BMP"); + } + // we established that bytes_read_so_far is positive and sensible. + // the first half of this test rejects offsets that are either too small positives, or + // negative, and guarantees that info.offset >= bytes_read_so_far > 0. this in turn + // ensures the number computed in the second half of the test can't overflow. + if (info.offset < bytes_read_so_far || info.offset - bytes_read_so_far > extra_data_limit) + { + return stbi__errpuc("bad offset", "Corrupt BMP"); + } + else + { + stbi__skip(s, info.offset - bytes_read_so_far); + } + } + + if (info.bpp == 24 && ma == 0xff000000) + s->img_n = 3; + else + s->img_n = ma ? 4 : 3; + if (req_comp && req_comp >= 3) // we can directly decode 3 or 4 + target = req_comp; + else + target = s->img_n; // if they want monochrome, we'll post-convert + + // sanity-check size + if (!stbi__mad3sizes_valid(target, s->img_x, s->img_y, 0)) + return stbi__errpuc("too large", "Corrupt BMP"); + + out = (stbi_uc*)stbi__malloc_mad3(target, s->img_x, s->img_y, 0); + if (!out) + return stbi__errpuc("outofmem", "Out of memory"); + if (info.bpp < 16) + { + int z = 0; + if (psize == 0 || psize > 256) + { + STBI_FREE(out); + return stbi__errpuc("invalid", "Corrupt BMP"); + } + for (i = 0; i < psize; ++i) + { + pal[i][2] = stbi__get8(s); + pal[i][1] = stbi__get8(s); + pal[i][0] = stbi__get8(s); + if (info.hsz != 12) + stbi__get8(s); + pal[i][3] = 255; + } + stbi__skip(s, info.offset - info.extra_read - info.hsz - psize * (info.hsz == 12 ? 3 : 4)); + if (info.bpp == 1) + width = (s->img_x + 7) >> 3; + else if (info.bpp == 4) + width = (s->img_x + 1) >> 1; + else if (info.bpp == 8) + width = s->img_x; + else + { + STBI_FREE(out); + return stbi__errpuc("bad bpp", "Corrupt BMP"); + } + pad = (-width) & 3; + if (info.bpp == 1) + { + for (j = 0; j < (int)s->img_y; ++j) + { + int bit_offset = 7, v = stbi__get8(s); + for (i = 0; i < (int)s->img_x; ++i) + { + int color = (v >> bit_offset) & 0x1; + out[z++] = pal[color][0]; + out[z++] = pal[color][1]; + out[z++] = pal[color][2]; + if (target == 4) + out[z++] = 255; + if (i + 1 == (int)s->img_x) + break; + if ((--bit_offset) < 0) + { + bit_offset = 7; + v = stbi__get8(s); + } + } + stbi__skip(s, pad); + } + } + else + { + for (j = 0; j < (int)s->img_y; ++j) + { + for (i = 0; i < (int)s->img_x; i += 2) + { + int v = stbi__get8(s), v2 = 0; + if (info.bpp == 4) + { + v2 = v & 15; + v >>= 4; + } + out[z++] = pal[v][0]; + out[z++] = pal[v][1]; + out[z++] = pal[v][2]; + if (target == 4) + out[z++] = 255; + if (i + 1 == (int)s->img_x) + break; + v = (info.bpp == 8) ? stbi__get8(s) : v2; + out[z++] = pal[v][0]; + out[z++] = pal[v][1]; + out[z++] = pal[v][2]; + if (target == 4) + out[z++] = 255; + } + stbi__skip(s, pad); + } + } + } + else + { + int rshift = 0, gshift = 0, bshift = 0, ashift = 0, rcount = 0, gcount = 0, bcount = 0, acount = 0; + int z = 0; + int easy = 0; + stbi__skip(s, info.offset - info.extra_read - info.hsz); + if (info.bpp == 24) + width = 3 * s->img_x; + else if (info.bpp == 16) + width = 2 * s->img_x; + else /* bpp = 32 and pad = 0 */ + width = 0; + pad = (-width) & 3; + if (info.bpp == 24) + { + easy = 1; + } + else if (info.bpp == 32) + { + if (mb == 0xff && mg == 0xff00 && mr == 0x00ff0000 && ma == 0xff000000) + easy = 2; + } + if (!easy) + { + if (!mr || !mg || !mb) + { + STBI_FREE(out); + return stbi__errpuc("bad masks", "Corrupt BMP"); + } + // right shift amt to put high bit in position #7 + rshift = stbi__high_bit(mr) - 7; + rcount = stbi__bitcount(mr); + gshift = stbi__high_bit(mg) - 7; + gcount = stbi__bitcount(mg); + bshift = stbi__high_bit(mb) - 7; + bcount = stbi__bitcount(mb); + ashift = stbi__high_bit(ma) - 7; + acount = stbi__bitcount(ma); + if (rcount > 8 || gcount > 8 || bcount > 8 || acount > 8) + { + STBI_FREE(out); + return stbi__errpuc("bad masks", "Corrupt BMP"); + } + } + for (j = 0; j < (int)s->img_y; ++j) + { + if (easy) + { + for (i = 0; i < (int)s->img_x; ++i) + { + unsigned char a; + out[z + 2] = stbi__get8(s); + out[z + 1] = stbi__get8(s); + out[z + 0] = stbi__get8(s); + z += 3; + a = (easy == 2 ? stbi__get8(s) : 255); + all_a |= a; + if (target == 4) + out[z++] = a; + } + } + else + { + int bpp = info.bpp; + for (i = 0; i < (int)s->img_x; ++i) + { + stbi__uint32 v = (bpp == 16 ? (stbi__uint32)stbi__get16le(s) : stbi__get32le(s)); + unsigned int a; + out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mr, rshift, rcount)); + out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mg, gshift, gcount)); + out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mb, bshift, bcount)); + a = (ma ? stbi__shiftsigned(v & ma, ashift, acount) : 255); + all_a |= a; + if (target == 4) + out[z++] = STBI__BYTECAST(a); + } + } + stbi__skip(s, pad); + } + } + + // if alpha channel is all 0s, replace with all 255s + if (target == 4 && all_a == 0) + for (i = 4 * s->img_x * s->img_y - 1; i >= 0; i -= 4) + out[i] = 255; + + if (flip_vertically) + { + stbi_uc t; + for (j = 0; j < (int)s->img_y >> 1; ++j) + { + stbi_uc* p1 = out + j * s->img_x * target; + stbi_uc* p2 = out + (s->img_y - 1 - j) * s->img_x * target; + for (i = 0; i < (int)s->img_x * target; ++i) + { + t = p1[i]; + p1[i] = p2[i]; + p2[i] = t; + } + } + } + + if (req_comp && req_comp != target) + { + out = stbi__convert_format(out, target, req_comp, s->img_x, s->img_y); + if (out == NULL) + return out; // stbi__convert_format frees input on failure + } + + *x = s->img_x; + *y = s->img_y; + if (comp) + *comp = s->img_n; + return out; +} + #endif + + // Targa Truevision - TGA + // by Jonathan Dummer + #ifndef STBI_NO_TGA +// returns STBI_rgb or whatever, 0 on error +static int stbi__tga_get_comp(int bits_per_pixel, int is_grey, int* is_rgb16) +{ + // only RGB or RGBA (incl. 16bit) or grey allowed + if (is_rgb16) + *is_rgb16 = 0; + switch (bits_per_pixel) + { + case 8: return STBI_grey; + case 16: + if (is_grey) + return STBI_grey_alpha; + // fallthrough + case 15: + if (is_rgb16) + *is_rgb16 = 1; + return STBI_rgb; + case 24: // fallthrough + case 32: return bits_per_pixel / 8; + default: return 0; + } +} + +static int stbi__tga_info(stbi__context* s, int* x, int* y, int* comp) +{ + int tga_w, tga_h, tga_comp, tga_image_type, tga_bits_per_pixel, tga_colormap_bpp; + int sz, tga_colormap_type; + stbi__get8(s); // discard Offset + tga_colormap_type = stbi__get8(s); // colormap type + if (tga_colormap_type > 1) + { + stbi__rewind(s); + return 0; // only RGB or indexed allowed + } + tga_image_type = stbi__get8(s); // image type + if (tga_colormap_type == 1) + { // colormapped (paletted) image + if (tga_image_type != 1 && tga_image_type != 9) + { + stbi__rewind(s); + return 0; + } + stbi__skip(s, 4); // skip index of first colormap entry and number of entries + sz = stbi__get8(s); // check bits per palette color entry + if ((sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32)) + { + stbi__rewind(s); + return 0; + } + stbi__skip(s, 4); // skip image x and y origin + tga_colormap_bpp = sz; + } + else + { // "normal" image w/o colormap - only RGB or grey allowed, +/- RLE + if ((tga_image_type != 2) && (tga_image_type != 3) && (tga_image_type != 10) && (tga_image_type != 11)) + { + stbi__rewind(s); + return 0; // only RGB or grey allowed, +/- RLE + } + stbi__skip(s, 9); // skip colormap specification and image x/y origin + tga_colormap_bpp = 0; + } + tga_w = stbi__get16le(s); + if (tga_w < 1) + { + stbi__rewind(s); + return 0; // test width + } + tga_h = stbi__get16le(s); + if (tga_h < 1) + { + stbi__rewind(s); + return 0; // test height + } + tga_bits_per_pixel = stbi__get8(s); // bits per pixel + stbi__get8(s); // ignore alpha bits + if (tga_colormap_bpp != 0) + { + if ((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16)) + { + // when using a colormap, tga_bits_per_pixel is the size of the indexes + // I don't think anything but 8 or 16bit indexes makes sense + stbi__rewind(s); + return 0; + } + tga_comp = stbi__tga_get_comp(tga_colormap_bpp, 0, NULL); + } + else + { + tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3) || (tga_image_type == 11), NULL); + } + if (!tga_comp) + { + stbi__rewind(s); + return 0; + } + if (x) + *x = tga_w; + if (y) + *y = tga_h; + if (comp) + *comp = tga_comp; + return 1; // seems to have passed everything +} + +static int stbi__tga_test(stbi__context* s) +{ + int res = 0; + int sz, tga_color_type; + stbi__get8(s); // discard Offset + tga_color_type = stbi__get8(s); // color type + if (tga_color_type > 1) + goto errorEnd; // only RGB or indexed allowed + sz = stbi__get8(s); // image type + if (tga_color_type == 1) + { // colormapped (paletted) image + if (sz != 1 && sz != 9) + goto errorEnd; // colortype 1 demands image type 1 or 9 + stbi__skip(s, 4); // skip index of first colormap entry and number of entries + sz = stbi__get8(s); // check bits per palette color entry + if ((sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32)) + goto errorEnd; + stbi__skip(s, 4); // skip image x and y origin + } + else + { // "normal" image w/o colormap + if ((sz != 2) && (sz != 3) && (sz != 10) && (sz != 11)) + goto errorEnd; // only RGB or grey allowed, +/- RLE + stbi__skip(s, 9); // skip colormap specification and image x/y origin + } + if (stbi__get16le(s) < 1) + goto errorEnd; // test width + if (stbi__get16le(s) < 1) + goto errorEnd; // test height + sz = stbi__get8(s); // bits per pixel + if ((tga_color_type == 1) && (sz != 8) && (sz != 16)) + goto errorEnd; // for colormapped images, bpp is size of an index + if ((sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32)) + goto errorEnd; + + res = 1; // if we got this far, everything's good and we can return 1 instead of 0 + +errorEnd: + stbi__rewind(s); + return res; +} + +// read 16bit value and convert to 24bit RGB +static void stbi__tga_read_rgb16(stbi__context* s, stbi_uc* out) +{ + stbi__uint16 px = (stbi__uint16)stbi__get16le(s); + stbi__uint16 fiveBitMask = 31; + // we have 3 channels with 5bits each + int r = (px >> 10) & fiveBitMask; + int g = (px >> 5) & fiveBitMask; + int b = px & fiveBitMask; + // Note that this saves the data in RGB(A) order, so it doesn't need to be swapped later + out[0] = (stbi_uc)((r * 255) / 31); + out[1] = (stbi_uc)((g * 255) / 31); + out[2] = (stbi_uc)((b * 255) / 31); + + // some people claim that the most significant bit might be used for alpha + // (possibly if an alpha-bit is set in the "image descriptor byte") + // but that only made 16bit test images completely translucent.. + // so let's treat all 15 and 16bit TGAs as RGB with no alpha. +} + +static void* stbi__tga_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) +{ + // read in the TGA header stuff + int tga_offset = stbi__get8(s); + int tga_indexed = stbi__get8(s); + int tga_image_type = stbi__get8(s); + int tga_is_RLE = 0; + int tga_palette_start = stbi__get16le(s); + int tga_palette_len = stbi__get16le(s); + int tga_palette_bits = stbi__get8(s); + int tga_x_origin = stbi__get16le(s); + int tga_y_origin = stbi__get16le(s); + int tga_width = stbi__get16le(s); + int tga_height = stbi__get16le(s); + int tga_bits_per_pixel = stbi__get8(s); + int tga_comp, tga_rgb16 = 0; + int tga_inverted = stbi__get8(s); + // int tga_alpha_bits = tga_inverted & 15; // the 4 lowest bits - unused (useless?) + // image data + unsigned char* tga_data; + unsigned char* tga_palette = NULL; + int i, j; + unsigned char raw_data[4] = {0}; + int RLE_count = 0; + int RLE_repeating = 0; + int read_next_pixel = 1; + STBI_NOTUSED(ri); + STBI_NOTUSED(tga_x_origin); // @TODO + STBI_NOTUSED(tga_y_origin); // @TODO + + if (tga_height > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + if (tga_width > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + + // do a tiny bit of precessing + if (tga_image_type >= 8) + { + tga_image_type -= 8; + tga_is_RLE = 1; + } + tga_inverted = 1 - ((tga_inverted >> 5) & 1); + + // If I'm paletted, then I'll use the number of bits from the palette + if (tga_indexed) + tga_comp = stbi__tga_get_comp(tga_palette_bits, 0, &tga_rgb16); + else + tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3), &tga_rgb16); + + if (!tga_comp) // shouldn't really happen, stbi__tga_test() should have ensured basic consistency + return stbi__errpuc("bad format", "Can't find out TGA pixelformat"); + + // tga info + *x = tga_width; + *y = tga_height; + if (comp) + *comp = tga_comp; + + if (!stbi__mad3sizes_valid(tga_width, tga_height, tga_comp, 0)) + return stbi__errpuc("too large", "Corrupt TGA"); + + tga_data = (unsigned char*)stbi__malloc_mad3(tga_width, tga_height, tga_comp, 0); + if (!tga_data) + return stbi__errpuc("outofmem", "Out of memory"); + + // skip to the data's starting position (offset usually = 0) + stbi__skip(s, tga_offset); + + if (!tga_indexed && !tga_is_RLE && !tga_rgb16) + { + for (i = 0; i < tga_height; ++i) + { + int row = tga_inverted ? tga_height - i - 1 : i; + stbi_uc* tga_row = tga_data + row * tga_width * tga_comp; + stbi__getn(s, tga_row, tga_width * tga_comp); + } + } + else + { + // do I need to load a palette? + if (tga_indexed) + { + if (tga_palette_len == 0) + { /* you have to have at least one entry! */ + STBI_FREE(tga_data); + return stbi__errpuc("bad palette", "Corrupt TGA"); + } + + // any data to skip? (offset usually = 0) + stbi__skip(s, tga_palette_start); + // load the palette + tga_palette = (unsigned char*)stbi__malloc_mad2(tga_palette_len, tga_comp, 0); + if (!tga_palette) + { + STBI_FREE(tga_data); + return stbi__errpuc("outofmem", "Out of memory"); + } + if (tga_rgb16) + { + stbi_uc* pal_entry = tga_palette; + STBI_ASSERT(tga_comp == STBI_rgb); + for (i = 0; i < tga_palette_len; ++i) + { + stbi__tga_read_rgb16(s, pal_entry); + pal_entry += tga_comp; + } + } + else if (!stbi__getn(s, tga_palette, tga_palette_len * tga_comp)) + { + STBI_FREE(tga_data); + STBI_FREE(tga_palette); + return stbi__errpuc("bad palette", "Corrupt TGA"); + } + } + // load the data + for (i = 0; i < tga_width * tga_height; ++i) + { + // if I'm in RLE mode, do I need to get a RLE stbi__pngchunk? + if (tga_is_RLE) + { + if (RLE_count == 0) + { + // yep, get the next byte as a RLE command + int RLE_cmd = stbi__get8(s); + RLE_count = 1 + (RLE_cmd & 127); + RLE_repeating = RLE_cmd >> 7; + read_next_pixel = 1; + } + else if (!RLE_repeating) + { + read_next_pixel = 1; + } + } + else + { + read_next_pixel = 1; + } + // OK, if I need to read a pixel, do it now + if (read_next_pixel) + { + // load however much data we did have + if (tga_indexed) + { + // read in index, then perform the lookup + int pal_idx = (tga_bits_per_pixel == 8) ? stbi__get8(s) : stbi__get16le(s); + if (pal_idx >= tga_palette_len) + { + // invalid index + pal_idx = 0; + } + pal_idx *= tga_comp; + for (j = 0; j < tga_comp; ++j) + { + raw_data[j] = tga_palette[pal_idx + j]; + } + } + else if (tga_rgb16) + { + STBI_ASSERT(tga_comp == STBI_rgb); + stbi__tga_read_rgb16(s, raw_data); + } + else + { + // read in the data raw + for (j = 0; j < tga_comp; ++j) + { + raw_data[j] = stbi__get8(s); + } + } + // clear the reading flag for the next pixel + read_next_pixel = 0; + } // end of reading a pixel + + // copy data + for (j = 0; j < tga_comp; ++j) + tga_data[i * tga_comp + j] = raw_data[j]; + + // in case we're in RLE mode, keep counting down + --RLE_count; + } + // do I need to invert the image? + if (tga_inverted) + { + for (j = 0; j * 2 < tga_height; ++j) + { + int index1 = j * tga_width * tga_comp; + int index2 = (tga_height - 1 - j) * tga_width * tga_comp; + for (i = tga_width * tga_comp; i > 0; --i) + { + unsigned char temp = tga_data[index1]; + tga_data[index1] = tga_data[index2]; + tga_data[index2] = temp; + ++index1; + ++index2; + } + } + } + // clear my palette, if I had one + if (tga_palette != NULL) + { + STBI_FREE(tga_palette); + } + } + + // swap RGB - if the source data was RGB16, it already is in the right order + if (tga_comp >= 3 && !tga_rgb16) + { + unsigned char* tga_pixel = tga_data; + for (i = 0; i < tga_width * tga_height; ++i) + { + unsigned char temp = tga_pixel[0]; + tga_pixel[0] = tga_pixel[2]; + tga_pixel[2] = temp; + tga_pixel += tga_comp; + } + } + + // convert to target component count + if (req_comp && req_comp != tga_comp) + tga_data = stbi__convert_format(tga_data, tga_comp, req_comp, tga_width, tga_height); + + // the things I do to get rid of an error message, and yet keep + // Microsoft's C compilers happy... [8^( + tga_palette_start = tga_palette_len = tga_palette_bits = tga_x_origin = tga_y_origin = 0; + STBI_NOTUSED(tga_palette_start); + // OK, done + return tga_data; +} + #endif + +// ************************************************************************************************* +// Photoshop PSD loader -- PD by Thatcher Ulrich, integration by Nicolas Schulz, tweaked by STB + + #ifndef STBI_NO_PSD +static int stbi__psd_test(stbi__context* s) +{ + int r = (stbi__get32be(s) == 0x38425053); + stbi__rewind(s); + return r; +} + +static int stbi__psd_decode_rle(stbi__context* s, stbi_uc* p, int pixelCount) +{ + int count, nleft, len; + + count = 0; + while ((nleft = pixelCount - count) > 0) + { + len = stbi__get8(s); + if (len == 128) + { + // No-op. + } + else if (len < 128) + { + // Copy next len+1 bytes literally. + len++; + if (len > nleft) + return 0; // corrupt data + count += len; + while (len) + { + *p = stbi__get8(s); + p += 4; + len--; + } + } + else if (len > 128) + { + stbi_uc val; + // Next -len+1 bytes in the dest are replicated from next source byte. + // (Interpret len as a negative 8-bit int.) + len = 257 - len; + if (len > nleft) + return 0; // corrupt data + val = stbi__get8(s); + count += len; + while (len) + { + *p = val; + p += 4; + len--; + } + } + } + + return 1; +} + +static void* stbi__psd_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri, int bpc) +{ + int pixelCount; + int channelCount, compression; + int channel, i; + int bitdepth; + int w, h; + stbi_uc* out; + STBI_NOTUSED(ri); + + // Check identifier + if (stbi__get32be(s) != 0x38425053) // "8BPS" + return stbi__errpuc("not PSD", "Corrupt PSD image"); + + // Check file type version. + if (stbi__get16be(s) != 1) + return stbi__errpuc("wrong version", "Unsupported version of PSD image"); + + // Skip 6 reserved bytes. + stbi__skip(s, 6); + + // Read the number of channels (R, G, B, A, etc). + channelCount = stbi__get16be(s); + if (channelCount < 0 || channelCount > 16) + return stbi__errpuc("wrong channel count", "Unsupported number of channels in PSD image"); + + // Read the rows and columns of the image. + h = stbi__get32be(s); + w = stbi__get32be(s); + + if (h > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + if (w > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + + // Make sure the depth is 8 bits. + bitdepth = stbi__get16be(s); + if (bitdepth != 8 && bitdepth != 16) + return stbi__errpuc("unsupported bit depth", "PSD bit depth is not 8 or 16 bit"); + + // Make sure the color mode is RGB. + // Valid options are: + // 0: Bitmap + // 1: Grayscale + // 2: Indexed color + // 3: RGB color + // 4: CMYK color + // 7: Multichannel + // 8: Duotone + // 9: Lab color + if (stbi__get16be(s) != 3) + return stbi__errpuc("wrong color format", "PSD is not in RGB color format"); + + // Skip the Mode Data. (It's the palette for indexed color; other info for other modes.) + stbi__skip(s, stbi__get32be(s)); + + // Skip the image resources. (resolution, pen tool paths, etc) + stbi__skip(s, stbi__get32be(s)); + + // Skip the reserved data. + stbi__skip(s, stbi__get32be(s)); + + // Find out if the data is compressed. + // Known values: + // 0: no compression + // 1: RLE compressed + compression = stbi__get16be(s); + if (compression > 1) + return stbi__errpuc("bad compression", "PSD has an unknown compression format"); + + // Check size + if (!stbi__mad3sizes_valid(4, w, h, 0)) + return stbi__errpuc("too large", "Corrupt PSD"); + + // Create the destination image. + + if (!compression && bitdepth == 16 && bpc == 16) + { + out = (stbi_uc*)stbi__malloc_mad3(8, w, h, 0); + ri->bits_per_channel = 16; + } + else + out = (stbi_uc*)stbi__malloc(4 * w * h); + + if (!out) + return stbi__errpuc("outofmem", "Out of memory"); + pixelCount = w * h; + + // Initialize the data to zero. + // memset( out, 0, pixelCount * 4 ); + + // Finally, the image data. + if (compression) + { + // RLE as used by .PSD and .TIFF + // Loop until you get the number of unpacked bytes you are expecting: + // Read the next source byte into n. + // If n is between 0 and 127 inclusive, copy the next n+1 bytes literally. + // Else if n is between -127 and -1 inclusive, copy the next byte -n+1 times. + // Else if n is 128, noop. + // Endloop + + // The RLE-compressed data is preceded by a 2-byte data count for each row in the data, + // which we're going to just skip. + stbi__skip(s, h * channelCount * 2); + + // Read the RLE data by channel. + for (channel = 0; channel < 4; channel++) + { + stbi_uc* p; + + p = out + channel; + if (channel >= channelCount) + { + // Fill this channel with default data. + for (i = 0; i < pixelCount; i++, p += 4) + *p = (channel == 3 ? 255 : 0); + } + else + { + // Read the RLE data. + if (!stbi__psd_decode_rle(s, p, pixelCount)) + { + STBI_FREE(out); + return stbi__errpuc("corrupt", "bad RLE data"); + } + } + } + } + else + { + // We're at the raw image data. It's each channel in order (Red, Green, Blue, Alpha, ...) + // where each channel consists of an 8-bit (or 16-bit) value for each pixel in the image. + + // Read the data by channel. + for (channel = 0; channel < 4; channel++) + { + if (channel >= channelCount) + { + // Fill this channel with default data. + if (bitdepth == 16 && bpc == 16) + { + stbi__uint16* q = ((stbi__uint16*)out) + channel; + stbi__uint16 val = channel == 3 ? 65535 : 0; + for (i = 0; i < pixelCount; i++, q += 4) + *q = val; + } + else + { + stbi_uc* p = out + channel; + stbi_uc val = channel == 3 ? 255 : 0; + for (i = 0; i < pixelCount; i++, p += 4) + *p = val; + } + } + else + { + if (ri->bits_per_channel == 16) + { // output bpc + stbi__uint16* q = ((stbi__uint16*)out) + channel; + for (i = 0; i < pixelCount; i++, q += 4) + *q = (stbi__uint16)stbi__get16be(s); + } + else + { + stbi_uc* p = out + channel; + if (bitdepth == 16) + { // input bpc + for (i = 0; i < pixelCount; i++, p += 4) + *p = (stbi_uc)(stbi__get16be(s) >> 8); + } + else + { + for (i = 0; i < pixelCount; i++, p += 4) + *p = stbi__get8(s); + } + } + } + } + } + + // remove weird white matte from PSD + if (channelCount >= 4) + { + if (ri->bits_per_channel == 16) + { + for (i = 0; i < w * h; ++i) + { + stbi__uint16* pixel = (stbi__uint16*)out + 4 * i; + if (pixel[3] != 0 && pixel[3] != 65535) + { + float a = pixel[3] / 65535.0f; + float ra = 1.0f / a; + float inv_a = 65535.0f * (1 - ra); + pixel[0] = (stbi__uint16)(pixel[0] * ra + inv_a); + pixel[1] = (stbi__uint16)(pixel[1] * ra + inv_a); + pixel[2] = (stbi__uint16)(pixel[2] * ra + inv_a); + } + } + } + else + { + for (i = 0; i < w * h; ++i) + { + unsigned char* pixel = out + 4 * i; + if (pixel[3] != 0 && pixel[3] != 255) + { + float a = pixel[3] / 255.0f; + float ra = 1.0f / a; + float inv_a = 255.0f * (1 - ra); + pixel[0] = (unsigned char)(pixel[0] * ra + inv_a); + pixel[1] = (unsigned char)(pixel[1] * ra + inv_a); + pixel[2] = (unsigned char)(pixel[2] * ra + inv_a); + } + } + } + } + + // convert to desired output format + if (req_comp && req_comp != 4) + { + if (ri->bits_per_channel == 16) + out = (stbi_uc*)stbi__convert_format16((stbi__uint16*)out, 4, req_comp, w, h); + else + out = stbi__convert_format(out, 4, req_comp, w, h); + if (out == NULL) + return out; // stbi__convert_format frees input on failure + } + + if (comp) + *comp = 4; + *y = h; + *x = w; + + return out; +} + #endif + +// ************************************************************************************************* +// Softimage PIC loader +// by Tom Seddon +// +// See http://softimage.wiki.softimage.com/index.php/INFO:_PIC_file_format +// See http://ozviz.wasp.uwa.edu.au/~pbourke/dataformats/softimagepic/ + + #ifndef STBI_NO_PIC +static int stbi__pic_is4(stbi__context* s, const char* str) +{ + int i; + for (i = 0; i < 4; ++i) + if (stbi__get8(s) != (stbi_uc)str[i]) + return 0; + + return 1; +} + +static int stbi__pic_test_core(stbi__context* s) +{ + int i; + + if (!stbi__pic_is4(s, "\x53\x80\xF6\x34")) + return 0; + + for (i = 0; i < 84; ++i) + stbi__get8(s); + + if (!stbi__pic_is4(s, "PICT")) + return 0; + + return 1; +} + +typedef struct { + stbi_uc size, type, channel; +} stbi__pic_packet; + +static stbi_uc* stbi__readval(stbi__context* s, int channel, stbi_uc* dest) +{ + int mask = 0x80, i; + + for (i = 0; i < 4; ++i, mask >>= 1) + { + if (channel & mask) + { + if (stbi__at_eof(s)) + return stbi__errpuc("bad file", "PIC file too short"); + dest[i] = stbi__get8(s); + } + } + + return dest; +} + +static void stbi__copyval(int channel, stbi_uc* dest, const stbi_uc* src) +{ + int mask = 0x80, i; + + for (i = 0; i < 4; ++i, mask >>= 1) + if (channel & mask) + dest[i] = src[i]; +} + +static stbi_uc* stbi__pic_load_core(stbi__context* s, int width, int height, int* comp, stbi_uc* result) +{ + int act_comp = 0, num_packets = 0, y, chained; + stbi__pic_packet packets[10]; + + // this will (should...) cater for even some bizarre stuff like having data + // for the same channel in multiple packets. + do + { + stbi__pic_packet* packet; + + if (num_packets == sizeof(packets) / sizeof(packets[0])) + return stbi__errpuc("bad format", "too many packets"); + + packet = &packets[num_packets++]; + + chained = stbi__get8(s); + packet->size = stbi__get8(s); + packet->type = stbi__get8(s); + packet->channel = stbi__get8(s); + + act_comp |= packet->channel; + + if (stbi__at_eof(s)) + return stbi__errpuc("bad file", "file too short (reading packets)"); + if (packet->size != 8) + return stbi__errpuc("bad format", "packet isn't 8bpp"); + } while (chained); + + *comp = (act_comp & 0x10 ? 4 : 3); // has alpha channel? + + for (y = 0; y < height; ++y) + { + int packet_idx; + + for (packet_idx = 0; packet_idx < num_packets; ++packet_idx) + { + stbi__pic_packet* packet = &packets[packet_idx]; + stbi_uc* dest = result + y * width * 4; + + switch (packet->type) + { + default: return stbi__errpuc("bad format", "packet has bad compression type"); + + case 0: + { // uncompressed + int x; + + for (x = 0; x < width; ++x, dest += 4) + if (!stbi__readval(s, packet->channel, dest)) + return 0; + break; + } + + case 1: // Pure RLE + { + int left = width, i; + + while (left > 0) + { + stbi_uc count, value[4]; + + count = stbi__get8(s); + if (stbi__at_eof(s)) + return stbi__errpuc("bad file", "file too short (pure read count)"); + + if (count > left) + count = (stbi_uc)left; + + if (!stbi__readval(s, packet->channel, value)) + return 0; + + for (i = 0; i < count; ++i, dest += 4) + stbi__copyval(packet->channel, dest, value); + left -= count; + } + } + break; + + case 2: + { // Mixed RLE + int left = width; + while (left > 0) + { + int count = stbi__get8(s), i; + if (stbi__at_eof(s)) + return stbi__errpuc("bad file", "file too short (mixed read count)"); + + if (count >= 128) + { // Repeated + stbi_uc value[4]; + + if (count == 128) + count = stbi__get16be(s); + else + count -= 127; + if (count > left) + return stbi__errpuc("bad file", "scanline overrun"); + + if (!stbi__readval(s, packet->channel, value)) + return 0; + + for (i = 0; i < count; ++i, dest += 4) + stbi__copyval(packet->channel, dest, value); + } + else + { // Raw + ++count; + if (count > left) + return stbi__errpuc("bad file", "scanline overrun"); + + for (i = 0; i < count; ++i, dest += 4) + if (!stbi__readval(s, packet->channel, dest)) + return 0; + } + left -= count; + } + break; + } + } + } + } + + return result; +} + +static void* stbi__pic_load(stbi__context* s, int* px, int* py, int* comp, int req_comp, stbi__result_info* ri) +{ + stbi_uc* result; + int i, x, y, internal_comp; + STBI_NOTUSED(ri); + + if (!comp) + comp = &internal_comp; + + for (i = 0; i < 92; ++i) + stbi__get8(s); + + x = stbi__get16be(s); + y = stbi__get16be(s); + + if (y > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + if (x > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + + if (stbi__at_eof(s)) + return stbi__errpuc("bad file", "file too short (pic header)"); + if (!stbi__mad3sizes_valid(x, y, 4, 0)) + return stbi__errpuc("too large", "PIC image too large to decode"); + + stbi__get32be(s); // skip `ratio' + stbi__get16be(s); // skip `fields' + stbi__get16be(s); // skip `pad' + + // intermediate buffer is RGBA + result = (stbi_uc*)stbi__malloc_mad3(x, y, 4, 0); + if (!result) + return stbi__errpuc("outofmem", "Out of memory"); + memset(result, 0xff, x * y * 4); + + if (!stbi__pic_load_core(s, x, y, comp, result)) + { + STBI_FREE(result); + result = 0; + } + *px = x; + *py = y; + if (req_comp == 0) + req_comp = *comp; + result = stbi__convert_format(result, 4, req_comp, x, y); + + return result; +} + +static int stbi__pic_test(stbi__context* s) +{ + int r = stbi__pic_test_core(s); + stbi__rewind(s); + return r; +} + #endif + +// ************************************************************************************************* +// GIF loader -- public domain by Jean-Marc Lienher -- simplified/shrunk by stb + + #ifndef STBI_NO_GIF +typedef struct { + stbi__int16 prefix; + stbi_uc first; + stbi_uc suffix; +} stbi__gif_lzw; + +typedef struct { + int w, h; + stbi_uc* out; // output buffer (always 4 components) + stbi_uc* background; // The current "background" as far as a gif is concerned + stbi_uc* history; + int flags, bgindex, ratio, transparent, eflags; + stbi_uc pal[256][4]; + stbi_uc lpal[256][4]; + stbi__gif_lzw codes[8192]; + stbi_uc* color_table; + int parse, step; + int lflags; + int start_x, start_y; + int max_x, max_y; + int cur_x, cur_y; + int line_size; + int delay; +} stbi__gif; + +static int stbi__gif_test_raw(stbi__context* s) +{ + int sz; + if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') + return 0; + sz = stbi__get8(s); + if (sz != '9' && sz != '7') + return 0; + if (stbi__get8(s) != 'a') + return 0; + return 1; +} + +static int stbi__gif_test(stbi__context* s) +{ + int r = stbi__gif_test_raw(s); + stbi__rewind(s); + return r; +} + +static void stbi__gif_parse_colortable(stbi__context* s, stbi_uc pal[256][4], int num_entries, int transp) +{ + int i; + for (i = 0; i < num_entries; ++i) + { + pal[i][2] = stbi__get8(s); + pal[i][1] = stbi__get8(s); + pal[i][0] = stbi__get8(s); + pal[i][3] = transp == i ? 0 : 255; + } +} + +static int stbi__gif_header(stbi__context* s, stbi__gif* g, int* comp, int is_info) +{ + stbi_uc version; + if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') + return stbi__err("not GIF", "Corrupt GIF"); + + version = stbi__get8(s); + if (version != '7' && version != '9') + return stbi__err("not GIF", "Corrupt GIF"); + if (stbi__get8(s) != 'a') + return stbi__err("not GIF", "Corrupt GIF"); + + stbi__g_failure_reason = ""; + g->w = stbi__get16le(s); + g->h = stbi__get16le(s); + g->flags = stbi__get8(s); + g->bgindex = stbi__get8(s); + g->ratio = stbi__get8(s); + g->transparent = -1; + + if (g->w > STBI_MAX_DIMENSIONS) + return stbi__err("too large", "Very large image (corrupt?)"); + if (g->h > STBI_MAX_DIMENSIONS) + return stbi__err("too large", "Very large image (corrupt?)"); + + if (comp != 0) + *comp = 4; // can't actually tell whether it's 3 or 4 until we parse the comments + + if (is_info) + return 1; + + if (g->flags & 0x80) + stbi__gif_parse_colortable(s, g->pal, 2 << (g->flags & 7), -1); + + return 1; +} + +static int stbi__gif_info_raw(stbi__context* s, int* x, int* y, int* comp) +{ + stbi__gif* g = (stbi__gif*)stbi__malloc(sizeof(stbi__gif)); + if (!g) + return stbi__err("outofmem", "Out of memory"); + if (!stbi__gif_header(s, g, comp, 1)) + { + STBI_FREE(g); + stbi__rewind(s); + return 0; + } + if (x) + *x = g->w; + if (y) + *y = g->h; + STBI_FREE(g); + return 1; +} + +static void stbi__out_gif_code(stbi__gif* g, stbi__uint16 code) +{ + stbi_uc *p, *c; + int idx; + + // recurse to decode the prefixes, since the linked-list is backwards, + // and working backwards through an interleaved image would be nasty + if (g->codes[code].prefix >= 0) + stbi__out_gif_code(g, g->codes[code].prefix); + + if (g->cur_y >= g->max_y) + return; + + idx = g->cur_x + g->cur_y; + p = &g->out[idx]; + g->history[idx / 4] = 1; + + c = &g->color_table[g->codes[code].suffix * 4]; + if (c[3] > 128) + { // don't render transparent pixels; + p[0] = c[2]; + p[1] = c[1]; + p[2] = c[0]; + p[3] = c[3]; + } + g->cur_x += 4; + + if (g->cur_x >= g->max_x) + { + g->cur_x = g->start_x; + g->cur_y += g->step; + + while (g->cur_y >= g->max_y && g->parse > 0) + { + g->step = (1 << g->parse) * g->line_size; + g->cur_y = g->start_y + (g->step >> 1); + --g->parse; + } + } +} + +static stbi_uc* stbi__process_gif_raster(stbi__context* s, stbi__gif* g) +{ + stbi_uc lzw_cs; + stbi__int32 len, init_code; + stbi__uint32 first; + stbi__int32 codesize, codemask, avail, oldcode, bits, valid_bits, clear; + stbi__gif_lzw* p; + + lzw_cs = stbi__get8(s); + if (lzw_cs > 12) + return NULL; + clear = 1 << lzw_cs; + first = 1; + codesize = lzw_cs + 1; + codemask = (1 << codesize) - 1; + bits = 0; + valid_bits = 0; + for (init_code = 0; init_code < clear; init_code++) + { + g->codes[init_code].prefix = -1; + g->codes[init_code].first = (stbi_uc)init_code; + g->codes[init_code].suffix = (stbi_uc)init_code; + } + + // support no starting clear code + avail = clear + 2; + oldcode = -1; + + len = 0; + for (;;) + { + if (valid_bits < codesize) + { + if (len == 0) + { + len = stbi__get8(s); // start new block + if (len == 0) + return g->out; + } + --len; + bits |= (stbi__int32)stbi__get8(s) << valid_bits; + valid_bits += 8; + } + else + { + stbi__int32 code = bits & codemask; + bits >>= codesize; + valid_bits -= codesize; + // @OPTIMIZE: is there some way we can accelerate the non-clear path? + if (code == clear) + { // clear code + codesize = lzw_cs + 1; + codemask = (1 << codesize) - 1; + avail = clear + 2; + oldcode = -1; + first = 0; + } + else if (code == clear + 1) + { // end of stream code + stbi__skip(s, len); + while ((len = stbi__get8(s)) > 0) + stbi__skip(s, len); + return g->out; + } + else if (code <= avail) + { + if (first) + { + return stbi__errpuc("no clear code", "Corrupt GIF"); + } + + if (oldcode >= 0) + { + p = &g->codes[avail++]; + if (avail > 8192) + { + return stbi__errpuc("too many codes", "Corrupt GIF"); + } + + p->prefix = (stbi__int16)oldcode; + p->first = g->codes[oldcode].first; + p->suffix = (code == avail) ? p->first : g->codes[code].first; + } + else if (code == avail) + return stbi__errpuc("illegal code in raster", "Corrupt GIF"); + + stbi__out_gif_code(g, (stbi__uint16)code); + + if ((avail & codemask) == 0 && avail <= 0x0FFF) + { + codesize++; + codemask = (1 << codesize) - 1; + } + + oldcode = code; + } + else + { + return stbi__errpuc("illegal code in raster", "Corrupt GIF"); + } + } + } +} + +// this function is designed to support animated gifs, although stb_image doesn't support it +// two back is the image from two frames ago, used for a very specific disposal format +static stbi_uc* stbi__gif_load_next(stbi__context* s, stbi__gif* g, int* comp, int req_comp, stbi_uc* two_back) +{ + int dispose; + int first_frame; + int pi; + int pcount; + STBI_NOTUSED(req_comp); + + // on first frame, any non-written pixels get the background colour (non-transparent) + first_frame = 0; + if (g->out == 0) + { + if (!stbi__gif_header(s, g, comp, 0)) + return 0; // stbi__g_failure_reason set by stbi__gif_header + if (!stbi__mad3sizes_valid(4, g->w, g->h, 0)) + return stbi__errpuc("too large", "GIF image is too large"); + pcount = g->w * g->h; + g->out = (stbi_uc*)stbi__malloc(4 * pcount); + g->background = (stbi_uc*)stbi__malloc(4 * pcount); + g->history = (stbi_uc*)stbi__malloc(pcount); + if (!g->out || !g->background || !g->history) + return stbi__errpuc("outofmem", "Out of memory"); + + // image is treated as "transparent" at the start - ie, nothing overwrites the current background; + // background colour is only used for pixels that are not rendered first frame, after that "background" + // color refers to the color that was there the previous frame. + memset(g->out, 0x00, 4 * pcount); + memset(g->background, 0x00, 4 * pcount); // state of the background (starts transparent) + memset(g->history, 0x00, pcount); // pixels that were affected previous frame + first_frame = 1; + } + else + { + // second frame - how do we dispose of the previous one? + dispose = (g->eflags & 0x1C) >> 2; + pcount = g->w * g->h; + + if ((dispose == 3) && (two_back == 0)) + { + dispose = 2; // if I don't have an image to revert back to, default to the old background + } + + if (dispose == 3) + { // use previous graphic + for (pi = 0; pi < pcount; ++pi) + { + if (g->history[pi]) + { + memcpy(&g->out[pi * 4], &two_back[pi * 4], 4); + } + } + } + else if (dispose == 2) + { + // restore what was changed last frame to background before that frame; + for (pi = 0; pi < pcount; ++pi) + { + if (g->history[pi]) + { + memcpy(&g->out[pi * 4], &g->background[pi * 4], 4); + } + } + } + else + { + // This is a non-disposal case eithe way, so just + // leave the pixels as is, and they will become the new background + // 1: do not dispose + // 0: not specified. + } + + // background is what out is after the undoing of the previou frame; + memcpy(g->background, g->out, 4 * g->w * g->h); + } + + // clear my history; + memset(g->history, 0x00, g->w * g->h); // pixels that were affected previous frame + + for (;;) + { + int tag = stbi__get8(s); + switch (tag) + { + case 0x2C: /* Image Descriptor */ + { + stbi__int32 x, y, w, h; + stbi_uc* o; + + x = stbi__get16le(s); + y = stbi__get16le(s); + w = stbi__get16le(s); + h = stbi__get16le(s); + if (((x + w) > (g->w)) || ((y + h) > (g->h))) + return stbi__errpuc("bad Image Descriptor", "Corrupt GIF"); + + g->line_size = g->w * 4; + g->start_x = x * 4; + g->start_y = y * g->line_size; + g->max_x = g->start_x + w * 4; + g->max_y = g->start_y + h * g->line_size; + g->cur_x = g->start_x; + g->cur_y = g->start_y; + + // if the width of the specified rectangle is 0, that means + // we may not see *any* pixels or the image is malformed; + // to make sure this is caught, move the current y down to + // max_y (which is what out_gif_code checks). + if (w == 0) + g->cur_y = g->max_y; + + g->lflags = stbi__get8(s); + + if (g->lflags & 0x40) + { + g->step = 8 * g->line_size; // first interlaced spacing + g->parse = 3; + } + else + { + g->step = g->line_size; + g->parse = 0; + } + + if (g->lflags & 0x80) + { + stbi__gif_parse_colortable(s, g->lpal, 2 << (g->lflags & 7), g->eflags & 0x01 ? g->transparent : -1); + g->color_table = (stbi_uc*)g->lpal; + } + else if (g->flags & 0x80) + { + g->color_table = (stbi_uc*)g->pal; + } + else + return stbi__errpuc("missing color table", "Corrupt GIF"); + + o = stbi__process_gif_raster(s, g); + if (!o) + return NULL; + + // if this was the first frame, + pcount = g->w * g->h; + if (first_frame && (g->bgindex > 0)) + { + // if first frame, any pixel not drawn to gets the background color + for (pi = 0; pi < pcount; ++pi) + { + if (g->history[pi] == 0) + { + g->pal[g->bgindex][3] = 255; // just in case it was made transparent, undo that; It will be reset next frame if need be; + memcpy(&g->out[pi * 4], &g->pal[g->bgindex], 4); + } + } + } + + return o; + } + + case 0x21: // Comment Extension. + { + int len; + int ext = stbi__get8(s); + if (ext == 0xF9) + { // Graphic Control Extension. + len = stbi__get8(s); + if (len == 4) + { + g->eflags = stbi__get8(s); + g->delay = 10 * stbi__get16le(s); // delay - 1/100th of a second, saving as 1/1000ths. + + // unset old transparent + if (g->transparent >= 0) + { + g->pal[g->transparent][3] = 255; + } + if (g->eflags & 0x01) + { + g->transparent = stbi__get8(s); + if (g->transparent >= 0) + { + g->pal[g->transparent][3] = 0; + } + } + else + { + // don't need transparent + stbi__skip(s, 1); + g->transparent = -1; + } + } + else + { + stbi__skip(s, len); + break; + } + } + while ((len = stbi__get8(s)) != 0) + { + stbi__skip(s, len); + } + break; + } + + case 0x3B: // gif stream termination code + return (stbi_uc*)s; // using '1' causes warning on some compilers + + default: return stbi__errpuc("unknown code", "Corrupt GIF"); + } + } +} + +static void* stbi__load_gif_main_outofmem(stbi__gif* g, stbi_uc* out, int** delays) +{ + STBI_FREE(g->out); + STBI_FREE(g->history); + STBI_FREE(g->background); + + if (out) + STBI_FREE(out); + if (delays && *delays) + STBI_FREE(*delays); + return stbi__errpuc("outofmem", "Out of memory"); +} + +static void* stbi__load_gif_main(stbi__context* s, int** delays, int* x, int* y, int* z, int* comp, int req_comp) +{ + if (stbi__gif_test(s)) + { + int layers = 0; + stbi_uc* u = 0; + stbi_uc* out = 0; + stbi_uc* two_back = 0; + stbi__gif g; + int stride; + int out_size = 0; + int delays_size = 0; + + STBI_NOTUSED(out_size); + STBI_NOTUSED(delays_size); + + memset(&g, 0, sizeof(g)); + if (delays) + { + *delays = 0; + } + + do + { + u = stbi__gif_load_next(s, &g, comp, req_comp, two_back); + if (u == (stbi_uc*)s) + u = 0; // end of animated gif marker + + if (u) + { + *x = g.w; + *y = g.h; + ++layers; + stride = g.w * g.h * 4; + + if (out) + { + void* tmp = (stbi_uc*)STBI_REALLOC_SIZED(out, out_size, layers * stride); + if (!tmp) + return stbi__load_gif_main_outofmem(&g, out, delays); + else + { + out = (stbi_uc*)tmp; + out_size = layers * stride; + } + + if (delays) + { + int* new_delays = (int*)STBI_REALLOC_SIZED(*delays, delays_size, sizeof(int) * layers); + if (!new_delays) + return stbi__load_gif_main_outofmem(&g, out, delays); + *delays = new_delays; + delays_size = layers * sizeof(int); + } + } + else + { + out = (stbi_uc*)stbi__malloc(layers * stride); + if (!out) + return stbi__load_gif_main_outofmem(&g, out, delays); + out_size = layers * stride; + if (delays) + { + *delays = (int*)stbi__malloc(layers * sizeof(int)); + if (!*delays) + return stbi__load_gif_main_outofmem(&g, out, delays); + delays_size = layers * sizeof(int); + } + } + memcpy(out + ((layers - 1) * stride), u, stride); + if (layers >= 2) + { + two_back = out - 2 * stride; + } + + if (delays) + { + (*delays)[layers - 1U] = g.delay; + } + } + } while (u != 0); + + // free temp buffer; + STBI_FREE(g.out); + STBI_FREE(g.history); + STBI_FREE(g.background); + + // do the final conversion after loading everything; + if (req_comp && req_comp != 4) + out = stbi__convert_format(out, 4, req_comp, layers * g.w, g.h); + + *z = layers; + return out; + } + else + { + return stbi__errpuc("not GIF", "Image was not as a gif type."); + } +} + +static void* stbi__gif_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) +{ + stbi_uc* u = 0; + stbi__gif g; + memset(&g, 0, sizeof(g)); + STBI_NOTUSED(ri); + + u = stbi__gif_load_next(s, &g, comp, req_comp, 0); + if (u == (stbi_uc*)s) + u = 0; // end of animated gif marker + if (u) + { + *x = g.w; + *y = g.h; + + // moved conversion to after successful load so that the same + // can be done for multiple frames. + if (req_comp && req_comp != 4) + u = stbi__convert_format(u, 4, req_comp, g.w, g.h); + } + else if (g.out) + { + // if there was an error and we allocated an image buffer, free it! + STBI_FREE(g.out); + } + + // free buffers needed for multiple frame loading; + STBI_FREE(g.history); + STBI_FREE(g.background); + + return u; +} + +static int stbi__gif_info(stbi__context* s, int* x, int* y, int* comp) +{ + return stbi__gif_info_raw(s, x, y, comp); +} + #endif + + // ************************************************************************************************* + // Radiance RGBE HDR loader + // originally by Nicolas Schulz + #ifndef STBI_NO_HDR +static int stbi__hdr_test_core(stbi__context* s, const char* signature) +{ + int i; + for (i = 0; signature[i]; ++i) + if (stbi__get8(s) != signature[i]) + return 0; + stbi__rewind(s); + return 1; +} + +static int stbi__hdr_test(stbi__context* s) +{ + int r = stbi__hdr_test_core(s, "#?RADIANCE\n"); + stbi__rewind(s); + if (!r) + { + r = stbi__hdr_test_core(s, "#?RGBE\n"); + stbi__rewind(s); + } + return r; +} + + #define STBI__HDR_BUFLEN 1024 +static char* stbi__hdr_gettoken(stbi__context* z, char* buffer) +{ + int len = 0; + char c = '\0'; + + c = (char)stbi__get8(z); + + while (!stbi__at_eof(z) && c != '\n') + { + buffer[len++] = c; + if (len == STBI__HDR_BUFLEN - 1) + { + // flush to end of line + while (!stbi__at_eof(z) && stbi__get8(z) != '\n') + ; + break; + } + c = (char)stbi__get8(z); + } + + buffer[len] = 0; + return buffer; +} + +static void stbi__hdr_convert(float* output, stbi_uc* input, int req_comp) +{ + if (input[3] != 0) + { + float f1; + // Exponent + f1 = (float)ldexp(1.0f, input[3] - (int)(128 + 8)); + if (req_comp <= 2) + output[0] = (input[0] + input[1] + input[2]) * f1 / 3; + else + { + output[0] = input[0] * f1; + output[1] = input[1] * f1; + output[2] = input[2] * f1; + } + if (req_comp == 2) + output[1] = 1; + if (req_comp == 4) + output[3] = 1; + } + else + { + switch (req_comp) + { + case 4: output[3] = 1; /* fallthrough */ + case 3: output[0] = output[1] = output[2] = 0; break; + case 2: output[1] = 1; /* fallthrough */ + case 1: output[0] = 0; break; + } + } +} + +static float* stbi__hdr_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) +{ + char buffer[STBI__HDR_BUFLEN]; + char* token; + int valid = 0; + int width, height; + stbi_uc* scanline; + float* hdr_data; + int len; + unsigned char count, value; + int i, j, k, c1, c2, z; + const char* headerToken; + STBI_NOTUSED(ri); + + // Check identifier + headerToken = stbi__hdr_gettoken(s, buffer); + if (strcmp(headerToken, "#?RADIANCE") != 0 && strcmp(headerToken, "#?RGBE") != 0) + return stbi__errpf("not HDR", "Corrupt HDR image"); + + // Parse header + for (;;) + { + token = stbi__hdr_gettoken(s, buffer); + if (token[0] == 0) + break; + if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) + valid = 1; + } + + if (!valid) + return stbi__errpf("unsupported format", "Unsupported HDR format"); + + // Parse width and height + // can't use sscanf() if we're not using stdio! + token = stbi__hdr_gettoken(s, buffer); + if (strncmp(token, "-Y ", 3)) + return stbi__errpf("unsupported data layout", "Unsupported HDR format"); + token += 3; + height = (int)strtol(token, &token, 10); + while (*token == ' ') + ++token; + if (strncmp(token, "+X ", 3)) + return stbi__errpf("unsupported data layout", "Unsupported HDR format"); + token += 3; + width = (int)strtol(token, NULL, 10); + + if (height > STBI_MAX_DIMENSIONS) + return stbi__errpf("too large", "Very large image (corrupt?)"); + if (width > STBI_MAX_DIMENSIONS) + return stbi__errpf("too large", "Very large image (corrupt?)"); + + *x = width; + *y = height; + + if (comp) + *comp = 3; + if (req_comp == 0) + req_comp = 3; + + if (!stbi__mad4sizes_valid(width, height, req_comp, sizeof(float), 0)) + return stbi__errpf("too large", "HDR image is too large"); + + // Read data + hdr_data = (float*)stbi__malloc_mad4(width, height, req_comp, sizeof(float), 0); + if (!hdr_data) + return stbi__errpf("outofmem", "Out of memory"); + + // Load image data + // image data is stored as some number of sca + if (width < 8 || width >= 32768) + { + // Read flat data + for (j = 0; j < height; ++j) + { + for (i = 0; i < width; ++i) + { + stbi_uc rgbe[4]; + main_decode_loop: + stbi__getn(s, rgbe, 4); + stbi__hdr_convert(hdr_data + j * width * req_comp + i * req_comp, rgbe, req_comp); + } + } + } + else + { + // Read RLE-encoded data + scanline = NULL; + + for (j = 0; j < height; ++j) + { + c1 = stbi__get8(s); + c2 = stbi__get8(s); + len = stbi__get8(s); + if (c1 != 2 || c2 != 2 || (len & 0x80)) + { + // not run-length encoded, so we have to actually use THIS data as a decoded + // pixel (note this can't be a valid pixel--one of RGB must be >= 128) + stbi_uc rgbe[4]; + rgbe[0] = (stbi_uc)c1; + rgbe[1] = (stbi_uc)c2; + rgbe[2] = (stbi_uc)len; + rgbe[3] = (stbi_uc)stbi__get8(s); + stbi__hdr_convert(hdr_data, rgbe, req_comp); + i = 1; + j = 0; + STBI_FREE(scanline); + goto main_decode_loop; // yes, this makes no sense + } + len <<= 8; + len |= stbi__get8(s); + if (len != width) + { + STBI_FREE(hdr_data); + STBI_FREE(scanline); + return stbi__errpf("invalid decoded scanline length", "corrupt HDR"); + } + if (scanline == NULL) + { + scanline = (stbi_uc*)stbi__malloc_mad2(width, 4, 0); + if (!scanline) + { + STBI_FREE(hdr_data); + return stbi__errpf("outofmem", "Out of memory"); + } + } + + for (k = 0; k < 4; ++k) + { + int nleft; + i = 0; + while ((nleft = width - i) > 0) + { + count = stbi__get8(s); + if (count > 128) + { + // Run + value = stbi__get8(s); + count -= 128; + if ((count == 0) || (count > nleft)) + { + STBI_FREE(hdr_data); + STBI_FREE(scanline); + return stbi__errpf("corrupt", "bad RLE data in HDR"); + } + for (z = 0; z < count; ++z) + scanline[i++ * 4 + k] = value; + } + else + { + // Dump + if ((count == 0) || (count > nleft)) + { + STBI_FREE(hdr_data); + STBI_FREE(scanline); + return stbi__errpf("corrupt", "bad RLE data in HDR"); + } + for (z = 0; z < count; ++z) + scanline[i++ * 4 + k] = stbi__get8(s); + } + } + } + for (i = 0; i < width; ++i) + stbi__hdr_convert(hdr_data + (j * width + i) * req_comp, scanline + i * 4, req_comp); + } + if (scanline) + STBI_FREE(scanline); + } + + return hdr_data; +} + +static int stbi__hdr_info(stbi__context* s, int* x, int* y, int* comp) +{ + char buffer[STBI__HDR_BUFLEN]; + char* token; + int valid = 0; + int dummy; + + if (!x) + x = &dummy; + if (!y) + y = &dummy; + if (!comp) + comp = &dummy; + + if (stbi__hdr_test(s) == 0) + { + stbi__rewind(s); + return 0; + } + + for (;;) + { + token = stbi__hdr_gettoken(s, buffer); + if (token[0] == 0) + break; + if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) + valid = 1; + } + + if (!valid) + { + stbi__rewind(s); + return 0; + } + token = stbi__hdr_gettoken(s, buffer); + if (strncmp(token, "-Y ", 3)) + { + stbi__rewind(s); + return 0; + } + token += 3; + *y = (int)strtol(token, &token, 10); + while (*token == ' ') + ++token; + if (strncmp(token, "+X ", 3)) + { + stbi__rewind(s); + return 0; + } + token += 3; + *x = (int)strtol(token, NULL, 10); + *comp = 3; + return 1; +} + #endif // STBI_NO_HDR + + #ifndef STBI_NO_BMP +static int stbi__bmp_info(stbi__context* s, int* x, int* y, int* comp) +{ + void* p; + stbi__bmp_data info; + + info.all_a = 255; + p = stbi__bmp_parse_header(s, &info); + if (p == NULL) + { + stbi__rewind(s); + return 0; + } + if (x) + *x = s->img_x; + if (y) + *y = s->img_y; + if (comp) + { + if (info.bpp == 24 && info.ma == 0xff000000) + *comp = 3; + else + *comp = info.ma ? 4 : 3; + } + return 1; +} + #endif + + #ifndef STBI_NO_PSD +static int stbi__psd_info(stbi__context* s, int* x, int* y, int* comp) +{ + int channelCount, dummy, depth; + if (!x) + x = &dummy; + if (!y) + y = &dummy; + if (!comp) + comp = &dummy; + if (stbi__get32be(s) != 0x38425053) + { + stbi__rewind(s); + return 0; + } + if (stbi__get16be(s) != 1) + { + stbi__rewind(s); + return 0; + } + stbi__skip(s, 6); + channelCount = stbi__get16be(s); + if (channelCount < 0 || channelCount > 16) + { + stbi__rewind(s); + return 0; + } + *y = stbi__get32be(s); + *x = stbi__get32be(s); + depth = stbi__get16be(s); + if (depth != 8 && depth != 16) + { + stbi__rewind(s); + return 0; + } + if (stbi__get16be(s) != 3) + { + stbi__rewind(s); + return 0; + } + *comp = 4; + return 1; +} + +static int stbi__psd_is16(stbi__context* s) +{ + int channelCount, depth; + if (stbi__get32be(s) != 0x38425053) + { + stbi__rewind(s); + return 0; + } + if (stbi__get16be(s) != 1) + { + stbi__rewind(s); + return 0; + } + stbi__skip(s, 6); + channelCount = stbi__get16be(s); + if (channelCount < 0 || channelCount > 16) + { + stbi__rewind(s); + return 0; + } + STBI_NOTUSED(stbi__get32be(s)); + STBI_NOTUSED(stbi__get32be(s)); + depth = stbi__get16be(s); + if (depth != 16) + { + stbi__rewind(s); + return 0; + } + return 1; +} + #endif + + #ifndef STBI_NO_PIC +static int stbi__pic_info(stbi__context* s, int* x, int* y, int* comp) +{ + int act_comp = 0, num_packets = 0, chained, dummy; + stbi__pic_packet packets[10]; + + if (!x) + x = &dummy; + if (!y) + y = &dummy; + if (!comp) + comp = &dummy; + + if (!stbi__pic_is4(s, "\x53\x80\xF6\x34")) + { + stbi__rewind(s); + return 0; + } + + stbi__skip(s, 88); + + *x = stbi__get16be(s); + *y = stbi__get16be(s); + if (stbi__at_eof(s)) + { + stbi__rewind(s); + return 0; + } + if ((*x) != 0 && (1 << 28) / (*x) < (*y)) + { + stbi__rewind(s); + return 0; + } + + stbi__skip(s, 8); + + do + { + stbi__pic_packet* packet; + + if (num_packets == sizeof(packets) / sizeof(packets[0])) + return 0; + + packet = &packets[num_packets++]; + chained = stbi__get8(s); + packet->size = stbi__get8(s); + packet->type = stbi__get8(s); + packet->channel = stbi__get8(s); + act_comp |= packet->channel; + + if (stbi__at_eof(s)) + { + stbi__rewind(s); + return 0; + } + if (packet->size != 8) + { + stbi__rewind(s); + return 0; + } + } while (chained); + + *comp = (act_comp & 0x10 ? 4 : 3); + + return 1; +} + #endif + +// ************************************************************************************************* +// Portable Gray Map and Portable Pixel Map loader +// by Ken Miller +// +// PGM: http://netpbm.sourceforge.net/doc/pgm.html +// PPM: http://netpbm.sourceforge.net/doc/ppm.html +// +// Known limitations: +// Does not support comments in the header section +// Does not support ASCII image data (formats P2 and P3) + + #ifndef STBI_NO_PNM + +static int stbi__pnm_test(stbi__context* s) +{ + char p, t; + p = (char)stbi__get8(s); + t = (char)stbi__get8(s); + if (p != 'P' || (t != '5' && t != '6')) + { + stbi__rewind(s); + return 0; + } + return 1; +} + +static void* stbi__pnm_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) +{ + stbi_uc* out; + STBI_NOTUSED(ri); + + ri->bits_per_channel = stbi__pnm_info(s, (int*)&s->img_x, (int*)&s->img_y, (int*)&s->img_n); + if (ri->bits_per_channel == 0) + return 0; + + if (s->img_y > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + if (s->img_x > STBI_MAX_DIMENSIONS) + return stbi__errpuc("too large", "Very large image (corrupt?)"); + + *x = s->img_x; + *y = s->img_y; + if (comp) + *comp = s->img_n; + + if (!stbi__mad4sizes_valid(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0)) + return stbi__errpuc("too large", "PNM too large"); + + out = (stbi_uc*)stbi__malloc_mad4(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0); + if (!out) + return stbi__errpuc("outofmem", "Out of memory"); + if (!stbi__getn(s, out, s->img_n * s->img_x * s->img_y * (ri->bits_per_channel / 8))) + { + STBI_FREE(out); + return stbi__errpuc("bad PNM", "PNM file truncated"); + } + + if (req_comp && req_comp != s->img_n) + { + if (ri->bits_per_channel == 16) + { + out = (stbi_uc*)stbi__convert_format16((stbi__uint16*)out, s->img_n, req_comp, s->img_x, s->img_y); + } + else + { + out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y); + } + if (out == NULL) + return out; // stbi__convert_format frees input on failure + } + return out; +} + +static int stbi__pnm_isspace(char c) +{ + return c == ' ' || c == '\t' || c == '\n' || c == '\v' || c == '\f' || c == '\r'; +} + +static void stbi__pnm_skip_whitespace(stbi__context* s, char* c) +{ + for (;;) + { + while (!stbi__at_eof(s) && stbi__pnm_isspace(*c)) + *c = (char)stbi__get8(s); + + if (stbi__at_eof(s) || *c != '#') + break; + + while (!stbi__at_eof(s) && *c != '\n' && *c != '\r') + *c = (char)stbi__get8(s); + } +} + +static int stbi__pnm_isdigit(char c) +{ + return c >= '0' && c <= '9'; +} + +static int stbi__pnm_getinteger(stbi__context* s, char* c) +{ + int value = 0; + + while (!stbi__at_eof(s) && stbi__pnm_isdigit(*c)) + { + value = value * 10 + (*c - '0'); + *c = (char)stbi__get8(s); + if ((value > 214748364) || (value == 214748364 && *c > '7')) + return stbi__err("integer parse overflow", "Parsing an integer in the PPM header overflowed a 32-bit int"); + } + + return value; +} + +static int stbi__pnm_info(stbi__context* s, int* x, int* y, int* comp) +{ + int maxv, dummy; + char c, p, t; + + if (!x) + x = &dummy; + if (!y) + y = &dummy; + if (!comp) + comp = &dummy; + + stbi__rewind(s); + + // Get identifier + p = (char)stbi__get8(s); + t = (char)stbi__get8(s); + if (p != 'P' || (t != '5' && t != '6')) + { + stbi__rewind(s); + return 0; + } + + *comp = (t == '6') ? 3 : 1; // '5' is 1-component .pgm; '6' is 3-component .ppm + + c = (char)stbi__get8(s); + stbi__pnm_skip_whitespace(s, &c); + + *x = stbi__pnm_getinteger(s, &c); // read width + if (*x == 0) + return stbi__err("invalid width", "PPM image header had zero or overflowing width"); + stbi__pnm_skip_whitespace(s, &c); + + *y = stbi__pnm_getinteger(s, &c); // read height + if (*y == 0) + return stbi__err("invalid width", "PPM image header had zero or overflowing width"); + stbi__pnm_skip_whitespace(s, &c); + + maxv = stbi__pnm_getinteger(s, &c); // read max value + if (maxv > 65535) + return stbi__err("max value > 65535", "PPM image supports only 8-bit and 16-bit images"); + else if (maxv > 255) + return 16; + else + return 8; +} + +static int stbi__pnm_is16(stbi__context* s) +{ + if (stbi__pnm_info(s, NULL, NULL, NULL) == 16) + return 1; + return 0; +} + #endif + +static int stbi__info_main(stbi__context* s, int* x, int* y, int* comp) +{ + #ifndef STBI_NO_JPEG + if (stbi__jpeg_info(s, x, y, comp)) + return 1; + #endif + + #ifndef STBI_NO_PNG + if (stbi__png_info(s, x, y, comp)) + return 1; + #endif + + #ifndef STBI_NO_GIF + if (stbi__gif_info(s, x, y, comp)) + return 1; + #endif + + #ifndef STBI_NO_BMP + if (stbi__bmp_info(s, x, y, comp)) + return 1; + #endif + + #ifndef STBI_NO_PSD + if (stbi__psd_info(s, x, y, comp)) + return 1; + #endif + + #ifndef STBI_NO_PIC + if (stbi__pic_info(s, x, y, comp)) + return 1; + #endif + + #ifndef STBI_NO_PNM + if (stbi__pnm_info(s, x, y, comp)) + return 1; + #endif + + #ifndef STBI_NO_HDR + if (stbi__hdr_info(s, x, y, comp)) + return 1; + #endif + + // test tga last because it's a crappy test! + #ifndef STBI_NO_TGA + if (stbi__tga_info(s, x, y, comp)) + return 1; + #endif + return stbi__err("unknown image type", "Image not of any known type, or corrupt"); +} + +static int stbi__is_16_main(stbi__context* s) +{ + #ifndef STBI_NO_PNG + if (stbi__png_is16(s)) + return 1; + #endif + + #ifndef STBI_NO_PSD + if (stbi__psd_is16(s)) + return 1; + #endif + + #ifndef STBI_NO_PNM + if (stbi__pnm_is16(s)) + return 1; + #endif + return 0; +} + + #ifndef STBI_NO_STDIO +STBIDEF int stbi_info(const char* filename, int* x, int* y, int* comp) +{ + FILE* f = stbi__fopen(filename, "rb"); + int result; + if (!f) + return stbi__err("can't fopen", "Unable to open file"); + result = stbi_info_from_file(f, x, y, comp); + fclose(f); + return result; +} + +STBIDEF int stbi_info_from_file(FILE* f, int* x, int* y, int* comp) +{ + int r; + stbi__context s; + long pos = ftell(f); + stbi__start_file(&s, f); + r = stbi__info_main(&s, x, y, comp); + fseek(f, pos, SEEK_SET); + return r; +} + +STBIDEF int stbi_is_16_bit(const char* filename) +{ + FILE* f = stbi__fopen(filename, "rb"); + int result; + if (!f) + return stbi__err("can't fopen", "Unable to open file"); + result = stbi_is_16_bit_from_file(f); + fclose(f); + return result; +} + +STBIDEF int stbi_is_16_bit_from_file(FILE* f) +{ + int r; + stbi__context s; + long pos = ftell(f); + stbi__start_file(&s, f); + r = stbi__is_16_main(&s); + fseek(f, pos, SEEK_SET); + return r; +} + #endif // !STBI_NO_STDIO + +STBIDEF int stbi_info_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp) +{ + stbi__context s; + stbi__start_mem(&s, buffer, len); + return stbi__info_main(&s, x, y, comp); +} + +STBIDEF int stbi_info_from_callbacks(const stbi_io_callbacks* c, void* user, int* x, int* y, int* comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks*)c, user); + return stbi__info_main(&s, x, y, comp); +} + +STBIDEF int stbi_is_16_bit_from_memory(const stbi_uc* buffer, int len) +{ + stbi__context s; + stbi__start_mem(&s, buffer, len); + return stbi__is_16_main(&s); +} + +STBIDEF int stbi_is_16_bit_from_callbacks(const stbi_io_callbacks* c, void* user) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks*)c, user); + return stbi__is_16_main(&s); +} + +#endif // STB_IMAGE_IMPLEMENTATION + + /* + revision history: + 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs + 2.19 (2018-02-11) fix warning + 2.18 (2018-01-30) fix warnings + 2.17 (2018-01-29) change sbti__shiftsigned to avoid clang -O2 bug + 1-bit BMP + *_is_16_bit api + avoid warnings + 2.16 (2017-07-23) all functions have 16-bit variants; + STBI_NO_STDIO works again; + compilation fixes; + fix rounding in unpremultiply; + optimize vertical flip; + disable raw_len validation; + documentation fixes + 2.15 (2017-03-18) fix png-1,2,4 bug; now all Imagenet JPGs decode; + warning fixes; disable run-time SSE detection on gcc; + uniform handling of optional "return" values; + thread-safe initialization of zlib tables + 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs + 2.13 (2016-11-29) add 16-bit API, only supported for PNG right now + 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes + 2.11 (2016-04-02) allocate large structures on the stack + remove white matting for transparent PSD + fix reported channel count for PNG & BMP + re-enable SSE2 in non-gcc 64-bit + support RGB-formatted JPEG + read 16-bit PNGs (only as 8-bit) + 2.10 (2016-01-22) avoid warning introduced in 2.09 by STBI_REALLOC_SIZED + 2.09 (2016-01-16) allow comments in PNM files + 16-bit-per-pixel TGA (not bit-per-component) + info() for TGA could break due to .hdr handling + info() for BMP to shares code instead of sloppy parse + can use STBI_REALLOC_SIZED if allocator doesn't support realloc + code cleanup + 2.08 (2015-09-13) fix to 2.07 cleanup, reading RGB PSD as RGBA + 2.07 (2015-09-13) fix compiler warnings + partial animated GIF support + limited 16-bpc PSD support + #ifdef unused functions + bug with < 92 byte PIC,PNM,HDR,TGA + 2.06 (2015-04-19) fix bug where PSD returns wrong '*comp' value + 2.05 (2015-04-19) fix bug in progressive JPEG handling, fix warning + 2.04 (2015-04-15) try to re-enable SIMD on MinGW 64-bit + 2.03 (2015-04-12) extra corruption checking (mmozeiko) + stbi_set_flip_vertically_on_load (nguillemot) + fix NEON support; fix mingw support + 2.02 (2015-01-19) fix incorrect assert, fix warning + 2.01 (2015-01-17) fix various warnings; suppress SIMD on gcc 32-bit without -msse2 + 2.00b (2014-12-25) fix STBI_MALLOC in progressive JPEG + 2.00 (2014-12-25) optimize JPG, including x86 SSE2 & NEON SIMD (ryg) + progressive JPEG (stb) + PGM/PPM support (Ken Miller) + STBI_MALLOC,STBI_REALLOC,STBI_FREE + GIF bugfix -- seemingly never worked + STBI_NO_*, STBI_ONLY_* + 1.48 (2014-12-14) fix incorrectly-named assert() + 1.47 (2014-12-14) 1/2/4-bit PNG support, both direct and paletted (Omar Cornut & stb) + optimize PNG (ryg) + fix bug in interlaced PNG with user-specified channel count (stb) + 1.46 (2014-08-26) + fix broken tRNS chunk (colorkey-style transparency) in non-paletted PNG + 1.45 (2014-08-16) + fix MSVC-ARM internal compiler error by wrapping malloc + 1.44 (2014-08-07) + various warning fixes from Ronny Chevalier + 1.43 (2014-07-15) + fix MSVC-only compiler problem in code changed in 1.42 + 1.42 (2014-07-09) + don't define _CRT_SECURE_NO_WARNINGS (affects user code) + fixes to stbi__cleanup_jpeg path + added STBI_ASSERT to avoid requiring assert.h + 1.41 (2014-06-25) + fix search&replace from 1.36 that messed up comments/error messages + 1.40 (2014-06-22) + fix gcc struct-initialization warning + 1.39 (2014-06-15) + fix to TGA optimization when req_comp != number of components in TGA; + fix to GIF loading because BMP wasn't rewinding (whoops, no GIFs in my test suite) + add support for BMP version 5 (more ignored fields) + 1.38 (2014-06-06) + suppress MSVC warnings on integer casts truncating values + fix accidental rename of 'skip' field of I/O + 1.37 (2014-06-04) + remove duplicate typedef + 1.36 (2014-06-03) + convert to header file single-file library + if de-iphone isn't set, load iphone images color-swapped instead of returning NULL + 1.35 (2014-05-27) + various warnings + fix broken STBI_SIMD path + fix bug where stbi_load_from_file no longer left file pointer in correct place + fix broken non-easy path for 32-bit BMP (possibly never used) + TGA optimization by Arseny Kapoulkine + 1.34 (unknown) + use STBI_NOTUSED in stbi__resample_row_generic(), fix one more leak in tga failure case + 1.33 (2011-07-14) + make stbi_is_hdr work in STBI_NO_HDR (as specified), minor compiler-friendly improvements + 1.32 (2011-07-13) + support for "info" function for all supported filetypes (SpartanJ) + 1.31 (2011-06-20) + a few more leak fixes, bug in PNG handling (SpartanJ) + 1.30 (2011-06-11) + added ability to load files via callbacks to accomidate custom input streams (Ben Wenger) + removed deprecated format-specific test/load functions + removed support for installable file formats (stbi_loader) -- would have been broken for IO callbacks anyway + error cases in bmp and tga give messages and don't leak (Raymond Barbiero, grisha) + fix inefficiency in decoding 32-bit BMP (David Woo) + 1.29 (2010-08-16) + various warning fixes from Aurelien Pocheville + 1.28 (2010-08-01) + fix bug in GIF palette transparency (SpartanJ) + 1.27 (2010-08-01) + cast-to-stbi_uc to fix warnings + 1.26 (2010-07-24) + fix bug in file buffering for PNG reported by SpartanJ + 1.25 (2010-07-17) + refix trans_data warning (Won Chun) + 1.24 (2010-07-12) + perf improvements reading from files on platforms with lock-heavy fgetc() + minor perf improvements for jpeg + deprecated type-specific functions so we'll get feedback if they're needed + attempt to fix trans_data warning (Won Chun) + 1.23 fixed bug in iPhone support + 1.22 (2010-07-10) + removed image *writing* support + stbi_info support from Jetro Lauha + GIF support from Jean-Marc Lienher + iPhone PNG-extensions from James Brown + warning-fixes from Nicolas Schulz and Janez Zemva (i.stbi__err. Janez (U+017D)emva) + 1.21 fix use of 'stbi_uc' in header (reported by jon blow) + 1.20 added support for Softimage PIC, by Tom Seddon + 1.19 bug in interlaced PNG corruption check (found by ryg) + 1.18 (2008-08-02) + fix a threading bug (local mutable static) + 1.17 support interlaced PNG + 1.16 major bugfix - stbi__convert_format converted one too many pixels + 1.15 initialize some fields for thread safety + 1.14 fix threadsafe conversion bug + header-file-only version (#define STBI_HEADER_FILE_ONLY before including) + 1.13 threadsafe + 1.12 const qualifiers in the API + 1.11 Support installable IDCT, colorspace conversion routines + 1.10 Fixes for 64-bit (don't use "unsigned long") + optimized upsampling by Fabian "ryg" Giesen + 1.09 Fix format-conversion for PSD code (bad global variables!) + 1.08 Thatcher Ulrich's PSD code integrated by Nicolas Schulz + 1.07 attempt to fix C++ warning/errors again + 1.06 attempt to fix C++ warning/errors again + 1.05 fix TGA loading to return correct *comp and use good luminance calc + 1.04 default float alpha is 1, not 255; use 'void *' for stbi_image_free + 1.03 bugfixes to STBI_NO_STDIO, STBI_NO_HDR + 1.02 support for (subset of) HDR files, float interface for preferred access to them + 1.01 fix bug: possible bug in handling right-side up bmps... not sure + fix bug: the stbi__bmp_load() and stbi__tga_load() functions didn't work at all + 1.00 interface to zlib that skips zlib header + 0.99 correct handling of alpha in palette + 0.98 TGA loader by lonesock; dynamically add loaders (untested) + 0.97 jpeg errors on too large a file; also catch another malloc failure + 0.96 fix detection of invalid v value - particleman@mollyrocket forum + 0.95 during header scan, seek to markers in case of padding + 0.94 STBI_NO_STDIO to disable stdio usage; rename all #defines the same + 0.93 handle jpegtran output; verbose errors + 0.92 read 4,8,16,24,32-bit BMP files of several formats + 0.91 output 24-bit Windows 3.0 BMP files + 0.90 fix a few more warnings; bump version number to approach 1.0 + 0.61 bugfixes due to Marc LeBlanc, Christopher Lloyd + 0.60 fix compiling as c++ + 0.59 fix warnings: merge Dave Moore's -Wall fixes + 0.58 fix bug: zlib uncompressed mode len/nlen was wrong endian + 0.57 fix bug: jpg last huffman symbol before marker was >9 bits but less than 16 available + 0.56 fix bug: zlib uncompressed mode len vs. nlen + 0.55 fix bug: restart_interval not initialized to 0 + 0.54 allow NULL for 'int *comp' + 0.53 fix bug in png 3->4; speedup png decoding + 0.52 png handles req_comp=3,4 directly; minor cleanup; jpeg comments + 0.51 obey req_comp requests, 1-component jpegs return as 1-component, + on 'test' only check type, not whether we support this variant + 0.50 (2006-11-19) + first released version + */ + +/* +------------------------------------------------------------------------------ +This software is available under 2 licenses -- choose whichever you prefer. +------------------------------------------------------------------------------ +ALTERNATIVE A - MIT License +Copyright (c) 2017 Sean Barrett +Permission is hereby granted, free of charge, to any person obtaining a copy of +this software and associated documentation files (the "Software"), to deal in +the Software without restriction, including without limitation the rights to +use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies +of the Software, and to permit persons to whom the Software is furnished to do +so, subject to the following conditions: +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. +------------------------------------------------------------------------------ +ALTERNATIVE B - Public Domain (www.unlicense.org) +This is free and unencumbered software released into the public domain. +Anyone is free to copy, modify, publish, use, compile, sell, or distribute this +software, either in source code form or as a compiled binary, for any purpose, +commercial or non-commercial, and by any means. +In jurisdictions that recognize copyright laws, the author or authors of this +software dedicate any and all copyright interest in the software to the public +domain. We make this dedication for the benefit of the public at large and to +the detriment of our heirs and successors. We intend this dedication to be an +overt act of relinquishment in perpetuity of all present and future rights to +this software under copyright law. +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN +ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +------------------------------------------------------------------------------ +*/ \ No newline at end of file From d63230a547e98a34e9781c2776a2d9fec0f803f1 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Mon, 30 Sep 2024 18:12:00 +0200 Subject: [PATCH 088/125] (fix): Why did I commit this --- Backends/stb_image.h | 9310 ------------------------------------------ 1 file changed, 9310 deletions(-) delete mode 100644 Backends/stb_image.h diff --git a/Backends/stb_image.h b/Backends/stb_image.h deleted file mode 100644 index 7a9c7bfac..000000000 --- a/Backends/stb_image.h +++ /dev/null @@ -1,9310 +0,0 @@ -/* stb_image - v2.30 - public domain image loader - http://nothings.org/stb - no warranty implied; use at your own risk - - Do this: - #define STB_IMAGE_IMPLEMENTATION - before you include this file in *one* C or C++ file to create the implementation. - - // i.e. it should look like this: - #include ... - #include ... - #include ... - #define STB_IMAGE_IMPLEMENTATION - #include "stb_image.h" - - You can #define STBI_ASSERT(x) before the #include to avoid using assert.h. - And #define STBI_MALLOC, STBI_REALLOC, and STBI_FREE to avoid using malloc,realloc,free - - - QUICK NOTES: - Primarily of interest to game developers and other people who can - avoid problematic images and only need the trivial interface - - JPEG baseline & progressive (12 bpc/arithmetic not supported, same as stock IJG lib) - PNG 1/2/4/8/16-bit-per-channel - - TGA (not sure what subset, if a subset) - BMP non-1bpp, non-RLE - PSD (composited view only, no extra channels, 8/16 bit-per-channel) - - GIF (*comp always reports as 4-channel) - HDR (radiance rgbE format) - PIC (Softimage PIC) - PNM (PPM and PGM binary only) - - Animated GIF still needs a proper API, but here's one way to do it: - http://gist.github.com/urraka/685d9a6340b26b830d49 - - - decode from memory or through FILE (define STBI_NO_STDIO to remove code) - - decode from arbitrary I/O callbacks - - SIMD acceleration on x86/x64 (SSE2) and ARM (NEON) - - Full documentation under "DOCUMENTATION" below. - - -LICENSE - - See end of file for license information. - -RECENT REVISION HISTORY: - - 2.30 (2024-05-31) avoid erroneous gcc warning - 2.29 (2023-05-xx) optimizations - 2.28 (2023-01-29) many error fixes, security errors, just tons of stuff - 2.27 (2021-07-11) document stbi_info better, 16-bit PNM support, bug fixes - 2.26 (2020-07-13) many minor fixes - 2.25 (2020-02-02) fix warnings - 2.24 (2020-02-02) fix warnings; thread-local failure_reason and flip_vertically - 2.23 (2019-08-11) fix clang static analysis warning - 2.22 (2019-03-04) gif fixes, fix warnings - 2.21 (2019-02-25) fix typo in comment - 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs - 2.19 (2018-02-11) fix warning - 2.18 (2018-01-30) fix warnings - 2.17 (2018-01-29) bugfix, 1-bit BMP, 16-bitness query, fix warnings - 2.16 (2017-07-23) all functions have 16-bit variants; optimizations; bugfixes - 2.15 (2017-03-18) fix png-1,2,4; all Imagenet JPGs; no runtime SSE detection on GCC - 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs - 2.13 (2016-12-04) experimental 16-bit API, only for PNG so far; fixes - 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes - 2.11 (2016-04-02) 16-bit PNGS; enable SSE2 in non-gcc x64 - RGB-format JPEG; remove white matting in PSD; - allocate large structures on the stack; - correct channel count for PNG & BMP - 2.10 (2016-01-22) avoid warning introduced in 2.09 - 2.09 (2016-01-16) 16-bit TGA; comments in PNM files; STBI_REALLOC_SIZED - - See end of file for full revision history. - - - ============================ Contributors ========================= - - Image formats Extensions, features - Sean Barrett (jpeg, png, bmp) Jetro Lauha (stbi_info) - Nicolas Schulz (hdr, psd) Martin "SpartanJ" Golini (stbi_info) - Jonathan Dummer (tga) James "moose2000" Brown (iPhone PNG) - Jean-Marc Lienher (gif) Ben "Disch" Wenger (io callbacks) - Tom Seddon (pic) Omar Cornut (1/2/4-bit PNG) - Thatcher Ulrich (psd) Nicolas Guillemot (vertical flip) - Ken Miller (pgm, ppm) Richard Mitton (16-bit PSD) - github:urraka (animated gif) Junggon Kim (PNM comments) - Christopher Forseth (animated gif) Daniel Gibson (16-bit TGA) - socks-the-fox (16-bit PNG) - Jeremy Sawicki (handle all ImageNet JPGs) - Optimizations & bugfixes Mikhail Morozov (1-bit BMP) - Fabian "ryg" Giesen Anael Seghezzi (is-16-bit query) - Arseny Kapoulkine Simon Breuss (16-bit PNM) - John-Mark Allen - Carmelo J Fdez-Aguera - - Bug & warning fixes - Marc LeBlanc David Woo Guillaume George Martins Mozeiko - Christpher Lloyd Jerry Jansson Joseph Thomson Blazej Dariusz Roszkowski - Phil Jordan Dave Moore Roy Eltham - Hayaki Saito Nathan Reed Won Chun - Luke Graham Johan Duparc Nick Verigakis the Horde3D community - Thomas Ruf Ronny Chevalier github:rlyeh - Janez Zemva John Bartholomew Michal Cichon github:romigrou - Jonathan Blow Ken Hamada Tero Hanninen github:svdijk - Eugene Golushkov Laurent Gomila Cort Stratton github:snagar - Aruelien Pocheville Sergio Gonzalez Thibault Reuille github:Zelex - Cass Everitt Ryamond Barbiero github:grim210 - Paul Du Bois Engin Manap Aldo Culquicondor github:sammyhw - Philipp Wiesemann Dale Weiler Oriol Ferrer Mesia github:phprus - Josh Tobin Neil Bickford Matthew Gregan github:poppolopoppo - Julian Raschke Gregory Mullen Christian Floisand github:darealshinji - Baldur Karlsson Kevin Schmidt JR Smith github:Michaelangel007 - Brad Weinberger Matvey Cherevko github:mosra - Luca Sas Alexander Veselov Zack Middleton [reserved] - Ryan C. Gordon [reserved] [reserved] - DO NOT ADD YOUR NAME HERE - - Jacko Dirks - - To add your name to the credits, pick a random blank space in the middle and fill it. - 80% of merge conflicts on stb PRs are due to people adding their name at the end - of the credits. -*/ - -#ifndef STBI_INCLUDE_STB_IMAGE_H - #define STBI_INCLUDE_STB_IMAGE_H - -// DOCUMENTATION -// -// Limitations: -// - no 12-bit-per-channel JPEG -// - no JPEGs with arithmetic coding -// - GIF always returns *comp=4 -// -// Basic usage (see HDR discussion below for HDR usage): -// int x,y,n; -// unsigned char *data = stbi_load(filename, &x, &y, &n, 0); -// // ... process data if not NULL ... -// // ... x = width, y = height, n = # 8-bit components per pixel ... -// // ... replace '0' with '1'..'4' to force that many components per pixel -// // ... but 'n' will always be the number that it would have been if you said 0 -// stbi_image_free(data); -// -// Standard parameters: -// int *x -- outputs image width in pixels -// int *y -- outputs image height in pixels -// int *channels_in_file -- outputs # of image components in image file -// int desired_channels -- if non-zero, # of image components requested in result -// -// The return value from an image loader is an 'unsigned char *' which points -// to the pixel data, or NULL on an allocation failure or if the image is -// corrupt or invalid. The pixel data consists of *y scanlines of *x pixels, -// with each pixel consisting of N interleaved 8-bit components; the first -// pixel pointed to is top-left-most in the image. There is no padding between -// image scanlines or between pixels, regardless of format. The number of -// components N is 'desired_channels' if desired_channels is non-zero, or -// *channels_in_file otherwise. If desired_channels is non-zero, -// *channels_in_file has the number of components that _would_ have been -// output otherwise. E.g. if you set desired_channels to 4, you will always -// get RGBA output, but you can check *channels_in_file to see if it's trivially -// opaque because e.g. there were only 3 channels in the source image. -// -// An output image with N components has the following components interleaved -// in this order in each pixel: -// -// N=#comp components -// 1 grey -// 2 grey, alpha -// 3 red, green, blue -// 4 red, green, blue, alpha -// -// If image loading fails for any reason, the return value will be NULL, -// and *x, *y, *channels_in_file will be unchanged. The function -// stbi_failure_reason() can be queried for an extremely brief, end-user -// unfriendly explanation of why the load failed. Define STBI_NO_FAILURE_STRINGS -// to avoid compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly -// more user-friendly ones. -// -// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized. -// -// To query the width, height and component count of an image without having to -// decode the full file, you can use the stbi_info family of functions: -// -// int x,y,n,ok; -// ok = stbi_info(filename, &x, &y, &n); -// // returns ok=1 and sets x, y, n if image is a supported format, -// // 0 otherwise. -// -// Note that stb_image pervasively uses ints in its public API for sizes, -// including sizes of memory buffers. This is now part of the API and thus -// hard to change without causing breakage. As a result, the various image -// loaders all have certain limits on image size; these differ somewhat -// by format but generally boil down to either just under 2GB or just under -// 1GB. When the decoded image would be larger than this, stb_image decoding -// will fail. -// -// Additionally, stb_image will reject image files that have any of their -// dimensions set to a larger value than the configurable STBI_MAX_DIMENSIONS, -// which defaults to 2**24 = 16777216 pixels. Due to the above memory limit, -// the only way to have an image with such dimensions load correctly -// is for it to have a rather extreme aspect ratio. Either way, the -// assumption here is that such larger images are likely to be malformed -// or malicious. If you do need to load an image with individual dimensions -// larger than that, and it still fits in the overall size limit, you can -// #define STBI_MAX_DIMENSIONS on your own to be something larger. -// -// =========================================================================== -// -// UNICODE: -// -// If compiling for Windows and you wish to use Unicode filenames, compile -// with -// #define STBI_WINDOWS_UTF8 -// and pass utf8-encoded filenames. Call stbi_convert_wchar_to_utf8 to convert -// Windows wchar_t filenames to utf8. -// -// =========================================================================== -// -// Philosophy -// -// stb libraries are designed with the following priorities: -// -// 1. easy to use -// 2. easy to maintain -// 3. good performance -// -// Sometimes I let "good performance" creep up in priority over "easy to maintain", -// and for best performance I may provide less-easy-to-use APIs that give higher -// performance, in addition to the easy-to-use ones. Nevertheless, it's important -// to keep in mind that from the standpoint of you, a client of this library, -// all you care about is #1 and #3, and stb libraries DO NOT emphasize #3 above all. -// -// Some secondary priorities arise directly from the first two, some of which -// provide more explicit reasons why performance can't be emphasized. -// -// - Portable ("ease of use") -// - Small source code footprint ("easy to maintain") -// - No dependencies ("ease of use") -// -// =========================================================================== -// -// I/O callbacks -// -// I/O callbacks allow you to read from arbitrary sources, like packaged -// files or some other source. Data read from callbacks are processed -// through a small internal buffer (currently 128 bytes) to try to reduce -// overhead. -// -// The three functions you must define are "read" (reads some bytes of data), -// "skip" (skips some bytes of data), "eof" (reports if the stream is at the end). -// -// =========================================================================== -// -// SIMD support -// -// The JPEG decoder will try to automatically use SIMD kernels on x86 when -// supported by the compiler. For ARM Neon support, you must explicitly -// request it. -// -// (The old do-it-yourself SIMD API is no longer supported in the current -// code.) -// -// On x86, SSE2 will automatically be used when available based on a run-time -// test; if not, the generic C versions are used as a fall-back. On ARM targets, -// the typical path is to have separate builds for NEON and non-NEON devices -// (at least this is true for iOS and Android). Therefore, the NEON support is -// toggled by a build flag: define STBI_NEON to get NEON loops. -// -// If for some reason you do not want to use any of SIMD code, or if -// you have issues compiling it, you can disable it entirely by -// defining STBI_NO_SIMD. -// -// =========================================================================== -// -// HDR image support (disable by defining STBI_NO_HDR) -// -// stb_image supports loading HDR images in general, and currently the Radiance -// .HDR file format specifically. You can still load any file through the existing -// interface; if you attempt to load an HDR file, it will be automatically remapped -// to LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1; -// both of these constants can be reconfigured through this interface: -// -// stbi_hdr_to_ldr_gamma(2.2f); -// stbi_hdr_to_ldr_scale(1.0f); -// -// (note, do not use _inverse_ constants; stbi_image will invert them -// appropriately). -// -// Additionally, there is a new, parallel interface for loading files as -// (linear) floats to preserve the full dynamic range: -// -// float *data = stbi_loadf(filename, &x, &y, &n, 0); -// -// If you load LDR images through this interface, those images will -// be promoted to floating point values, run through the inverse of -// constants corresponding to the above: -// -// stbi_ldr_to_hdr_scale(1.0f); -// stbi_ldr_to_hdr_gamma(2.2f); -// -// Finally, given a filename (or an open file or memory block--see header -// file for details) containing image data, you can query for the "most -// appropriate" interface to use (that is, whether the image is HDR or -// not), using: -// -// stbi_is_hdr(char *filename); -// -// =========================================================================== -// -// iPhone PNG support: -// -// We optionally support converting iPhone-formatted PNGs (which store -// premultiplied BGRA) back to RGB, even though they're internally encoded -// differently. To enable this conversion, call -// stbi_convert_iphone_png_to_rgb(1). -// -// Call stbi_set_unpremultiply_on_load(1) as well to force a divide per -// pixel to remove any premultiplied alpha *only* if the image file explicitly -// says there's premultiplied data (currently only happens in iPhone images, -// and only if iPhone convert-to-rgb processing is on). -// -// =========================================================================== -// -// ADDITIONAL CONFIGURATION -// -// - You can suppress implementation of any of the decoders to reduce -// your code footprint by #defining one or more of the following -// symbols before creating the implementation. -// -// STBI_NO_JPEG -// STBI_NO_PNG -// STBI_NO_BMP -// STBI_NO_PSD -// STBI_NO_TGA -// STBI_NO_GIF -// STBI_NO_HDR -// STBI_NO_PIC -// STBI_NO_PNM (.ppm and .pgm) -// -// - You can request *only* certain decoders and suppress all other ones -// (this will be more forward-compatible, as addition of new decoders -// doesn't require you to disable them explicitly): -// -// STBI_ONLY_JPEG -// STBI_ONLY_PNG -// STBI_ONLY_BMP -// STBI_ONLY_PSD -// STBI_ONLY_TGA -// STBI_ONLY_GIF -// STBI_ONLY_HDR -// STBI_ONLY_PIC -// STBI_ONLY_PNM (.ppm and .pgm) -// -// - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still -// want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB -// -// - If you define STBI_MAX_DIMENSIONS, stb_image will reject images greater -// than that size (in either width or height) without further processing. -// This is to let programs in the wild set an upper bound to prevent -// denial-of-service attacks on untrusted data, as one could generate a -// valid image of gigantic dimensions and force stb_image to allocate a -// huge block of memory and spend disproportionate time decoding it. By -// default this is set to (1 << 24), which is 16777216, but that's still -// very big. - - #ifndef STBI_NO_STDIO - #include - #endif // STBI_NO_STDIO - - #define STBI_VERSION 1 - -enum { - STBI_default = 0, // only used for desired_channels - - STBI_grey = 1, - STBI_grey_alpha = 2, - STBI_rgb = 3, - STBI_rgb_alpha = 4 -}; - - #include -typedef unsigned char stbi_uc; -typedef unsigned short stbi_us; - - #ifdef __cplusplus -extern "C" { - #endif - - #ifndef STBIDEF - #ifdef STB_IMAGE_STATIC - #define STBIDEF static - #else - #define STBIDEF extern - #endif - #endif - -////////////////////////////////////////////////////////////////////////////// -// -// PRIMARY API - works on images of any type -// - -// -// load image by filename, open file, or memory buffer -// - -typedef struct { - int (*read)(void* user, char* data, int size); // fill 'data' with 'size' bytes. return number of bytes actually read - void (*skip)(void* user, int n); // skip the next 'n' bytes, or 'unget' the last -n bytes if negative - int (*eof)(void* user); // returns nonzero if we are at end of file/data -} stbi_io_callbacks; - -//////////////////////////////////// -// -// 8-bits-per-channel interface -// - -STBIDEF stbi_uc* stbi_load_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels); -STBIDEF stbi_uc* stbi_load_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels); - - #ifndef STBI_NO_STDIO -STBIDEF stbi_uc* stbi_load(const char* filename, int* x, int* y, int* channels_in_file, int desired_channels); -STBIDEF stbi_uc* stbi_load_from_file(FILE* f, int* x, int* y, int* channels_in_file, int desired_channels); - // for stbi_load_from_file, file pointer is left pointing immediately after image - #endif - - #ifndef STBI_NO_GIF -STBIDEF stbi_uc* stbi_load_gif_from_memory(const stbi_uc* buffer, int len, int** delays, int* x, int* y, int* z, int* comp, int req_comp); - #endif - - #ifdef STBI_WINDOWS_UTF8 -STBIDEF int stbi_convert_wchar_to_utf8(char* buffer, size_t bufferlen, const wchar_t* input); - #endif - -//////////////////////////////////// -// -// 16-bits-per-channel interface -// - -STBIDEF stbi_us* stbi_load_16_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels); -STBIDEF stbi_us* stbi_load_16_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels); - - #ifndef STBI_NO_STDIO -STBIDEF stbi_us* stbi_load_16(const char* filename, int* x, int* y, int* channels_in_file, int desired_channels); -STBIDEF stbi_us* stbi_load_from_file_16(FILE* f, int* x, int* y, int* channels_in_file, int desired_channels); - #endif - - //////////////////////////////////// - // - // float-per-channel interface - // - #ifndef STBI_NO_LINEAR -STBIDEF float* stbi_loadf_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels); -STBIDEF float* stbi_loadf_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels); - - #ifndef STBI_NO_STDIO -STBIDEF float* stbi_loadf(const char* filename, int* x, int* y, int* channels_in_file, int desired_channels); -STBIDEF float* stbi_loadf_from_file(FILE* f, int* x, int* y, int* channels_in_file, int desired_channels); - #endif - #endif - - #ifndef STBI_NO_HDR -STBIDEF void stbi_hdr_to_ldr_gamma(float gamma); -STBIDEF void stbi_hdr_to_ldr_scale(float scale); - #endif // STBI_NO_HDR - - #ifndef STBI_NO_LINEAR -STBIDEF void stbi_ldr_to_hdr_gamma(float gamma); -STBIDEF void stbi_ldr_to_hdr_scale(float scale); - #endif // STBI_NO_LINEAR - -// stbi_is_hdr is always defined, but always returns false if STBI_NO_HDR -STBIDEF int stbi_is_hdr_from_callbacks(const stbi_io_callbacks* clbk, void* user); -STBIDEF int stbi_is_hdr_from_memory(const stbi_uc* buffer, int len); - #ifndef STBI_NO_STDIO -STBIDEF int stbi_is_hdr(const char* filename); -STBIDEF int stbi_is_hdr_from_file(FILE* f); - #endif // STBI_NO_STDIO - -// get a VERY brief reason for failure -// on most compilers (and ALL modern mainstream compilers) this is threadsafe -STBIDEF const char* stbi_failure_reason(void); - -// free the loaded image -- this is just free() -STBIDEF void stbi_image_free(void* retval_from_stbi_load); - -// get image dimensions & components without fully decoding -STBIDEF int stbi_info_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp); -STBIDEF int stbi_info_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* comp); -STBIDEF int stbi_is_16_bit_from_memory(const stbi_uc* buffer, int len); -STBIDEF int stbi_is_16_bit_from_callbacks(const stbi_io_callbacks* clbk, void* user); - - #ifndef STBI_NO_STDIO -STBIDEF int stbi_info(const char* filename, int* x, int* y, int* comp); -STBIDEF int stbi_info_from_file(FILE* f, int* x, int* y, int* comp); -STBIDEF int stbi_is_16_bit(const char* filename); -STBIDEF int stbi_is_16_bit_from_file(FILE* f); - #endif - -// for image formats that explicitly notate that they have premultiplied alpha, -// we just return the colors as stored in the file. set this flag to force -// unpremultiplication. results are undefined if the unpremultiply overflow. -STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply); - -// indicate whether we should process iphone images back to canonical format, -// or just pass them through "as-is" -STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert); - -// flip the image vertically, so the first pixel in the output array is the bottom left -STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip); - -// as above, but only applies to images loaded on the thread that calls the function -// this function is only available if your compiler supports thread-local variables; -// calling it will fail to link if your compiler doesn't -STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply); -STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert); -STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip); - -// ZLIB client - used by PNG, available for other purposes - -STBIDEF char* stbi_zlib_decode_malloc_guesssize(const char* buffer, int len, int initial_size, int* outlen); -STBIDEF char* stbi_zlib_decode_malloc_guesssize_headerflag(const char* buffer, int len, int initial_size, int* outlen, int parse_header); -STBIDEF char* stbi_zlib_decode_malloc(const char* buffer, int len, int* outlen); -STBIDEF int stbi_zlib_decode_buffer(char* obuffer, int olen, const char* ibuffer, int ilen); - -STBIDEF char* stbi_zlib_decode_noheader_malloc(const char* buffer, int len, int* outlen); -STBIDEF int stbi_zlib_decode_noheader_buffer(char* obuffer, int olen, const char* ibuffer, int ilen); - - #ifdef __cplusplus -} - #endif - -// -// -//// end header file ///////////////////////////////////////////////////// -#endif // STBI_INCLUDE_STB_IMAGE_H - -#ifdef STB_IMAGE_IMPLEMENTATION - - #if defined(STBI_ONLY_JPEG) || defined(STBI_ONLY_PNG) || defined(STBI_ONLY_BMP) || defined(STBI_ONLY_TGA) || defined(STBI_ONLY_GIF) || \ - defined(STBI_ONLY_PSD) || defined(STBI_ONLY_HDR) || defined(STBI_ONLY_PIC) || defined(STBI_ONLY_PNM) || defined(STBI_ONLY_ZLIB) - #ifndef STBI_ONLY_JPEG - #define STBI_NO_JPEG - #endif - #ifndef STBI_ONLY_PNG - #define STBI_NO_PNG - #endif - #ifndef STBI_ONLY_BMP - #define STBI_NO_BMP - #endif - #ifndef STBI_ONLY_PSD - #define STBI_NO_PSD - #endif - #ifndef STBI_ONLY_TGA - #define STBI_NO_TGA - #endif - #ifndef STBI_ONLY_GIF - #define STBI_NO_GIF - #endif - #ifndef STBI_ONLY_HDR - #define STBI_NO_HDR - #endif - #ifndef STBI_ONLY_PIC - #define STBI_NO_PIC - #endif - #ifndef STBI_ONLY_PNM - #define STBI_NO_PNM - #endif - #endif - - #if defined(STBI_NO_PNG) && !defined(STBI_SUPPORT_ZLIB) && !defined(STBI_NO_ZLIB) - #define STBI_NO_ZLIB - #endif - - #include - #include - #include // ptrdiff_t on osx - #include - #include - - #if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) - #include // ldexp, pow - #endif - - #ifndef STBI_NO_STDIO - #include - #endif - - #ifndef STBI_ASSERT - #include - #define STBI_ASSERT(x) assert(x) - #endif - - #ifdef __cplusplus - #define STBI_EXTERN extern "C" - #else - #define STBI_EXTERN extern - #endif - - #ifndef _MSC_VER - #ifdef __cplusplus - #define stbi_inline inline - #else - #define stbi_inline - #endif - #else - #define stbi_inline __forceinline - #endif - - #ifndef STBI_NO_THREAD_LOCALS - #if defined(__cplusplus) && __cplusplus >= 201103L - #define STBI_THREAD_LOCAL thread_local - #elif defined(__GNUC__) && __GNUC__ < 5 - #define STBI_THREAD_LOCAL __thread - #elif defined(_MSC_VER) - #define STBI_THREAD_LOCAL __declspec(thread) - #elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L && !defined(__STDC_NO_THREADS__) - #define STBI_THREAD_LOCAL _Thread_local - #endif - - #ifndef STBI_THREAD_LOCAL - #if defined(__GNUC__) - #define STBI_THREAD_LOCAL __thread - #endif - #endif - #endif - - #if defined(_MSC_VER) || defined(__SYMBIAN32__) -typedef unsigned short stbi__uint16; -typedef signed short stbi__int16; -typedef unsigned int stbi__uint32; -typedef signed int stbi__int32; - #else - #include -typedef uint16_t stbi__uint16; -typedef int16_t stbi__int16; -typedef uint32_t stbi__uint32; -typedef int32_t stbi__int32; - #endif - -// should produce compiler error if size is wrong -typedef unsigned char validate_uint32[sizeof(stbi__uint32) == 4 ? 1 : -1]; - - #ifdef _MSC_VER - #define STBI_NOTUSED(v) (void)(v) - #else - #define STBI_NOTUSED(v) (void)sizeof(v) - #endif - - #ifdef _MSC_VER - #define STBI_HAS_LROTL - #endif - - #ifdef STBI_HAS_LROTL - #define stbi_lrot(x, y) _lrotl(x, y) - #else - #define stbi_lrot(x, y) (((x) << (y)) | ((x) >> (-(y) & 31))) - #endif - - #if defined(STBI_MALLOC) && defined(STBI_FREE) && (defined(STBI_REALLOC) || defined(STBI_REALLOC_SIZED)) - // ok - #elif !defined(STBI_MALLOC) && !defined(STBI_FREE) && !defined(STBI_REALLOC) && !defined(STBI_REALLOC_SIZED) - // ok - #else - #error "Must define all or none of STBI_MALLOC, STBI_FREE, and STBI_REALLOC (or STBI_REALLOC_SIZED)." - #endif - - #ifndef STBI_MALLOC - #define STBI_MALLOC(sz) malloc(sz) - #define STBI_REALLOC(p, newsz) realloc(p, newsz) - #define STBI_FREE(p) free(p) - #endif - - #ifndef STBI_REALLOC_SIZED - #define STBI_REALLOC_SIZED(p, oldsz, newsz) STBI_REALLOC(p, newsz) - #endif - - // x86/x64 detection - #if defined(__x86_64__) || defined(_M_X64) - #define STBI__X64_TARGET - #elif defined(__i386) || defined(_M_IX86) - #define STBI__X86_TARGET - #endif - - #if defined(__GNUC__) && defined(STBI__X86_TARGET) && !defined(__SSE2__) && !defined(STBI_NO_SIMD) - // gcc doesn't support sse2 intrinsics unless you compile with -msse2, - // which in turn means it gets to use SSE2 everywhere. This is unfortunate, - // but previous attempts to provide the SSE2 functions with runtime - // detection caused numerous issues. The way architecture extensions are - // exposed in GCC/Clang is, sadly, not really suited for one-file libs. - // New behavior: if compiled with -msse2, we use SSE2 without any - // detection; if not, we don't use it at all. - #define STBI_NO_SIMD - #endif - - #if defined(__MINGW32__) && defined(STBI__X86_TARGET) && !defined(STBI_MINGW_ENABLE_SSE2) && !defined(STBI_NO_SIMD) - // Note that __MINGW32__ doesn't actually mean 32-bit, so we have to avoid STBI__X64_TARGET - // - // 32-bit MinGW wants ESP to be 16-byte aligned, but this is not in the - // Windows ABI and VC++ as well as Windows DLLs don't maintain that invariant. - // As a result, enabling SSE2 on 32-bit MinGW is dangerous when not - // simultaneously enabling "-mstackrealign". - // - // See https://github.com/nothings/stb/issues/81 for more information. - // - // So default to no SSE2 on 32-bit MinGW. If you've read this far and added - // -mstackrealign to your build settings, feel free to #define STBI_MINGW_ENABLE_SSE2. - #define STBI_NO_SIMD - #endif - - #if !defined(STBI_NO_SIMD) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET)) - #define STBI_SSE2 - #include - - #ifdef _MSC_VER - - #if _MSC_VER >= 1400 // not VC6 - #include // __cpuid -static int stbi__cpuid3(void) -{ - int info[4]; - __cpuid(info, 1); - return info[3]; -} - #else -static int stbi__cpuid3(void) -{ - int res; - __asm { - mov eax,1 - cpuid - mov res,edx - } - return res; -} - #endif - - #define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name - - #if !defined(STBI_NO_JPEG) && defined(STBI_SSE2) -static int stbi__sse2_available(void) -{ - int info3 = stbi__cpuid3(); - return ((info3 >> 26) & 1) != 0; -} - #endif - - #else // assume GCC-style if not VC++ - #define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16))) - - #if !defined(STBI_NO_JPEG) && defined(STBI_SSE2) -static int stbi__sse2_available(void) -{ - // If we're even attempting to compile this on GCC/Clang, that means - // -msse2 is on, which means the compiler is allowed to use SSE2 - // instructions at will, and so are we. - return 1; -} - #endif - - #endif - #endif - - // ARM NEON - #if defined(STBI_NO_SIMD) && defined(STBI_NEON) - #undef STBI_NEON - #endif - - #ifdef STBI_NEON - #include - #ifdef _MSC_VER - #define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name - #else - #define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16))) - #endif - #endif - - #ifndef STBI_SIMD_ALIGN - #define STBI_SIMD_ALIGN(type, name) type name - #endif - - #ifndef STBI_MAX_DIMENSIONS - #define STBI_MAX_DIMENSIONS (1 << 24) - #endif - -/////////////////////////////////////////////// -// -// stbi__context struct and start_xxx functions - -// stbi__context structure is our basic context used by all images, so it -// contains all the IO context, plus some basic image information -typedef struct { - stbi__uint32 img_x, img_y; - int img_n, img_out_n; - - stbi_io_callbacks io; - void* io_user_data; - - int read_from_callbacks; - int buflen; - stbi_uc buffer_start[128]; - int callback_already_read; - - stbi_uc *img_buffer, *img_buffer_end; - stbi_uc *img_buffer_original, *img_buffer_original_end; -} stbi__context; - -static void stbi__refill_buffer(stbi__context* s); - -// initialize a memory-decode context -static void stbi__start_mem(stbi__context* s, const stbi_uc* buffer, int len) -{ - s->io.read = NULL; - s->read_from_callbacks = 0; - s->callback_already_read = 0; - s->img_buffer = s->img_buffer_original = (stbi_uc*)buffer; - s->img_buffer_end = s->img_buffer_original_end = (stbi_uc*)buffer + len; -} - -// initialize a callback-based context -static void stbi__start_callbacks(stbi__context* s, stbi_io_callbacks* c, void* user) -{ - s->io = *c; - s->io_user_data = user; - s->buflen = sizeof(s->buffer_start); - s->read_from_callbacks = 1; - s->callback_already_read = 0; - s->img_buffer = s->img_buffer_original = s->buffer_start; - stbi__refill_buffer(s); - s->img_buffer_original_end = s->img_buffer_end; -} - - #ifndef STBI_NO_STDIO - -static int stbi__stdio_read(void* user, char* data, int size) -{ - return (int)fread(data, 1, size, (FILE*)user); -} - -static void stbi__stdio_skip(void* user, int n) -{ - int ch; - fseek((FILE*)user, n, SEEK_CUR); - ch = fgetc((FILE*)user); /* have to read a byte to reset feof()'s flag */ - if (ch != EOF) - { - ungetc(ch, (FILE*)user); /* push byte back onto stream if valid. */ - } -} - -static int stbi__stdio_eof(void* user) -{ - return feof((FILE*)user) || ferror((FILE*)user); -} - -static stbi_io_callbacks stbi__stdio_callbacks = { - stbi__stdio_read, - stbi__stdio_skip, - stbi__stdio_eof, -}; - -static void stbi__start_file(stbi__context* s, FILE* f) -{ - stbi__start_callbacks(s, &stbi__stdio_callbacks, (void*)f); -} - - // static void stop_file(stbi__context *s) { } - - #endif // !STBI_NO_STDIO - -static void stbi__rewind(stbi__context* s) -{ - // conceptually rewind SHOULD rewind to the beginning of the stream, - // but we just rewind to the beginning of the initial buffer, because - // we only use it after doing 'test', which only ever looks at at most 92 bytes - s->img_buffer = s->img_buffer_original; - s->img_buffer_end = s->img_buffer_original_end; -} - -enum { STBI_ORDER_RGB, STBI_ORDER_BGR }; - -typedef struct { - int bits_per_channel; - int num_channels; - int channel_order; -} stbi__result_info; - - #ifndef STBI_NO_JPEG -static int stbi__jpeg_test(stbi__context* s); -static void* stbi__jpeg_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static int stbi__jpeg_info(stbi__context* s, int* x, int* y, int* comp); - #endif - - #ifndef STBI_NO_PNG -static int stbi__png_test(stbi__context* s); -static void* stbi__png_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static int stbi__png_info(stbi__context* s, int* x, int* y, int* comp); -static int stbi__png_is16(stbi__context* s); - #endif - - #ifndef STBI_NO_BMP -static int stbi__bmp_test(stbi__context* s); -static void* stbi__bmp_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static int stbi__bmp_info(stbi__context* s, int* x, int* y, int* comp); - #endif - - #ifndef STBI_NO_TGA -static int stbi__tga_test(stbi__context* s); -static void* stbi__tga_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static int stbi__tga_info(stbi__context* s, int* x, int* y, int* comp); - #endif - - #ifndef STBI_NO_PSD -static int stbi__psd_test(stbi__context* s); -static void* stbi__psd_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri, int bpc); -static int stbi__psd_info(stbi__context* s, int* x, int* y, int* comp); -static int stbi__psd_is16(stbi__context* s); - #endif - - #ifndef STBI_NO_HDR -static int stbi__hdr_test(stbi__context* s); -static float* stbi__hdr_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static int stbi__hdr_info(stbi__context* s, int* x, int* y, int* comp); - #endif - - #ifndef STBI_NO_PIC -static int stbi__pic_test(stbi__context* s); -static void* stbi__pic_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static int stbi__pic_info(stbi__context* s, int* x, int* y, int* comp); - #endif - - #ifndef STBI_NO_GIF -static int stbi__gif_test(stbi__context* s); -static void* stbi__gif_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static void* stbi__load_gif_main(stbi__context* s, int** delays, int* x, int* y, int* z, int* comp, int req_comp); -static int stbi__gif_info(stbi__context* s, int* x, int* y, int* comp); - #endif - - #ifndef STBI_NO_PNM -static int stbi__pnm_test(stbi__context* s); -static void* stbi__pnm_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri); -static int stbi__pnm_info(stbi__context* s, int* x, int* y, int* comp); -static int stbi__pnm_is16(stbi__context* s); - #endif - -static - #ifdef STBI_THREAD_LOCAL - STBI_THREAD_LOCAL - #endif - const char* stbi__g_failure_reason; - -STBIDEF const char* stbi_failure_reason(void) -{ - return stbi__g_failure_reason; -} - - #ifndef STBI_NO_FAILURE_STRINGS -static int stbi__err(const char* str) -{ - stbi__g_failure_reason = str; - return 0; -} - #endif - -static void* stbi__malloc(size_t size) -{ - return STBI_MALLOC(size); -} - -// stb_image uses ints pervasively, including for offset calculations. -// therefore the largest decoded image size we can support with the -// current code, even on 64-bit targets, is INT_MAX. this is not a -// significant limitation for the intended use case. -// -// we do, however, need to make sure our size calculations don't -// overflow. hence a few helper functions for size calculations that -// multiply integers together, making sure that they're non-negative -// and no overflow occurs. - -// return 1 if the sum is valid, 0 on overflow. -// negative terms are considered invalid. -static int stbi__addsizes_valid(int a, int b) -{ - if (b < 0) - return 0; - // now 0 <= b <= INT_MAX, hence also - // 0 <= INT_MAX - b <= INTMAX. - // And "a + b <= INT_MAX" (which might overflow) is the - // same as a <= INT_MAX - b (no overflow) - return a <= INT_MAX - b; -} - -// returns 1 if the product is valid, 0 on overflow. -// negative factors are considered invalid. -static int stbi__mul2sizes_valid(int a, int b) -{ - if (a < 0 || b < 0) - return 0; - if (b == 0) - return 1; // mul-by-0 is always safe - // portable way to check for no overflows in a*b - return a <= INT_MAX / b; -} - - #if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR) -// returns 1 if "a*b + add" has no negative terms/factors and doesn't overflow -static int stbi__mad2sizes_valid(int a, int b, int add) -{ - return stbi__mul2sizes_valid(a, b) && stbi__addsizes_valid(a * b, add); -} - #endif - -// returns 1 if "a*b*c + add" has no negative terms/factors and doesn't overflow -static int stbi__mad3sizes_valid(int a, int b, int c, int add) -{ - return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a * b, c) && stbi__addsizes_valid(a * b * c, add); -} - - // returns 1 if "a*b*c*d + add" has no negative terms/factors and doesn't overflow - #if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM) -static int stbi__mad4sizes_valid(int a, int b, int c, int d, int add) -{ - return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a * b, c) && stbi__mul2sizes_valid(a * b * c, d) && - stbi__addsizes_valid(a * b * c * d, add); -} - #endif - - #if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR) -// mallocs with size overflow checking -static void* stbi__malloc_mad2(int a, int b, int add) -{ - if (!stbi__mad2sizes_valid(a, b, add)) - return NULL; - return stbi__malloc(a * b + add); -} - #endif - -static void* stbi__malloc_mad3(int a, int b, int c, int add) -{ - if (!stbi__mad3sizes_valid(a, b, c, add)) - return NULL; - return stbi__malloc(a * b * c + add); -} - - #if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM) -static void* stbi__malloc_mad4(int a, int b, int c, int d, int add) -{ - if (!stbi__mad4sizes_valid(a, b, c, d, add)) - return NULL; - return stbi__malloc(a * b * c * d + add); -} - #endif - -// returns 1 if the sum of two signed ints is valid (between -2^31 and 2^31-1 inclusive), 0 on overflow. -static int stbi__addints_valid(int a, int b) -{ - if ((a >= 0) != (b >= 0)) - return 1; // a and b have different signs, so no overflow - if (a < 0 && b < 0) - return a >= INT_MIN - b; // same as a + b >= INT_MIN; INT_MIN - b cannot overflow since b < 0. - return a <= INT_MAX - b; -} - -// returns 1 if the product of two ints fits in a signed short, 0 on overflow. -static int stbi__mul2shorts_valid(int a, int b) -{ - if (b == 0 || b == -1) - return 1; // multiplication by 0 is always 0; check for -1 so SHRT_MIN/b doesn't overflow - if ((a >= 0) == (b >= 0)) - return a <= SHRT_MAX / b; // product is positive, so similar to mul2sizes_valid - if (b < 0) - return a <= SHRT_MIN / b; // same as a * b >= SHRT_MIN - return a >= SHRT_MIN / b; -} - -// stbi__err - error -// stbi__errpf - error returning pointer to float -// stbi__errpuc - error returning pointer to unsigned char - - #ifdef STBI_NO_FAILURE_STRINGS - #define stbi__err(x, y) 0 - #elif defined(STBI_FAILURE_USERMSG) - #define stbi__err(x, y) stbi__err(y) - #else - #define stbi__err(x, y) stbi__err(x) - #endif - - #define stbi__errpf(x, y) ((float*)(size_t)(stbi__err(x, y) ? NULL : NULL)) - #define stbi__errpuc(x, y) ((unsigned char*)(size_t)(stbi__err(x, y) ? NULL : NULL)) - -STBIDEF void stbi_image_free(void* retval_from_stbi_load) -{ - STBI_FREE(retval_from_stbi_load); -} - - #ifndef STBI_NO_LINEAR -static float* stbi__ldr_to_hdr(stbi_uc* data, int x, int y, int comp); - #endif - - #ifndef STBI_NO_HDR -static stbi_uc* stbi__hdr_to_ldr(float* data, int x, int y, int comp); - #endif - -static int stbi__vertically_flip_on_load_global = 0; - -STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip) -{ - stbi__vertically_flip_on_load_global = flag_true_if_should_flip; -} - - #ifndef STBI_THREAD_LOCAL - #define stbi__vertically_flip_on_load stbi__vertically_flip_on_load_global - #else -static STBI_THREAD_LOCAL int stbi__vertically_flip_on_load_local, stbi__vertically_flip_on_load_set; - -STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip) -{ - stbi__vertically_flip_on_load_local = flag_true_if_should_flip; - stbi__vertically_flip_on_load_set = 1; -} - - #define stbi__vertically_flip_on_load \ - (stbi__vertically_flip_on_load_set ? stbi__vertically_flip_on_load_local : stbi__vertically_flip_on_load_global) - #endif // STBI_THREAD_LOCAL - -static void* stbi__load_main(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri, int bpc) -{ - memset(ri, 0, sizeof(*ri)); // make sure it's initialized if we add new fields - ri->bits_per_channel = 8; // default is 8 so most paths don't have to be changed - ri->channel_order = STBI_ORDER_RGB; // all current input & output are this, but this is here so we can add BGR order - ri->num_channels = 0; - - // test the formats with a very explicit header first (at least a FOURCC - // or distinctive magic number first) - #ifndef STBI_NO_PNG - if (stbi__png_test(s)) - return stbi__png_load(s, x, y, comp, req_comp, ri); - #endif - #ifndef STBI_NO_BMP - if (stbi__bmp_test(s)) - return stbi__bmp_load(s, x, y, comp, req_comp, ri); - #endif - #ifndef STBI_NO_GIF - if (stbi__gif_test(s)) - return stbi__gif_load(s, x, y, comp, req_comp, ri); - #endif - #ifndef STBI_NO_PSD - if (stbi__psd_test(s)) - return stbi__psd_load(s, x, y, comp, req_comp, ri, bpc); - #else - STBI_NOTUSED(bpc); - #endif - #ifndef STBI_NO_PIC - if (stbi__pic_test(s)) - return stbi__pic_load(s, x, y, comp, req_comp, ri); - #endif - - // then the formats that can end up attempting to load with just 1 or 2 - // bytes matching expectations; these are prone to false positives, so - // try them later - #ifndef STBI_NO_JPEG - if (stbi__jpeg_test(s)) - return stbi__jpeg_load(s, x, y, comp, req_comp, ri); - #endif - #ifndef STBI_NO_PNM - if (stbi__pnm_test(s)) - return stbi__pnm_load(s, x, y, comp, req_comp, ri); - #endif - - #ifndef STBI_NO_HDR - if (stbi__hdr_test(s)) - { - float* hdr = stbi__hdr_load(s, x, y, comp, req_comp, ri); - return stbi__hdr_to_ldr(hdr, *x, *y, req_comp ? req_comp : *comp); - } - #endif - - #ifndef STBI_NO_TGA - // test tga last because it's a crappy test! - if (stbi__tga_test(s)) - return stbi__tga_load(s, x, y, comp, req_comp, ri); - #endif - - return stbi__errpuc("unknown image type", "Image not of any known type, or corrupt"); -} - -static stbi_uc* stbi__convert_16_to_8(stbi__uint16* orig, int w, int h, int channels) -{ - int i; - int img_len = w * h * channels; - stbi_uc* reduced; - - reduced = (stbi_uc*)stbi__malloc(img_len); - if (reduced == NULL) - return stbi__errpuc("outofmem", "Out of memory"); - - for (i = 0; i < img_len; ++i) - reduced[i] = (stbi_uc)((orig[i] >> 8) & 0xFF); // top half of each byte is sufficient approx of 16->8 bit scaling - - STBI_FREE(orig); - return reduced; -} - -static stbi__uint16* stbi__convert_8_to_16(stbi_uc* orig, int w, int h, int channels) -{ - int i; - int img_len = w * h * channels; - stbi__uint16* enlarged; - - enlarged = (stbi__uint16*)stbi__malloc(img_len * 2); - if (enlarged == NULL) - return (stbi__uint16*)stbi__errpuc("outofmem", "Out of memory"); - - for (i = 0; i < img_len; ++i) - enlarged[i] = (stbi__uint16)((orig[i] << 8) + orig[i]); // replicate to high and low byte, maps 0->0, 255->0xffff - - STBI_FREE(orig); - return enlarged; -} - -static void stbi__vertical_flip(void* image, int w, int h, int bytes_per_pixel) -{ - int row; - size_t bytes_per_row = (size_t)w * bytes_per_pixel; - stbi_uc temp[2048]; - stbi_uc* bytes = (stbi_uc*)image; - - for (row = 0; row < (h >> 1); row++) - { - stbi_uc* row0 = bytes + row * bytes_per_row; - stbi_uc* row1 = bytes + (h - row - 1) * bytes_per_row; - // swap row0 with row1 - size_t bytes_left = bytes_per_row; - while (bytes_left) - { - size_t bytes_copy = (bytes_left < sizeof(temp)) ? bytes_left : sizeof(temp); - memcpy(temp, row0, bytes_copy); - memcpy(row0, row1, bytes_copy); - memcpy(row1, temp, bytes_copy); - row0 += bytes_copy; - row1 += bytes_copy; - bytes_left -= bytes_copy; - } - } -} - - #ifndef STBI_NO_GIF -static void stbi__vertical_flip_slices(void* image, int w, int h, int z, int bytes_per_pixel) -{ - int slice; - int slice_size = w * h * bytes_per_pixel; - - stbi_uc* bytes = (stbi_uc*)image; - for (slice = 0; slice < z; ++slice) - { - stbi__vertical_flip(bytes, w, h, bytes_per_pixel); - bytes += slice_size; - } -} - #endif - -static unsigned char* stbi__load_and_postprocess_8bit(stbi__context* s, int* x, int* y, int* comp, int req_comp) -{ - stbi__result_info ri; - void* result = stbi__load_main(s, x, y, comp, req_comp, &ri, 8); - - if (result == NULL) - return NULL; - - // it is the responsibility of the loaders to make sure we get either 8 or 16 bit. - STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16); - - if (ri.bits_per_channel != 8) - { - result = stbi__convert_16_to_8((stbi__uint16*)result, *x, *y, req_comp == 0 ? *comp : req_comp); - ri.bits_per_channel = 8; - } - - // @TODO: move stbi__convert_format to here - - if (stbi__vertically_flip_on_load) - { - int channels = req_comp ? req_comp : *comp; - stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi_uc)); - } - - return (unsigned char*)result; -} - -static stbi__uint16* stbi__load_and_postprocess_16bit(stbi__context* s, int* x, int* y, int* comp, int req_comp) -{ - stbi__result_info ri; - void* result = stbi__load_main(s, x, y, comp, req_comp, &ri, 16); - - if (result == NULL) - return NULL; - - // it is the responsibility of the loaders to make sure we get either 8 or 16 bit. - STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16); - - if (ri.bits_per_channel != 16) - { - result = stbi__convert_8_to_16((stbi_uc*)result, *x, *y, req_comp == 0 ? *comp : req_comp); - ri.bits_per_channel = 16; - } - - // @TODO: move stbi__convert_format16 to here - // @TODO: special case RGB-to-Y (and RGBA-to-YA) for 8-bit-to-16-bit case to keep more precision - - if (stbi__vertically_flip_on_load) - { - int channels = req_comp ? req_comp : *comp; - stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi__uint16)); - } - - return (stbi__uint16*)result; -} - - #if !defined(STBI_NO_HDR) && !defined(STBI_NO_LINEAR) -static void stbi__float_postprocess(float* result, int* x, int* y, int* comp, int req_comp) -{ - if (stbi__vertically_flip_on_load && result != NULL) - { - int channels = req_comp ? req_comp : *comp; - stbi__vertical_flip(result, *x, *y, channels * sizeof(float)); - } -} - #endif - - #ifndef STBI_NO_STDIO - - #if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) -STBI_EXTERN __declspec( - dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char* str, int cbmb, wchar_t* widestr, int cchwide); -STBI_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t* widestr, int cchwide, - char* str, int cbmb, const char* defchar, int* used_default); - #endif - - #if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) -STBIDEF int stbi_convert_wchar_to_utf8(char* buffer, size_t bufferlen, const wchar_t* input) -{ - return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int)bufferlen, NULL, NULL); -} - #endif - -static FILE* stbi__fopen(const char* filename, const char* mode) -{ - FILE* f; - #if defined(_WIN32) && defined(STBI_WINDOWS_UTF8) - wchar_t wMode[64]; - wchar_t wFilename[1024]; - if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename) / sizeof(*wFilename))) - return 0; - - if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode) / sizeof(*wMode))) - return 0; - - #if defined(_MSC_VER) && _MSC_VER >= 1400 - if (0 != _wfopen_s(&f, wFilename, wMode)) - f = 0; - #else - f = _wfopen(wFilename, wMode); - #endif - - #elif defined(_MSC_VER) && _MSC_VER >= 1400 - if (0 != fopen_s(&f, filename, mode)) - f = 0; - #else - f = fopen(filename, mode); - #endif - return f; -} - -STBIDEF stbi_uc* stbi_load(const char* filename, int* x, int* y, int* comp, int req_comp) -{ - FILE* f = stbi__fopen(filename, "rb"); - unsigned char* result; - if (!f) - return stbi__errpuc("can't fopen", "Unable to open file"); - result = stbi_load_from_file(f, x, y, comp, req_comp); - fclose(f); - return result; -} - -STBIDEF stbi_uc* stbi_load_from_file(FILE* f, int* x, int* y, int* comp, int req_comp) -{ - unsigned char* result; - stbi__context s; - stbi__start_file(&s, f); - result = stbi__load_and_postprocess_8bit(&s, x, y, comp, req_comp); - if (result) - { - // need to 'unget' all the characters in the IO buffer - fseek(f, -(int)(s.img_buffer_end - s.img_buffer), SEEK_CUR); - } - return result; -} - -STBIDEF stbi__uint16* stbi_load_from_file_16(FILE* f, int* x, int* y, int* comp, int req_comp) -{ - stbi__uint16* result; - stbi__context s; - stbi__start_file(&s, f); - result = stbi__load_and_postprocess_16bit(&s, x, y, comp, req_comp); - if (result) - { - // need to 'unget' all the characters in the IO buffer - fseek(f, -(int)(s.img_buffer_end - s.img_buffer), SEEK_CUR); - } - return result; -} - -STBIDEF stbi_us* stbi_load_16(const char* filename, int* x, int* y, int* comp, int req_comp) -{ - FILE* f = stbi__fopen(filename, "rb"); - stbi__uint16* result; - if (!f) - return (stbi_us*)stbi__errpuc("can't fopen", "Unable to open file"); - result = stbi_load_from_file_16(f, x, y, comp, req_comp); - fclose(f); - return result; -} - - #endif //! STBI_NO_STDIO - -STBIDEF stbi_us* stbi_load_16_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* channels_in_file, int desired_channels) -{ - stbi__context s; - stbi__start_mem(&s, buffer, len); - return stbi__load_and_postprocess_16bit(&s, x, y, channels_in_file, desired_channels); -} - -STBIDEF stbi_us* stbi_load_16_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* channels_in_file, int desired_channels) -{ - stbi__context s; - stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); - return stbi__load_and_postprocess_16bit(&s, x, y, channels_in_file, desired_channels); -} - -STBIDEF stbi_uc* stbi_load_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp, int req_comp) -{ - stbi__context s; - stbi__start_mem(&s, buffer, len); - return stbi__load_and_postprocess_8bit(&s, x, y, comp, req_comp); -} - -STBIDEF stbi_uc* stbi_load_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* comp, int req_comp) -{ - stbi__context s; - stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); - return stbi__load_and_postprocess_8bit(&s, x, y, comp, req_comp); -} - - #ifndef STBI_NO_GIF -STBIDEF stbi_uc* stbi_load_gif_from_memory(const stbi_uc* buffer, int len, int** delays, int* x, int* y, int* z, int* comp, int req_comp) -{ - unsigned char* result; - stbi__context s; - stbi__start_mem(&s, buffer, len); - - result = (unsigned char*)stbi__load_gif_main(&s, delays, x, y, z, comp, req_comp); - if (stbi__vertically_flip_on_load) - { - stbi__vertical_flip_slices(result, *x, *y, *z, *comp); - } - - return result; -} - #endif - - #ifndef STBI_NO_LINEAR -static float* stbi__loadf_main(stbi__context* s, int* x, int* y, int* comp, int req_comp) -{ - unsigned char* data; - #ifndef STBI_NO_HDR - if (stbi__hdr_test(s)) - { - stbi__result_info ri; - float* hdr_data = stbi__hdr_load(s, x, y, comp, req_comp, &ri); - if (hdr_data) - stbi__float_postprocess(hdr_data, x, y, comp, req_comp); - return hdr_data; - } - #endif - data = stbi__load_and_postprocess_8bit(s, x, y, comp, req_comp); - if (data) - return stbi__ldr_to_hdr(data, *x, *y, req_comp ? req_comp : *comp); - return stbi__errpf("unknown image type", "Image not of any known type, or corrupt"); -} - -STBIDEF float* stbi_loadf_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp, int req_comp) -{ - stbi__context s; - stbi__start_mem(&s, buffer, len); - return stbi__loadf_main(&s, x, y, comp, req_comp); -} - -STBIDEF float* stbi_loadf_from_callbacks(const stbi_io_callbacks* clbk, void* user, int* x, int* y, int* comp, int req_comp) -{ - stbi__context s; - stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); - return stbi__loadf_main(&s, x, y, comp, req_comp); -} - - #ifndef STBI_NO_STDIO -STBIDEF float* stbi_loadf(const char* filename, int* x, int* y, int* comp, int req_comp) -{ - float* result; - FILE* f = stbi__fopen(filename, "rb"); - if (!f) - return stbi__errpf("can't fopen", "Unable to open file"); - result = stbi_loadf_from_file(f, x, y, comp, req_comp); - fclose(f); - return result; -} - -STBIDEF float* stbi_loadf_from_file(FILE* f, int* x, int* y, int* comp, int req_comp) -{ - stbi__context s; - stbi__start_file(&s, f); - return stbi__loadf_main(&s, x, y, comp, req_comp); -} - #endif // !STBI_NO_STDIO - - #endif // !STBI_NO_LINEAR - -// these is-hdr-or-not is defined independent of whether STBI_NO_LINEAR is -// defined, for API simplicity; if STBI_NO_LINEAR is defined, it always -// reports false! - -STBIDEF int stbi_is_hdr_from_memory(const stbi_uc* buffer, int len) -{ - #ifndef STBI_NO_HDR - stbi__context s; - stbi__start_mem(&s, buffer, len); - return stbi__hdr_test(&s); - #else - STBI_NOTUSED(buffer); - STBI_NOTUSED(len); - return 0; - #endif -} - - #ifndef STBI_NO_STDIO -STBIDEF int stbi_is_hdr(const char* filename) -{ - FILE* f = stbi__fopen(filename, "rb"); - int result = 0; - if (f) - { - result = stbi_is_hdr_from_file(f); - fclose(f); - } - return result; -} - -STBIDEF int stbi_is_hdr_from_file(FILE* f) -{ - #ifndef STBI_NO_HDR - long pos = ftell(f); - int res; - stbi__context s; - stbi__start_file(&s, f); - res = stbi__hdr_test(&s); - fseek(f, pos, SEEK_SET); - return res; - #else - STBI_NOTUSED(f); - return 0; - #endif -} - #endif // !STBI_NO_STDIO - -STBIDEF int stbi_is_hdr_from_callbacks(const stbi_io_callbacks* clbk, void* user) -{ - #ifndef STBI_NO_HDR - stbi__context s; - stbi__start_callbacks(&s, (stbi_io_callbacks*)clbk, user); - return stbi__hdr_test(&s); - #else - STBI_NOTUSED(clbk); - STBI_NOTUSED(user); - return 0; - #endif -} - - #ifndef STBI_NO_LINEAR -static float stbi__l2h_gamma = 2.2f, stbi__l2h_scale = 1.0f; - -STBIDEF void stbi_ldr_to_hdr_gamma(float gamma) -{ - stbi__l2h_gamma = gamma; -} -STBIDEF void stbi_ldr_to_hdr_scale(float scale) -{ - stbi__l2h_scale = scale; -} - #endif - -static float stbi__h2l_gamma_i = 1.0f / 2.2f, stbi__h2l_scale_i = 1.0f; - -STBIDEF void stbi_hdr_to_ldr_gamma(float gamma) -{ - stbi__h2l_gamma_i = 1 / gamma; -} -STBIDEF void stbi_hdr_to_ldr_scale(float scale) -{ - stbi__h2l_scale_i = 1 / scale; -} - -////////////////////////////////////////////////////////////////////////////// -// -// Common code used by all image loaders -// - -enum { STBI__SCAN_load = 0, STBI__SCAN_type, STBI__SCAN_header }; - -static void stbi__refill_buffer(stbi__context* s) -{ - int n = (s->io.read)(s->io_user_data, (char*)s->buffer_start, s->buflen); - s->callback_already_read += (int)(s->img_buffer - s->img_buffer_original); - if (n == 0) - { - // at end of file, treat same as if from memory, but need to handle case - // where s->img_buffer isn't pointing to safe memory, e.g. 0-byte file - s->read_from_callbacks = 0; - s->img_buffer = s->buffer_start; - s->img_buffer_end = s->buffer_start + 1; - *s->img_buffer = 0; - } - else - { - s->img_buffer = s->buffer_start; - s->img_buffer_end = s->buffer_start + n; - } -} - -stbi_inline static stbi_uc stbi__get8(stbi__context* s) -{ - if (s->img_buffer < s->img_buffer_end) - return *s->img_buffer++; - if (s->read_from_callbacks) - { - stbi__refill_buffer(s); - return *s->img_buffer++; - } - return 0; -} - - #if defined(STBI_NO_JPEG) && defined(STBI_NO_HDR) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM) - // nothing - #else -stbi_inline static int stbi__at_eof(stbi__context* s) -{ - if (s->io.read) - { - if (!(s->io.eof)(s->io_user_data)) - return 0; - // if feof() is true, check if buffer = end - // special case: we've only got the special 0 character at the end - if (s->read_from_callbacks == 0) - return 1; - } - - return s->img_buffer >= s->img_buffer_end; -} - #endif - - #if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && \ - defined(STBI_NO_GIF) && defined(STBI_NO_PIC) - // nothing - #else -static void stbi__skip(stbi__context* s, int n) -{ - if (n == 0) - return; // already there! - if (n < 0) - { - s->img_buffer = s->img_buffer_end; - return; - } - if (s->io.read) - { - int blen = (int)(s->img_buffer_end - s->img_buffer); - if (blen < n) - { - s->img_buffer = s->img_buffer_end; - (s->io.skip)(s->io_user_data, n - blen); - return; - } - } - s->img_buffer += n; -} - #endif - - #if defined(STBI_NO_PNG) && defined(STBI_NO_TGA) && defined(STBI_NO_HDR) && defined(STBI_NO_PNM) - // nothing - #else -static int stbi__getn(stbi__context* s, stbi_uc* buffer, int n) -{ - if (s->io.read) - { - int blen = (int)(s->img_buffer_end - s->img_buffer); - if (blen < n) - { - int res, count; - - memcpy(buffer, s->img_buffer, blen); - - count = (s->io.read)(s->io_user_data, (char*)buffer + blen, n - blen); - res = (count == (n - blen)); - s->img_buffer = s->img_buffer_end; - return res; - } - } - - if (s->img_buffer + n <= s->img_buffer_end) - { - memcpy(buffer, s->img_buffer, n); - s->img_buffer += n; - return 1; - } - else - return 0; -} - #endif - - #if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC) - // nothing - #else -static int stbi__get16be(stbi__context* s) -{ - int z = stbi__get8(s); - return (z << 8) + stbi__get8(s); -} - #endif - - #if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC) - // nothing - #else -static stbi__uint32 stbi__get32be(stbi__context* s) -{ - stbi__uint32 z = stbi__get16be(s); - return (z << 16) + stbi__get16be(s); -} - #endif - - #if defined(STBI_NO_BMP) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) - // nothing - #else -static int stbi__get16le(stbi__context* s) -{ - int z = stbi__get8(s); - return z + (stbi__get8(s) << 8); -} - #endif - - #ifndef STBI_NO_BMP -static stbi__uint32 stbi__get32le(stbi__context* s) -{ - stbi__uint32 z = stbi__get16le(s); - z += (stbi__uint32)stbi__get16le(s) << 16; - return z; -} - #endif - - #define STBI__BYTECAST(x) ((stbi_uc)((x) & 255)) // truncate int to byte without warnings - - #if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && \ - defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM) - // nothing - #else -////////////////////////////////////////////////////////////////////////////// -// -// generic converter from built-in img_n to req_comp -// individual types do this automatically as much as possible (e.g. jpeg -// does all cases internally since it needs to colorspace convert anyway, -// and it never has alpha, so very few cases ). png can automatically -// interleave an alpha=255 channel, but falls back to this for other cases -// -// assume data buffer is malloced, so malloc a new one and free that one -// only failure mode is malloc failing - -static stbi_uc stbi__compute_y(int r, int g, int b) -{ - return (stbi_uc)(((r * 77) + (g * 150) + (29 * b)) >> 8); -} - #endif - - #if defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && \ - defined(STBI_NO_PIC) && defined(STBI_NO_PNM) - // nothing - #else -static unsigned char* stbi__convert_format(unsigned char* data, int img_n, int req_comp, unsigned int x, unsigned int y) -{ - int i, j; - unsigned char* good; - - if (req_comp == img_n) - return data; - STBI_ASSERT(req_comp >= 1 && req_comp <= 4); - - good = (unsigned char*)stbi__malloc_mad3(req_comp, x, y, 0); - if (good == NULL) - { - STBI_FREE(data); - return stbi__errpuc("outofmem", "Out of memory"); - } - - for (j = 0; j < (int)y; ++j) - { - unsigned char* src = data + j * x * img_n; - unsigned char* dest = good + j * x * req_comp; - - #define STBI__COMBO(a, b) ((a) * 8 + (b)) - #define STBI__CASE(a, b) \ - case STBI__COMBO(a, b): \ - for (i = x - 1; i >= 0; --i, src += a, dest += b) - // convert source image with img_n components to one with req_comp components; - // avoid switch per pixel, so use switch per scanline and massive macros - switch (STBI__COMBO(img_n, req_comp)) - { - STBI__CASE(1, 2) - { - dest[0] = src[0]; - dest[1] = 255; - } - break; - STBI__CASE(1, 3) - { - dest[0] = dest[1] = dest[2] = src[0]; - } - break; - STBI__CASE(1, 4) - { - dest[0] = dest[1] = dest[2] = src[0]; - dest[3] = 255; - } - break; - STBI__CASE(2, 1) - { - dest[0] = src[0]; - } - break; - STBI__CASE(2, 3) - { - dest[0] = dest[1] = dest[2] = src[0]; - } - break; - STBI__CASE(2, 4) - { - dest[0] = dest[1] = dest[2] = src[0]; - dest[3] = src[1]; - } - break; - STBI__CASE(3, 4) - { - dest[0] = src[0]; - dest[1] = src[1]; - dest[2] = src[2]; - dest[3] = 255; - } - break; - STBI__CASE(3, 1) - { - dest[0] = stbi__compute_y(src[0], src[1], src[2]); - } - break; - STBI__CASE(3, 2) - { - dest[0] = stbi__compute_y(src[0], src[1], src[2]); - dest[1] = 255; - } - break; - STBI__CASE(4, 1) - { - dest[0] = stbi__compute_y(src[0], src[1], src[2]); - } - break; - STBI__CASE(4, 2) - { - dest[0] = stbi__compute_y(src[0], src[1], src[2]); - dest[1] = src[3]; - } - break; - STBI__CASE(4, 3) - { - dest[0] = src[0]; - dest[1] = src[1]; - dest[2] = src[2]; - } - break; - default: - STBI_ASSERT(0); - STBI_FREE(data); - STBI_FREE(good); - return stbi__errpuc("unsupported", "Unsupported format conversion"); - } - #undef STBI__CASE - } - - STBI_FREE(data); - return good; -} - #endif - - #if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) - // nothing - #else -static stbi__uint16 stbi__compute_y_16(int r, int g, int b) -{ - return (stbi__uint16)(((r * 77) + (g * 150) + (29 * b)) >> 8); -} - #endif - - #if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) - // nothing - #else -static stbi__uint16* stbi__convert_format16(stbi__uint16* data, int img_n, int req_comp, unsigned int x, unsigned int y) -{ - int i, j; - stbi__uint16* good; - - if (req_comp == img_n) - return data; - STBI_ASSERT(req_comp >= 1 && req_comp <= 4); - - good = (stbi__uint16*)stbi__malloc(req_comp * x * y * 2); - if (good == NULL) - { - STBI_FREE(data); - return (stbi__uint16*)stbi__errpuc("outofmem", "Out of memory"); - } - - for (j = 0; j < (int)y; ++j) - { - stbi__uint16* src = data + j * x * img_n; - stbi__uint16* dest = good + j * x * req_comp; - - #define STBI__COMBO(a, b) ((a) * 8 + (b)) - #define STBI__CASE(a, b) \ - case STBI__COMBO(a, b): \ - for (i = x - 1; i >= 0; --i, src += a, dest += b) - // convert source image with img_n components to one with req_comp components; - // avoid switch per pixel, so use switch per scanline and massive macros - switch (STBI__COMBO(img_n, req_comp)) - { - STBI__CASE(1, 2) - { - dest[0] = src[0]; - dest[1] = 0xffff; - } - break; - STBI__CASE(1, 3) - { - dest[0] = dest[1] = dest[2] = src[0]; - } - break; - STBI__CASE(1, 4) - { - dest[0] = dest[1] = dest[2] = src[0]; - dest[3] = 0xffff; - } - break; - STBI__CASE(2, 1) - { - dest[0] = src[0]; - } - break; - STBI__CASE(2, 3) - { - dest[0] = dest[1] = dest[2] = src[0]; - } - break; - STBI__CASE(2, 4) - { - dest[0] = dest[1] = dest[2] = src[0]; - dest[3] = src[1]; - } - break; - STBI__CASE(3, 4) - { - dest[0] = src[0]; - dest[1] = src[1]; - dest[2] = src[2]; - dest[3] = 0xffff; - } - break; - STBI__CASE(3, 1) - { - dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); - } - break; - STBI__CASE(3, 2) - { - dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); - dest[1] = 0xffff; - } - break; - STBI__CASE(4, 1) - { - dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); - } - break; - STBI__CASE(4, 2) - { - dest[0] = stbi__compute_y_16(src[0], src[1], src[2]); - dest[1] = src[3]; - } - break; - STBI__CASE(4, 3) - { - dest[0] = src[0]; - dest[1] = src[1]; - dest[2] = src[2]; - } - break; - default: - STBI_ASSERT(0); - STBI_FREE(data); - STBI_FREE(good); - return (stbi__uint16*)stbi__errpuc("unsupported", "Unsupported format conversion"); - } - #undef STBI__CASE - } - - STBI_FREE(data); - return good; -} - #endif - - #ifndef STBI_NO_LINEAR -static float* stbi__ldr_to_hdr(stbi_uc* data, int x, int y, int comp) -{ - int i, k, n; - float* output; - if (!data) - return NULL; - output = (float*)stbi__malloc_mad4(x, y, comp, sizeof(float), 0); - if (output == NULL) - { - STBI_FREE(data); - return stbi__errpf("outofmem", "Out of memory"); - } - // compute number of non-alpha components - if (comp & 1) - n = comp; - else - n = comp - 1; - for (i = 0; i < x * y; ++i) - { - for (k = 0; k < n; ++k) - { - output[i * comp + k] = (float)(pow(data[i * comp + k] / 255.0f, stbi__l2h_gamma) * stbi__l2h_scale); - } - } - if (n < comp) - { - for (i = 0; i < x * y; ++i) - { - output[i * comp + n] = data[i * comp + n] / 255.0f; - } - } - STBI_FREE(data); - return output; -} - #endif - - #ifndef STBI_NO_HDR - #define stbi__float2int(x) ((int)(x)) -static stbi_uc* stbi__hdr_to_ldr(float* data, int x, int y, int comp) -{ - int i, k, n; - stbi_uc* output; - if (!data) - return NULL; - output = (stbi_uc*)stbi__malloc_mad3(x, y, comp, 0); - if (output == NULL) - { - STBI_FREE(data); - return stbi__errpuc("outofmem", "Out of memory"); - } - // compute number of non-alpha components - if (comp & 1) - n = comp; - else - n = comp - 1; - for (i = 0; i < x * y; ++i) - { - for (k = 0; k < n; ++k) - { - float z = (float)pow(data[i * comp + k] * stbi__h2l_scale_i, stbi__h2l_gamma_i) * 255 + 0.5f; - if (z < 0) - z = 0; - if (z > 255) - z = 255; - output[i * comp + k] = (stbi_uc)stbi__float2int(z); - } - if (k < comp) - { - float z = data[i * comp + k] * 255 + 0.5f; - if (z < 0) - z = 0; - if (z > 255) - z = 255; - output[i * comp + k] = (stbi_uc)stbi__float2int(z); - } - } - STBI_FREE(data); - return output; -} - #endif - -////////////////////////////////////////////////////////////////////////////// -// -// "baseline" JPEG/JFIF decoder -// -// simple implementation -// - doesn't support delayed output of y-dimension -// - simple interface (only one output format: 8-bit interleaved RGB) -// - doesn't try to recover corrupt jpegs -// - doesn't allow partial loading, loading multiple at once -// - still fast on x86 (copying globals into locals doesn't help x86) -// - allocates lots of intermediate memory (full size of all components) -// - non-interleaved case requires this anyway -// - allows good upsampling (see next) -// high-quality -// - upsampled channels are bilinearly interpolated, even across blocks -// - quality integer IDCT derived from IJG's 'slow' -// performance -// - fast huffman; reasonable integer IDCT -// - some SIMD kernels for common paths on targets with SSE2/NEON -// - uses a lot of intermediate memory, could cache poorly - - #ifndef STBI_NO_JPEG - - // huffman decoding acceleration - #define FAST_BITS 9 // larger handles more cases; smaller stomps less cache - -typedef struct { - stbi_uc fast[1 << FAST_BITS]; - // weirdly, repacking this into AoS is a 10% speed loss, instead of a win - stbi__uint16 code[256]; - stbi_uc values[256]; - stbi_uc size[257]; - unsigned int maxcode[18]; - int delta[17]; // old 'firstsymbol' - old 'firstcode' -} stbi__huffman; - -typedef struct { - stbi__context* s; - stbi__huffman huff_dc[4]; - stbi__huffman huff_ac[4]; - stbi__uint16 dequant[4][64]; - stbi__int16 fast_ac[4][1 << FAST_BITS]; - - // sizes for components, interleaved MCUs - int img_h_max, img_v_max; - int img_mcu_x, img_mcu_y; - int img_mcu_w, img_mcu_h; - - // definition of jpeg image component - struct { - int id; - int h, v; - int tq; - int hd, ha; - int dc_pred; - - int x, y, w2, h2; - stbi_uc* data; - void *raw_data, *raw_coeff; - stbi_uc* linebuf; - short* coeff; // progressive only - int coeff_w, coeff_h; // number of 8x8 coefficient blocks - } img_comp[4]; - - stbi__uint32 code_buffer; // jpeg entropy-coded buffer - int code_bits; // number of valid bits - unsigned char marker; // marker seen while filling entropy buffer - int nomore; // flag if we saw a marker so must stop - - int progressive; - int spec_start; - int spec_end; - int succ_high; - int succ_low; - int eob_run; - int jfif; - int app14_color_transform; // Adobe APP14 tag - int rgb; - - int scan_n, order[4]; - int restart_interval, todo; - - // kernels - void (*idct_block_kernel)(stbi_uc* out, int out_stride, short data[64]); - void (*YCbCr_to_RGB_kernel)(stbi_uc* out, const stbi_uc* y, const stbi_uc* pcb, const stbi_uc* pcr, int count, int step); - stbi_uc* (*resample_row_hv_2_kernel)(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs); -} stbi__jpeg; - -static int stbi__build_huffman(stbi__huffman* h, int* count) -{ - int i, j, k = 0; - unsigned int code; - // build size list for each symbol (from JPEG spec) - for (i = 0; i < 16; ++i) - { - for (j = 0; j < count[i]; ++j) - { - h->size[k++] = (stbi_uc)(i + 1); - if (k >= 257) - return stbi__err("bad size list", "Corrupt JPEG"); - } - } - h->size[k] = 0; - - // compute actual symbols (from jpeg spec) - code = 0; - k = 0; - for (j = 1; j <= 16; ++j) - { - // compute delta to add to code to compute symbol id - h->delta[j] = k - code; - if (h->size[k] == j) - { - while (h->size[k] == j) - h->code[k++] = (stbi__uint16)(code++); - if (code - 1 >= (1u << j)) - return stbi__err("bad code lengths", "Corrupt JPEG"); - } - // compute largest code + 1 for this size, preshifted as needed later - h->maxcode[j] = code << (16 - j); - code <<= 1; - } - h->maxcode[j] = 0xffffffff; - - // build non-spec acceleration table; 255 is flag for not-accelerated - memset(h->fast, 255, 1 << FAST_BITS); - for (i = 0; i < k; ++i) - { - int s = h->size[i]; - if (s <= FAST_BITS) - { - int c = h->code[i] << (FAST_BITS - s); - int m = 1 << (FAST_BITS - s); - for (j = 0; j < m; ++j) - { - h->fast[c + j] = (stbi_uc)i; - } - } - } - return 1; -} - -// build a table that decodes both magnitude and value of small ACs in -// one go. -static void stbi__build_fast_ac(stbi__int16* fast_ac, stbi__huffman* h) -{ - int i; - for (i = 0; i < (1 << FAST_BITS); ++i) - { - stbi_uc fast = h->fast[i]; - fast_ac[i] = 0; - if (fast < 255) - { - int rs = h->values[fast]; - int run = (rs >> 4) & 15; - int magbits = rs & 15; - int len = h->size[fast]; - - if (magbits && len + magbits <= FAST_BITS) - { - // magnitude code followed by receive_extend code - int k = ((i << len) & ((1 << FAST_BITS) - 1)) >> (FAST_BITS - magbits); - int m = 1 << (magbits - 1); - if (k < m) - k += (~0U << magbits) + 1; - // if the result is small enough, we can fit it in fast_ac table - if (k >= -128 && k <= 127) - fast_ac[i] = (stbi__int16)((k * 256) + (run * 16) + (len + magbits)); - } - } - } -} - -static void stbi__grow_buffer_unsafe(stbi__jpeg* j) -{ - do - { - unsigned int b = j->nomore ? 0 : stbi__get8(j->s); - if (b == 0xff) - { - int c = stbi__get8(j->s); - while (c == 0xff) - c = stbi__get8(j->s); // consume fill bytes - if (c != 0) - { - j->marker = (unsigned char)c; - j->nomore = 1; - return; - } - } - j->code_buffer |= b << (24 - j->code_bits); - j->code_bits += 8; - } while (j->code_bits <= 24); -} - -// (1 << n) - 1 -static const stbi__uint32 stbi__bmask[17] = {0, 1, 3, 7, 15, 31, 63, 127, 255, 511, 1023, 2047, 4095, 8191, 16383, 32767, 65535}; - -// decode a jpeg huffman value from the bitstream -stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg* j, stbi__huffman* h) -{ - unsigned int temp; - int c, k; - - if (j->code_bits < 16) - stbi__grow_buffer_unsafe(j); - - // look at the top FAST_BITS and determine what symbol ID it is, - // if the code is <= FAST_BITS - c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS) - 1); - k = h->fast[c]; - if (k < 255) - { - int s = h->size[k]; - if (s > j->code_bits) - return -1; - j->code_buffer <<= s; - j->code_bits -= s; - return h->values[k]; - } - - // naive test is to shift the code_buffer down so k bits are - // valid, then test against maxcode. To speed this up, we've - // preshifted maxcode left so that it has (16-k) 0s at the - // end; in other words, regardless of the number of bits, it - // wants to be compared against something shifted to have 16; - // that way we don't need to shift inside the loop. - temp = j->code_buffer >> 16; - for (k = FAST_BITS + 1;; ++k) - if (temp < h->maxcode[k]) - break; - if (k == 17) - { - // error! code not found - j->code_bits -= 16; - return -1; - } - - if (k > j->code_bits) - return -1; - - // convert the huffman code to the symbol id - c = ((j->code_buffer >> (32 - k)) & stbi__bmask[k]) + h->delta[k]; - if (c < 0 || c >= 256) // symbol id out of bounds! - return -1; - STBI_ASSERT((((j->code_buffer) >> (32 - h->size[c])) & stbi__bmask[h->size[c]]) == h->code[c]); - - // convert the id to a symbol - j->code_bits -= k; - j->code_buffer <<= k; - return h->values[c]; -} - -// bias[n] = (-1<code_bits < n) - stbi__grow_buffer_unsafe(j); - if (j->code_bits < n) - return 0; // ran out of bits from stream, return 0s intead of continuing - - sgn = j->code_buffer >> 31; // sign bit always in MSB; 0 if MSB clear (positive), 1 if MSB set (negative) - k = stbi_lrot(j->code_buffer, n); - j->code_buffer = k & ~stbi__bmask[n]; - k &= stbi__bmask[n]; - j->code_bits -= n; - return k + (stbi__jbias[n] & (sgn - 1)); -} - -// get some unsigned bits -stbi_inline static int stbi__jpeg_get_bits(stbi__jpeg* j, int n) -{ - unsigned int k; - if (j->code_bits < n) - stbi__grow_buffer_unsafe(j); - if (j->code_bits < n) - return 0; // ran out of bits from stream, return 0s intead of continuing - k = stbi_lrot(j->code_buffer, n); - j->code_buffer = k & ~stbi__bmask[n]; - k &= stbi__bmask[n]; - j->code_bits -= n; - return k; -} - -stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg* j) -{ - unsigned int k; - if (j->code_bits < 1) - stbi__grow_buffer_unsafe(j); - if (j->code_bits < 1) - return 0; // ran out of bits from stream, return 0s intead of continuing - k = j->code_buffer; - j->code_buffer <<= 1; - --j->code_bits; - return k & 0x80000000; -} - -// given a value that's at position X in the zigzag stream, -// where does it appear in the 8x8 matrix coded as row-major? -static const stbi_uc stbi__jpeg_dezigzag[64 + 15] = {0, 1, 8, 16, 9, 2, 3, 10, 17, 24, 32, 25, 18, 11, 4, 5, 12, 19, 26, 33, 40, 48, 41, 34, 27, 20, - 13, 6, 7, 14, 21, 28, 35, 42, 49, 56, 57, 50, 43, 36, 29, 22, 15, 23, 30, 37, 44, 51, 58, 59, 52, 45, 38, 31, 39, 46, 53, 60, 61, 54, 47, 55, 62, - 63, - // let corrupt input sample past end - 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63}; - -// decode one 64-entry block-- -static int stbi__jpeg_decode_block(stbi__jpeg* j, short data[64], stbi__huffman* hdc, stbi__huffman* hac, stbi__int16* fac, int b, - stbi__uint16* dequant) -{ - int diff, dc, k; - int t; - - if (j->code_bits < 16) - stbi__grow_buffer_unsafe(j); - t = stbi__jpeg_huff_decode(j, hdc); - if (t < 0 || t > 15) - return stbi__err("bad huffman code", "Corrupt JPEG"); - - // 0 all the ac values now so we can do it 32-bits at a time - memset(data, 0, 64 * sizeof(data[0])); - - diff = t ? stbi__extend_receive(j, t) : 0; - if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) - return stbi__err("bad delta", "Corrupt JPEG"); - dc = j->img_comp[b].dc_pred + diff; - j->img_comp[b].dc_pred = dc; - if (!stbi__mul2shorts_valid(dc, dequant[0])) - return stbi__err("can't merge dc and ac", "Corrupt JPEG"); - data[0] = (short)(dc * dequant[0]); - - // decode AC components, see JPEG spec - k = 1; - do - { - unsigned int zig; - int c, r, s; - if (j->code_bits < 16) - stbi__grow_buffer_unsafe(j); - c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS) - 1); - r = fac[c]; - if (r) - { // fast-AC path - k += (r >> 4) & 15; // run - s = r & 15; // combined length - if (s > j->code_bits) - return stbi__err("bad huffman code", "Combined length longer than code bits available"); - j->code_buffer <<= s; - j->code_bits -= s; - // decode into unzigzag'd location - zig = stbi__jpeg_dezigzag[k++]; - data[zig] = (short)((r >> 8) * dequant[zig]); - } - else - { - int rs = stbi__jpeg_huff_decode(j, hac); - if (rs < 0) - return stbi__err("bad huffman code", "Corrupt JPEG"); - s = rs & 15; - r = rs >> 4; - if (s == 0) - { - if (rs != 0xf0) - break; // end block - k += 16; - } - else - { - k += r; - // decode into unzigzag'd location - zig = stbi__jpeg_dezigzag[k++]; - data[zig] = (short)(stbi__extend_receive(j, s) * dequant[zig]); - } - } - } while (k < 64); - return 1; -} - -static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg* j, short data[64], stbi__huffman* hdc, int b) -{ - int diff, dc; - int t; - if (j->spec_end != 0) - return stbi__err("can't merge dc and ac", "Corrupt JPEG"); - - if (j->code_bits < 16) - stbi__grow_buffer_unsafe(j); - - if (j->succ_high == 0) - { - // first scan for DC coefficient, must be first - memset(data, 0, 64 * sizeof(data[0])); // 0 all the ac values now - t = stbi__jpeg_huff_decode(j, hdc); - if (t < 0 || t > 15) - return stbi__err("can't merge dc and ac", "Corrupt JPEG"); - diff = t ? stbi__extend_receive(j, t) : 0; - - if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) - return stbi__err("bad delta", "Corrupt JPEG"); - dc = j->img_comp[b].dc_pred + diff; - j->img_comp[b].dc_pred = dc; - if (!stbi__mul2shorts_valid(dc, 1 << j->succ_low)) - return stbi__err("can't merge dc and ac", "Corrupt JPEG"); - data[0] = (short)(dc * (1 << j->succ_low)); - } - else - { - // refinement scan for DC coefficient - if (stbi__jpeg_get_bit(j)) - data[0] += (short)(1 << j->succ_low); - } - return 1; -} - -// @OPTIMIZE: store non-zigzagged during the decode passes, -// and only de-zigzag when dequantizing -static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg* j, short data[64], stbi__huffman* hac, stbi__int16* fac) -{ - int k; - if (j->spec_start == 0) - return stbi__err("can't merge dc and ac", "Corrupt JPEG"); - - if (j->succ_high == 0) - { - int shift = j->succ_low; - - if (j->eob_run) - { - --j->eob_run; - return 1; - } - - k = j->spec_start; - do - { - unsigned int zig; - int c, r, s; - if (j->code_bits < 16) - stbi__grow_buffer_unsafe(j); - c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS) - 1); - r = fac[c]; - if (r) - { // fast-AC path - k += (r >> 4) & 15; // run - s = r & 15; // combined length - if (s > j->code_bits) - return stbi__err("bad huffman code", "Combined length longer than code bits available"); - j->code_buffer <<= s; - j->code_bits -= s; - zig = stbi__jpeg_dezigzag[k++]; - data[zig] = (short)((r >> 8) * (1 << shift)); - } - else - { - int rs = stbi__jpeg_huff_decode(j, hac); - if (rs < 0) - return stbi__err("bad huffman code", "Corrupt JPEG"); - s = rs & 15; - r = rs >> 4; - if (s == 0) - { - if (r < 15) - { - j->eob_run = (1 << r); - if (r) - j->eob_run += stbi__jpeg_get_bits(j, r); - --j->eob_run; - break; - } - k += 16; - } - else - { - k += r; - zig = stbi__jpeg_dezigzag[k++]; - data[zig] = (short)(stbi__extend_receive(j, s) * (1 << shift)); - } - } - } while (k <= j->spec_end); - } - else - { - // refinement scan for these AC coefficients - - short bit = (short)(1 << j->succ_low); - - if (j->eob_run) - { - --j->eob_run; - for (k = j->spec_start; k <= j->spec_end; ++k) - { - short* p = &data[stbi__jpeg_dezigzag[k]]; - if (*p != 0) - if (stbi__jpeg_get_bit(j)) - if ((*p & bit) == 0) - { - if (*p > 0) - *p += bit; - else - *p -= bit; - } - } - } - else - { - k = j->spec_start; - do - { - int r, s; - int rs = stbi__jpeg_huff_decode(j, hac); // @OPTIMIZE see if we can use the fast path here, advance-by-r is so slow, eh - if (rs < 0) - return stbi__err("bad huffman code", "Corrupt JPEG"); - s = rs & 15; - r = rs >> 4; - if (s == 0) - { - if (r < 15) - { - j->eob_run = (1 << r) - 1; - if (r) - j->eob_run += stbi__jpeg_get_bits(j, r); - r = 64; // force end of block - } - else - { - // r=15 s=0 should write 16 0s, so we just do - // a run of 15 0s and then write s (which is 0), - // so we don't have to do anything special here - } - } - else - { - if (s != 1) - return stbi__err("bad huffman code", "Corrupt JPEG"); - // sign bit - if (stbi__jpeg_get_bit(j)) - s = bit; - else - s = -bit; - } - - // advance by r - while (k <= j->spec_end) - { - short* p = &data[stbi__jpeg_dezigzag[k++]]; - if (*p != 0) - { - if (stbi__jpeg_get_bit(j)) - if ((*p & bit) == 0) - { - if (*p > 0) - *p += bit; - else - *p -= bit; - } - } - else - { - if (r == 0) - { - *p = (short)s; - break; - } - --r; - } - } - } while (k <= j->spec_end); - } - } - return 1; -} - -// take a -128..127 value and stbi__clamp it and convert to 0..255 -stbi_inline static stbi_uc stbi__clamp(int x) -{ - // trick to use a single test to catch both cases - if ((unsigned int)x > 255) - { - if (x < 0) - return 0; - if (x > 255) - return 255; - } - return (stbi_uc)x; -} - - #define stbi__f2f(x) ((int)(((x) * 4096 + 0.5))) - #define stbi__fsh(x) ((x) * 4096) - - // derived from jidctint -- DCT_ISLOW - #define STBI__IDCT_1D(s0, s1, s2, s3, s4, s5, s6, s7) \ - int t0, t1, t2, t3, p1, p2, p3, p4, p5, x0, x1, x2, x3; \ - p2 = s2; \ - p3 = s6; \ - p1 = (p2 + p3) * stbi__f2f(0.5411961f); \ - t2 = p1 + p3 * stbi__f2f(-1.847759065f); \ - t3 = p1 + p2 * stbi__f2f(0.765366865f); \ - p2 = s0; \ - p3 = s4; \ - t0 = stbi__fsh(p2 + p3); \ - t1 = stbi__fsh(p2 - p3); \ - x0 = t0 + t3; \ - x3 = t0 - t3; \ - x1 = t1 + t2; \ - x2 = t1 - t2; \ - t0 = s7; \ - t1 = s5; \ - t2 = s3; \ - t3 = s1; \ - p3 = t0 + t2; \ - p4 = t1 + t3; \ - p1 = t0 + t3; \ - p2 = t1 + t2; \ - p5 = (p3 + p4) * stbi__f2f(1.175875602f); \ - t0 = t0 * stbi__f2f(0.298631336f); \ - t1 = t1 * stbi__f2f(2.053119869f); \ - t2 = t2 * stbi__f2f(3.072711026f); \ - t3 = t3 * stbi__f2f(1.501321110f); \ - p1 = p5 + p1 * stbi__f2f(-0.899976223f); \ - p2 = p5 + p2 * stbi__f2f(-2.562915447f); \ - p3 = p3 * stbi__f2f(-1.961570560f); \ - p4 = p4 * stbi__f2f(-0.390180644f); \ - t3 += p1 + p4; \ - t2 += p2 + p3; \ - t1 += p2 + p4; \ - t0 += p1 + p3; - -static void stbi__idct_block(stbi_uc* out, int out_stride, short data[64]) -{ - int i, val[64], *v = val; - stbi_uc* o; - short* d = data; - - // columns - for (i = 0; i < 8; ++i, ++d, ++v) - { - // if all zeroes, shortcut -- this avoids dequantizing 0s and IDCTing - if (d[8] == 0 && d[16] == 0 && d[24] == 0 && d[32] == 0 && d[40] == 0 && d[48] == 0 && d[56] == 0) - { - // no shortcut 0 seconds - // (1|2|3|4|5|6|7)==0 0 seconds - // all separate -0.047 seconds - // 1 && 2|3 && 4|5 && 6|7: -0.047 seconds - int dcterm = d[0] * 4; - v[0] = v[8] = v[16] = v[24] = v[32] = v[40] = v[48] = v[56] = dcterm; - } - else - { - STBI__IDCT_1D(d[0], d[8], d[16], d[24], d[32], d[40], d[48], d[56]) - // constants scaled things up by 1<<12; let's bring them back - // down, but keep 2 extra bits of precision - x0 += 512; - x1 += 512; - x2 += 512; - x3 += 512; - v[0] = (x0 + t3) >> 10; - v[56] = (x0 - t3) >> 10; - v[8] = (x1 + t2) >> 10; - v[48] = (x1 - t2) >> 10; - v[16] = (x2 + t1) >> 10; - v[40] = (x2 - t1) >> 10; - v[24] = (x3 + t0) >> 10; - v[32] = (x3 - t0) >> 10; - } - } - - for (i = 0, v = val, o = out; i < 8; ++i, v += 8, o += out_stride) - { - // no fast case since the first 1D IDCT spread components out - STBI__IDCT_1D(v[0], v[1], v[2], v[3], v[4], v[5], v[6], v[7]) - // constants scaled things up by 1<<12, plus we had 1<<2 from first - // loop, plus horizontal and vertical each scale by sqrt(8) so together - // we've got an extra 1<<3, so 1<<17 total we need to remove. - // so we want to round that, which means adding 0.5 * 1<<17, - // aka 65536. Also, we'll end up with -128 to 127 that we want - // to encode as 0..255 by adding 128, so we'll add that before the shift - x0 += 65536 + (128 << 17); - x1 += 65536 + (128 << 17); - x2 += 65536 + (128 << 17); - x3 += 65536 + (128 << 17); - // tried computing the shifts into temps, or'ing the temps to see - // if any were out of range, but that was slower - o[0] = stbi__clamp((x0 + t3) >> 17); - o[7] = stbi__clamp((x0 - t3) >> 17); - o[1] = stbi__clamp((x1 + t2) >> 17); - o[6] = stbi__clamp((x1 - t2) >> 17); - o[2] = stbi__clamp((x2 + t1) >> 17); - o[5] = stbi__clamp((x2 - t1) >> 17); - o[3] = stbi__clamp((x3 + t0) >> 17); - o[4] = stbi__clamp((x3 - t0) >> 17); - } -} - - #ifdef STBI_SSE2 -// sse2 integer IDCT. not the fastest possible implementation but it -// produces bit-identical results to the generic C version so it's -// fully "transparent". -static void stbi__idct_simd(stbi_uc* out, int out_stride, short data[64]) -{ - // This is constructed to match our regular (generic) integer IDCT exactly. - __m128i row0, row1, row2, row3, row4, row5, row6, row7; - __m128i tmp; - - // dot product constant: even elems=x, odd elems=y - #define dct_const(x, y) _mm_setr_epi16((x), (y), (x), (y), (x), (y), (x), (y)) - - // out(0) = c0[even]*x + c0[odd]*y (c0, x, y 16-bit, out 32-bit) - // out(1) = c1[even]*x + c1[odd]*y - #define dct_rot(out0, out1, x, y, c0, c1) \ - __m128i c0##lo = _mm_unpacklo_epi16((x), (y)); \ - __m128i c0##hi = _mm_unpackhi_epi16((x), (y)); \ - __m128i out0##_l = _mm_madd_epi16(c0##lo, c0); \ - __m128i out0##_h = _mm_madd_epi16(c0##hi, c0); \ - __m128i out1##_l = _mm_madd_epi16(c0##lo, c1); \ - __m128i out1##_h = _mm_madd_epi16(c0##hi, c1) - - // out = in << 12 (in 16-bit, out 32-bit) - #define dct_widen(out, in) \ - __m128i out##_l = _mm_srai_epi32(_mm_unpacklo_epi16(_mm_setzero_si128(), (in)), 4); \ - __m128i out##_h = _mm_srai_epi32(_mm_unpackhi_epi16(_mm_setzero_si128(), (in)), 4) - - // wide add - #define dct_wadd(out, a, b) \ - __m128i out##_l = _mm_add_epi32(a##_l, b##_l); \ - __m128i out##_h = _mm_add_epi32(a##_h, b##_h) - - // wide sub - #define dct_wsub(out, a, b) \ - __m128i out##_l = _mm_sub_epi32(a##_l, b##_l); \ - __m128i out##_h = _mm_sub_epi32(a##_h, b##_h) - - // butterfly a/b, add bias, then shift by "s" and pack - #define dct_bfly32o(out0, out1, a, b, bias, s) \ - { \ - __m128i abiased_l = _mm_add_epi32(a##_l, bias); \ - __m128i abiased_h = _mm_add_epi32(a##_h, bias); \ - dct_wadd(sum, abiased, b); \ - dct_wsub(dif, abiased, b); \ - out0 = _mm_packs_epi32(_mm_srai_epi32(sum_l, s), _mm_srai_epi32(sum_h, s)); \ - out1 = _mm_packs_epi32(_mm_srai_epi32(dif_l, s), _mm_srai_epi32(dif_h, s)); \ - } - - // 8-bit interleave step (for transposes) - #define dct_interleave8(a, b) \ - tmp = a; \ - a = _mm_unpacklo_epi8(a, b); \ - b = _mm_unpackhi_epi8(tmp, b) - - // 16-bit interleave step (for transposes) - #define dct_interleave16(a, b) \ - tmp = a; \ - a = _mm_unpacklo_epi16(a, b); \ - b = _mm_unpackhi_epi16(tmp, b) - - #define dct_pass(bias, shift) \ - { \ - /* even part */ \ - dct_rot(t2e, t3e, row2, row6, rot0_0, rot0_1); \ - __m128i sum04 = _mm_add_epi16(row0, row4); \ - __m128i dif04 = _mm_sub_epi16(row0, row4); \ - dct_widen(t0e, sum04); \ - dct_widen(t1e, dif04); \ - dct_wadd(x0, t0e, t3e); \ - dct_wsub(x3, t0e, t3e); \ - dct_wadd(x1, t1e, t2e); \ - dct_wsub(x2, t1e, t2e); \ - /* odd part */ \ - dct_rot(y0o, y2o, row7, row3, rot2_0, rot2_1); \ - dct_rot(y1o, y3o, row5, row1, rot3_0, rot3_1); \ - __m128i sum17 = _mm_add_epi16(row1, row7); \ - __m128i sum35 = _mm_add_epi16(row3, row5); \ - dct_rot(y4o, y5o, sum17, sum35, rot1_0, rot1_1); \ - dct_wadd(x4, y0o, y4o); \ - dct_wadd(x5, y1o, y5o); \ - dct_wadd(x6, y2o, y5o); \ - dct_wadd(x7, y3o, y4o); \ - dct_bfly32o(row0, row7, x0, x7, bias, shift); \ - dct_bfly32o(row1, row6, x1, x6, bias, shift); \ - dct_bfly32o(row2, row5, x2, x5, bias, shift); \ - dct_bfly32o(row3, row4, x3, x4, bias, shift); \ - } - - __m128i rot0_0 = dct_const(stbi__f2f(0.5411961f), stbi__f2f(0.5411961f) + stbi__f2f(-1.847759065f)); - __m128i rot0_1 = dct_const(stbi__f2f(0.5411961f) + stbi__f2f(0.765366865f), stbi__f2f(0.5411961f)); - __m128i rot1_0 = dct_const(stbi__f2f(1.175875602f) + stbi__f2f(-0.899976223f), stbi__f2f(1.175875602f)); - __m128i rot1_1 = dct_const(stbi__f2f(1.175875602f), stbi__f2f(1.175875602f) + stbi__f2f(-2.562915447f)); - __m128i rot2_0 = dct_const(stbi__f2f(-1.961570560f) + stbi__f2f(0.298631336f), stbi__f2f(-1.961570560f)); - __m128i rot2_1 = dct_const(stbi__f2f(-1.961570560f), stbi__f2f(-1.961570560f) + stbi__f2f(3.072711026f)); - __m128i rot3_0 = dct_const(stbi__f2f(-0.390180644f) + stbi__f2f(2.053119869f), stbi__f2f(-0.390180644f)); - __m128i rot3_1 = dct_const(stbi__f2f(-0.390180644f), stbi__f2f(-0.390180644f) + stbi__f2f(1.501321110f)); - - // rounding biases in column/row passes, see stbi__idct_block for explanation. - __m128i bias_0 = _mm_set1_epi32(512); - __m128i bias_1 = _mm_set1_epi32(65536 + (128 << 17)); - - // load - row0 = _mm_load_si128((const __m128i*)(data + 0 * 8)); - row1 = _mm_load_si128((const __m128i*)(data + 1 * 8)); - row2 = _mm_load_si128((const __m128i*)(data + 2 * 8)); - row3 = _mm_load_si128((const __m128i*)(data + 3 * 8)); - row4 = _mm_load_si128((const __m128i*)(data + 4 * 8)); - row5 = _mm_load_si128((const __m128i*)(data + 5 * 8)); - row6 = _mm_load_si128((const __m128i*)(data + 6 * 8)); - row7 = _mm_load_si128((const __m128i*)(data + 7 * 8)); - - // column pass - dct_pass(bias_0, 10); - - { - // 16bit 8x8 transpose pass 1 - dct_interleave16(row0, row4); - dct_interleave16(row1, row5); - dct_interleave16(row2, row6); - dct_interleave16(row3, row7); - - // transpose pass 2 - dct_interleave16(row0, row2); - dct_interleave16(row1, row3); - dct_interleave16(row4, row6); - dct_interleave16(row5, row7); - - // transpose pass 3 - dct_interleave16(row0, row1); - dct_interleave16(row2, row3); - dct_interleave16(row4, row5); - dct_interleave16(row6, row7); - } - - // row pass - dct_pass(bias_1, 17); - - { - // pack - __m128i p0 = _mm_packus_epi16(row0, row1); // a0a1a2a3...a7b0b1b2b3...b7 - __m128i p1 = _mm_packus_epi16(row2, row3); - __m128i p2 = _mm_packus_epi16(row4, row5); - __m128i p3 = _mm_packus_epi16(row6, row7); - - // 8bit 8x8 transpose pass 1 - dct_interleave8(p0, p2); // a0e0a1e1... - dct_interleave8(p1, p3); // c0g0c1g1... - - // transpose pass 2 - dct_interleave8(p0, p1); // a0c0e0g0... - dct_interleave8(p2, p3); // b0d0f0h0... - - // transpose pass 3 - dct_interleave8(p0, p2); // a0b0c0d0... - dct_interleave8(p1, p3); // a4b4c4d4... - - // store - _mm_storel_epi64((__m128i*)out, p0); - out += out_stride; - _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p0, 0x4e)); - out += out_stride; - _mm_storel_epi64((__m128i*)out, p2); - out += out_stride; - _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p2, 0x4e)); - out += out_stride; - _mm_storel_epi64((__m128i*)out, p1); - out += out_stride; - _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p1, 0x4e)); - out += out_stride; - _mm_storel_epi64((__m128i*)out, p3); - out += out_stride; - _mm_storel_epi64((__m128i*)out, _mm_shuffle_epi32(p3, 0x4e)); - } - - #undef dct_const - #undef dct_rot - #undef dct_widen - #undef dct_wadd - #undef dct_wsub - #undef dct_bfly32o - #undef dct_interleave8 - #undef dct_interleave16 - #undef dct_pass -} - - #endif // STBI_SSE2 - - #ifdef STBI_NEON - -// NEON integer IDCT. should produce bit-identical -// results to the generic C version. -static void stbi__idct_simd(stbi_uc* out, int out_stride, short data[64]) -{ - int16x8_t row0, row1, row2, row3, row4, row5, row6, row7; - - int16x4_t rot0_0 = vdup_n_s16(stbi__f2f(0.5411961f)); - int16x4_t rot0_1 = vdup_n_s16(stbi__f2f(-1.847759065f)); - int16x4_t rot0_2 = vdup_n_s16(stbi__f2f(0.765366865f)); - int16x4_t rot1_0 = vdup_n_s16(stbi__f2f(1.175875602f)); - int16x4_t rot1_1 = vdup_n_s16(stbi__f2f(-0.899976223f)); - int16x4_t rot1_2 = vdup_n_s16(stbi__f2f(-2.562915447f)); - int16x4_t rot2_0 = vdup_n_s16(stbi__f2f(-1.961570560f)); - int16x4_t rot2_1 = vdup_n_s16(stbi__f2f(-0.390180644f)); - int16x4_t rot3_0 = vdup_n_s16(stbi__f2f(0.298631336f)); - int16x4_t rot3_1 = vdup_n_s16(stbi__f2f(2.053119869f)); - int16x4_t rot3_2 = vdup_n_s16(stbi__f2f(3.072711026f)); - int16x4_t rot3_3 = vdup_n_s16(stbi__f2f(1.501321110f)); - - #define dct_long_mul(out, inq, coeff) \ - int32x4_t out##_l = vmull_s16(vget_low_s16(inq), coeff); \ - int32x4_t out##_h = vmull_s16(vget_high_s16(inq), coeff) - - #define dct_long_mac(out, acc, inq, coeff) \ - int32x4_t out##_l = vmlal_s16(acc##_l, vget_low_s16(inq), coeff); \ - int32x4_t out##_h = vmlal_s16(acc##_h, vget_high_s16(inq), coeff) - - #define dct_widen(out, inq) \ - int32x4_t out##_l = vshll_n_s16(vget_low_s16(inq), 12); \ - int32x4_t out##_h = vshll_n_s16(vget_high_s16(inq), 12) - - // wide add - #define dct_wadd(out, a, b) \ - int32x4_t out##_l = vaddq_s32(a##_l, b##_l); \ - int32x4_t out##_h = vaddq_s32(a##_h, b##_h) - - // wide sub - #define dct_wsub(out, a, b) \ - int32x4_t out##_l = vsubq_s32(a##_l, b##_l); \ - int32x4_t out##_h = vsubq_s32(a##_h, b##_h) - - // butterfly a/b, then shift using "shiftop" by "s" and pack - #define dct_bfly32o(out0, out1, a, b, shiftop, s) \ - { \ - dct_wadd(sum, a, b); \ - dct_wsub(dif, a, b); \ - out0 = vcombine_s16(shiftop(sum_l, s), shiftop(sum_h, s)); \ - out1 = vcombine_s16(shiftop(dif_l, s), shiftop(dif_h, s)); \ - } - - #define dct_pass(shiftop, shift) \ - { \ - /* even part */ \ - int16x8_t sum26 = vaddq_s16(row2, row6); \ - dct_long_mul(p1e, sum26, rot0_0); \ - dct_long_mac(t2e, p1e, row6, rot0_1); \ - dct_long_mac(t3e, p1e, row2, rot0_2); \ - int16x8_t sum04 = vaddq_s16(row0, row4); \ - int16x8_t dif04 = vsubq_s16(row0, row4); \ - dct_widen(t0e, sum04); \ - dct_widen(t1e, dif04); \ - dct_wadd(x0, t0e, t3e); \ - dct_wsub(x3, t0e, t3e); \ - dct_wadd(x1, t1e, t2e); \ - dct_wsub(x2, t1e, t2e); \ - /* odd part */ \ - int16x8_t sum15 = vaddq_s16(row1, row5); \ - int16x8_t sum17 = vaddq_s16(row1, row7); \ - int16x8_t sum35 = vaddq_s16(row3, row5); \ - int16x8_t sum37 = vaddq_s16(row3, row7); \ - int16x8_t sumodd = vaddq_s16(sum17, sum35); \ - dct_long_mul(p5o, sumodd, rot1_0); \ - dct_long_mac(p1o, p5o, sum17, rot1_1); \ - dct_long_mac(p2o, p5o, sum35, rot1_2); \ - dct_long_mul(p3o, sum37, rot2_0); \ - dct_long_mul(p4o, sum15, rot2_1); \ - dct_wadd(sump13o, p1o, p3o); \ - dct_wadd(sump24o, p2o, p4o); \ - dct_wadd(sump23o, p2o, p3o); \ - dct_wadd(sump14o, p1o, p4o); \ - dct_long_mac(x4, sump13o, row7, rot3_0); \ - dct_long_mac(x5, sump24o, row5, rot3_1); \ - dct_long_mac(x6, sump23o, row3, rot3_2); \ - dct_long_mac(x7, sump14o, row1, rot3_3); \ - dct_bfly32o(row0, row7, x0, x7, shiftop, shift); \ - dct_bfly32o(row1, row6, x1, x6, shiftop, shift); \ - dct_bfly32o(row2, row5, x2, x5, shiftop, shift); \ - dct_bfly32o(row3, row4, x3, x4, shiftop, shift); \ - } - - // load - row0 = vld1q_s16(data + 0 * 8); - row1 = vld1q_s16(data + 1 * 8); - row2 = vld1q_s16(data + 2 * 8); - row3 = vld1q_s16(data + 3 * 8); - row4 = vld1q_s16(data + 4 * 8); - row5 = vld1q_s16(data + 5 * 8); - row6 = vld1q_s16(data + 6 * 8); - row7 = vld1q_s16(data + 7 * 8); - - // add DC bias - row0 = vaddq_s16(row0, vsetq_lane_s16(1024, vdupq_n_s16(0), 0)); - - // column pass - dct_pass(vrshrn_n_s32, 10); - - // 16bit 8x8 transpose - { - // these three map to a single VTRN.16, VTRN.32, and VSWP, respectively. - // whether compilers actually get this is another story, sadly. - #define dct_trn16(x, y) \ - { \ - int16x8x2_t t = vtrnq_s16(x, y); \ - x = t.val[0]; \ - y = t.val[1]; \ - } - #define dct_trn32(x, y) \ - { \ - int32x4x2_t t = vtrnq_s32(vreinterpretq_s32_s16(x), vreinterpretq_s32_s16(y)); \ - x = vreinterpretq_s16_s32(t.val[0]); \ - y = vreinterpretq_s16_s32(t.val[1]); \ - } - #define dct_trn64(x, y) \ - { \ - int16x8_t x0 = x; \ - int16x8_t y0 = y; \ - x = vcombine_s16(vget_low_s16(x0), vget_low_s16(y0)); \ - y = vcombine_s16(vget_high_s16(x0), vget_high_s16(y0)); \ - } - - // pass 1 - dct_trn16(row0, row1); // a0b0a2b2a4b4a6b6 - dct_trn16(row2, row3); - dct_trn16(row4, row5); - dct_trn16(row6, row7); - - // pass 2 - dct_trn32(row0, row2); // a0b0c0d0a4b4c4d4 - dct_trn32(row1, row3); - dct_trn32(row4, row6); - dct_trn32(row5, row7); - - // pass 3 - dct_trn64(row0, row4); // a0b0c0d0e0f0g0h0 - dct_trn64(row1, row5); - dct_trn64(row2, row6); - dct_trn64(row3, row7); - - #undef dct_trn16 - #undef dct_trn32 - #undef dct_trn64 - } - - // row pass - // vrshrn_n_s32 only supports shifts up to 16, we need - // 17. so do a non-rounding shift of 16 first then follow - // up with a rounding shift by 1. - dct_pass(vshrn_n_s32, 16); - - { - // pack and round - uint8x8_t p0 = vqrshrun_n_s16(row0, 1); - uint8x8_t p1 = vqrshrun_n_s16(row1, 1); - uint8x8_t p2 = vqrshrun_n_s16(row2, 1); - uint8x8_t p3 = vqrshrun_n_s16(row3, 1); - uint8x8_t p4 = vqrshrun_n_s16(row4, 1); - uint8x8_t p5 = vqrshrun_n_s16(row5, 1); - uint8x8_t p6 = vqrshrun_n_s16(row6, 1); - uint8x8_t p7 = vqrshrun_n_s16(row7, 1); - - // again, these can translate into one instruction, but often don't. - #define dct_trn8_8(x, y) \ - { \ - uint8x8x2_t t = vtrn_u8(x, y); \ - x = t.val[0]; \ - y = t.val[1]; \ - } - #define dct_trn8_16(x, y) \ - { \ - uint16x4x2_t t = vtrn_u16(vreinterpret_u16_u8(x), vreinterpret_u16_u8(y)); \ - x = vreinterpret_u8_u16(t.val[0]); \ - y = vreinterpret_u8_u16(t.val[1]); \ - } - #define dct_trn8_32(x, y) \ - { \ - uint32x2x2_t t = vtrn_u32(vreinterpret_u32_u8(x), vreinterpret_u32_u8(y)); \ - x = vreinterpret_u8_u32(t.val[0]); \ - y = vreinterpret_u8_u32(t.val[1]); \ - } - - // sadly can't use interleaved stores here since we only write - // 8 bytes to each scan line! - - // 8x8 8-bit transpose pass 1 - dct_trn8_8(p0, p1); - dct_trn8_8(p2, p3); - dct_trn8_8(p4, p5); - dct_trn8_8(p6, p7); - - // pass 2 - dct_trn8_16(p0, p2); - dct_trn8_16(p1, p3); - dct_trn8_16(p4, p6); - dct_trn8_16(p5, p7); - - // pass 3 - dct_trn8_32(p0, p4); - dct_trn8_32(p1, p5); - dct_trn8_32(p2, p6); - dct_trn8_32(p3, p7); - - // store - vst1_u8(out, p0); - out += out_stride; - vst1_u8(out, p1); - out += out_stride; - vst1_u8(out, p2); - out += out_stride; - vst1_u8(out, p3); - out += out_stride; - vst1_u8(out, p4); - out += out_stride; - vst1_u8(out, p5); - out += out_stride; - vst1_u8(out, p6); - out += out_stride; - vst1_u8(out, p7); - - #undef dct_trn8_8 - #undef dct_trn8_16 - #undef dct_trn8_32 - } - - #undef dct_long_mul - #undef dct_long_mac - #undef dct_widen - #undef dct_wadd - #undef dct_wsub - #undef dct_bfly32o - #undef dct_pass -} - - #endif // STBI_NEON - - #define STBI__MARKER_none 0xff -// if there's a pending marker from the entropy stream, return that -// otherwise, fetch from the stream and get a marker. if there's no -// marker, return 0xff, which is never a valid marker value -static stbi_uc stbi__get_marker(stbi__jpeg* j) -{ - stbi_uc x; - if (j->marker != STBI__MARKER_none) - { - x = j->marker; - j->marker = STBI__MARKER_none; - return x; - } - x = stbi__get8(j->s); - if (x != 0xff) - return STBI__MARKER_none; - while (x == 0xff) - x = stbi__get8(j->s); // consume repeated 0xff fill bytes - return x; -} - - // in each scan, we'll have scan_n components, and the order - // of the components is specified by order[] - #define STBI__RESTART(x) ((x) >= 0xd0 && (x) <= 0xd7) - -// after a restart interval, stbi__jpeg_reset the entropy decoder and -// the dc prediction -static void stbi__jpeg_reset(stbi__jpeg* j) -{ - j->code_bits = 0; - j->code_buffer = 0; - j->nomore = 0; - j->img_comp[0].dc_pred = j->img_comp[1].dc_pred = j->img_comp[2].dc_pred = j->img_comp[3].dc_pred = 0; - j->marker = STBI__MARKER_none; - j->todo = j->restart_interval ? j->restart_interval : 0x7fffffff; - j->eob_run = 0; - // no more than 1<<31 MCUs if no restart_interal? that's plenty safe, - // since we don't even allow 1<<30 pixels -} - -static int stbi__parse_entropy_coded_data(stbi__jpeg* z) -{ - stbi__jpeg_reset(z); - if (!z->progressive) - { - if (z->scan_n == 1) - { - int i, j; - STBI_SIMD_ALIGN(short, data[64]); - int n = z->order[0]; - // non-interleaved data, we just need to process one block at a time, - // in trivial scanline order - // number of blocks to do just depends on how many actual "pixels" this - // component has, independent of interleaved MCU blocking and such - int w = (z->img_comp[n].x + 7) >> 3; - int h = (z->img_comp[n].y + 7) >> 3; - for (j = 0; j < h; ++j) - { - for (i = 0; i < w; ++i) - { - int ha = z->img_comp[n].ha; - if (!stbi__jpeg_decode_block(z, data, z->huff_dc + z->img_comp[n].hd, z->huff_ac + ha, z->fast_ac[ha], n, - z->dequant[z->img_comp[n].tq])) - return 0; - z->idct_block_kernel(z->img_comp[n].data + z->img_comp[n].w2 * j * 8 + i * 8, z->img_comp[n].w2, data); - // every data block is an MCU, so countdown the restart interval - if (--z->todo <= 0) - { - if (z->code_bits < 24) - stbi__grow_buffer_unsafe(z); - // if it's NOT a restart, then just bail, so we get corrupt data - // rather than no data - if (!STBI__RESTART(z->marker)) - return 1; - stbi__jpeg_reset(z); - } - } - } - return 1; - } - else - { // interleaved - int i, j, k, x, y; - STBI_SIMD_ALIGN(short, data[64]); - for (j = 0; j < z->img_mcu_y; ++j) - { - for (i = 0; i < z->img_mcu_x; ++i) - { - // scan an interleaved mcu... process scan_n components in order - for (k = 0; k < z->scan_n; ++k) - { - int n = z->order[k]; - // scan out an mcu's worth of this component; that's just determined - // by the basic H and V specified for the component - for (y = 0; y < z->img_comp[n].v; ++y) - { - for (x = 0; x < z->img_comp[n].h; ++x) - { - int x2 = (i * z->img_comp[n].h + x) * 8; - int y2 = (j * z->img_comp[n].v + y) * 8; - int ha = z->img_comp[n].ha; - if (!stbi__jpeg_decode_block(z, data, z->huff_dc + z->img_comp[n].hd, z->huff_ac + ha, z->fast_ac[ha], n, - z->dequant[z->img_comp[n].tq])) - return 0; - z->idct_block_kernel(z->img_comp[n].data + z->img_comp[n].w2 * y2 + x2, z->img_comp[n].w2, data); - } - } - } - // after all interleaved components, that's an interleaved MCU, - // so now count down the restart interval - if (--z->todo <= 0) - { - if (z->code_bits < 24) - stbi__grow_buffer_unsafe(z); - if (!STBI__RESTART(z->marker)) - return 1; - stbi__jpeg_reset(z); - } - } - } - return 1; - } - } - else - { - if (z->scan_n == 1) - { - int i, j; - int n = z->order[0]; - // non-interleaved data, we just need to process one block at a time, - // in trivial scanline order - // number of blocks to do just depends on how many actual "pixels" this - // component has, independent of interleaved MCU blocking and such - int w = (z->img_comp[n].x + 7) >> 3; - int h = (z->img_comp[n].y + 7) >> 3; - for (j = 0; j < h; ++j) - { - for (i = 0; i < w; ++i) - { - short* data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w); - if (z->spec_start == 0) - { - if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n)) - return 0; - } - else - { - int ha = z->img_comp[n].ha; - if (!stbi__jpeg_decode_block_prog_ac(z, data, &z->huff_ac[ha], z->fast_ac[ha])) - return 0; - } - // every data block is an MCU, so countdown the restart interval - if (--z->todo <= 0) - { - if (z->code_bits < 24) - stbi__grow_buffer_unsafe(z); - if (!STBI__RESTART(z->marker)) - return 1; - stbi__jpeg_reset(z); - } - } - } - return 1; - } - else - { // interleaved - int i, j, k, x, y; - for (j = 0; j < z->img_mcu_y; ++j) - { - for (i = 0; i < z->img_mcu_x; ++i) - { - // scan an interleaved mcu... process scan_n components in order - for (k = 0; k < z->scan_n; ++k) - { - int n = z->order[k]; - // scan out an mcu's worth of this component; that's just determined - // by the basic H and V specified for the component - for (y = 0; y < z->img_comp[n].v; ++y) - { - for (x = 0; x < z->img_comp[n].h; ++x) - { - int x2 = (i * z->img_comp[n].h + x); - int y2 = (j * z->img_comp[n].v + y); - short* data = z->img_comp[n].coeff + 64 * (x2 + y2 * z->img_comp[n].coeff_w); - if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n)) - return 0; - } - } - } - // after all interleaved components, that's an interleaved MCU, - // so now count down the restart interval - if (--z->todo <= 0) - { - if (z->code_bits < 24) - stbi__grow_buffer_unsafe(z); - if (!STBI__RESTART(z->marker)) - return 1; - stbi__jpeg_reset(z); - } - } - } - return 1; - } - } -} - -static void stbi__jpeg_dequantize(short* data, stbi__uint16* dequant) -{ - int i; - for (i = 0; i < 64; ++i) - data[i] *= dequant[i]; -} - -static void stbi__jpeg_finish(stbi__jpeg* z) -{ - if (z->progressive) - { - // dequantize and idct the data - int i, j, n; - for (n = 0; n < z->s->img_n; ++n) - { - int w = (z->img_comp[n].x + 7) >> 3; - int h = (z->img_comp[n].y + 7) >> 3; - for (j = 0; j < h; ++j) - { - for (i = 0; i < w; ++i) - { - short* data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w); - stbi__jpeg_dequantize(data, z->dequant[z->img_comp[n].tq]); - z->idct_block_kernel(z->img_comp[n].data + z->img_comp[n].w2 * j * 8 + i * 8, z->img_comp[n].w2, data); - } - } - } - } -} - -static int stbi__process_marker(stbi__jpeg* z, int m) -{ - int L; - switch (m) - { - case STBI__MARKER_none: // no marker found - return stbi__err("expected marker", "Corrupt JPEG"); - - case 0xDD: // DRI - specify restart interval - if (stbi__get16be(z->s) != 4) - return stbi__err("bad DRI len", "Corrupt JPEG"); - z->restart_interval = stbi__get16be(z->s); - return 1; - - case 0xDB: // DQT - define quantization table - L = stbi__get16be(z->s) - 2; - while (L > 0) - { - int q = stbi__get8(z->s); - int p = q >> 4, sixteen = (p != 0); - int t = q & 15, i; - if (p != 0 && p != 1) - return stbi__err("bad DQT type", "Corrupt JPEG"); - if (t > 3) - return stbi__err("bad DQT table", "Corrupt JPEG"); - - for (i = 0; i < 64; ++i) - z->dequant[t][stbi__jpeg_dezigzag[i]] = (stbi__uint16)(sixteen ? stbi__get16be(z->s) : stbi__get8(z->s)); - L -= (sixteen ? 129 : 65); - } - return L == 0; - - case 0xC4: // DHT - define huffman table - L = stbi__get16be(z->s) - 2; - while (L > 0) - { - stbi_uc* v; - int sizes[16], i, n = 0; - int q = stbi__get8(z->s); - int tc = q >> 4; - int th = q & 15; - if (tc > 1 || th > 3) - return stbi__err("bad DHT header", "Corrupt JPEG"); - for (i = 0; i < 16; ++i) - { - sizes[i] = stbi__get8(z->s); - n += sizes[i]; - } - if (n > 256) - return stbi__err("bad DHT header", "Corrupt JPEG"); // Loop over i < n would write past end of values! - L -= 17; - if (tc == 0) - { - if (!stbi__build_huffman(z->huff_dc + th, sizes)) - return 0; - v = z->huff_dc[th].values; - } - else - { - if (!stbi__build_huffman(z->huff_ac + th, sizes)) - return 0; - v = z->huff_ac[th].values; - } - for (i = 0; i < n; ++i) - v[i] = stbi__get8(z->s); - if (tc != 0) - stbi__build_fast_ac(z->fast_ac[th], z->huff_ac + th); - L -= n; - } - return L == 0; - } - - // check for comment block or APP blocks - if ((m >= 0xE0 && m <= 0xEF) || m == 0xFE) - { - L = stbi__get16be(z->s); - if (L < 2) - { - if (m == 0xFE) - return stbi__err("bad COM len", "Corrupt JPEG"); - else - return stbi__err("bad APP len", "Corrupt JPEG"); - } - L -= 2; - - if (m == 0xE0 && L >= 5) - { // JFIF APP0 segment - static const unsigned char tag[5] = {'J', 'F', 'I', 'F', '\0'}; - int ok = 1; - int i; - for (i = 0; i < 5; ++i) - if (stbi__get8(z->s) != tag[i]) - ok = 0; - L -= 5; - if (ok) - z->jfif = 1; - } - else if (m == 0xEE && L >= 12) - { // Adobe APP14 segment - static const unsigned char tag[6] = {'A', 'd', 'o', 'b', 'e', '\0'}; - int ok = 1; - int i; - for (i = 0; i < 6; ++i) - if (stbi__get8(z->s) != tag[i]) - ok = 0; - L -= 6; - if (ok) - { - stbi__get8(z->s); // version - stbi__get16be(z->s); // flags0 - stbi__get16be(z->s); // flags1 - z->app14_color_transform = stbi__get8(z->s); // color transform - L -= 6; - } - } - - stbi__skip(z->s, L); - return 1; - } - - return stbi__err("unknown marker", "Corrupt JPEG"); -} - -// after we see SOS -static int stbi__process_scan_header(stbi__jpeg* z) -{ - int i; - int Ls = stbi__get16be(z->s); - z->scan_n = stbi__get8(z->s); - if (z->scan_n < 1 || z->scan_n > 4 || z->scan_n > (int)z->s->img_n) - return stbi__err("bad SOS component count", "Corrupt JPEG"); - if (Ls != 6 + 2 * z->scan_n) - return stbi__err("bad SOS len", "Corrupt JPEG"); - for (i = 0; i < z->scan_n; ++i) - { - int id = stbi__get8(z->s), which; - int q = stbi__get8(z->s); - for (which = 0; which < z->s->img_n; ++which) - if (z->img_comp[which].id == id) - break; - if (which == z->s->img_n) - return 0; // no match - z->img_comp[which].hd = q >> 4; - if (z->img_comp[which].hd > 3) - return stbi__err("bad DC huff", "Corrupt JPEG"); - z->img_comp[which].ha = q & 15; - if (z->img_comp[which].ha > 3) - return stbi__err("bad AC huff", "Corrupt JPEG"); - z->order[i] = which; - } - - { - int aa; - z->spec_start = stbi__get8(z->s); - z->spec_end = stbi__get8(z->s); // should be 63, but might be 0 - aa = stbi__get8(z->s); - z->succ_high = (aa >> 4); - z->succ_low = (aa & 15); - if (z->progressive) - { - if (z->spec_start > 63 || z->spec_end > 63 || z->spec_start > z->spec_end || z->succ_high > 13 || z->succ_low > 13) - return stbi__err("bad SOS", "Corrupt JPEG"); - } - else - { - if (z->spec_start != 0) - return stbi__err("bad SOS", "Corrupt JPEG"); - if (z->succ_high != 0 || z->succ_low != 0) - return stbi__err("bad SOS", "Corrupt JPEG"); - z->spec_end = 63; - } - } - - return 1; -} - -static int stbi__free_jpeg_components(stbi__jpeg* z, int ncomp, int why) -{ - int i; - for (i = 0; i < ncomp; ++i) - { - if (z->img_comp[i].raw_data) - { - STBI_FREE(z->img_comp[i].raw_data); - z->img_comp[i].raw_data = NULL; - z->img_comp[i].data = NULL; - } - if (z->img_comp[i].raw_coeff) - { - STBI_FREE(z->img_comp[i].raw_coeff); - z->img_comp[i].raw_coeff = 0; - z->img_comp[i].coeff = 0; - } - if (z->img_comp[i].linebuf) - { - STBI_FREE(z->img_comp[i].linebuf); - z->img_comp[i].linebuf = NULL; - } - } - return why; -} - -static int stbi__process_frame_header(stbi__jpeg* z, int scan) -{ - stbi__context* s = z->s; - int Lf, p, i, q, h_max = 1, v_max = 1, c; - Lf = stbi__get16be(s); - if (Lf < 11) - return stbi__err("bad SOF len", "Corrupt JPEG"); // JPEG - p = stbi__get8(s); - if (p != 8) - return stbi__err("only 8-bit", "JPEG format not supported: 8-bit only"); // JPEG baseline - s->img_y = stbi__get16be(s); - if (s->img_y == 0) - return stbi__err("no header height", "JPEG format not supported: delayed height"); // Legal, but we don't handle it--but neither does IJG - s->img_x = stbi__get16be(s); - if (s->img_x == 0) - return stbi__err("0 width", "Corrupt JPEG"); // JPEG requires - if (s->img_y > STBI_MAX_DIMENSIONS) - return stbi__err("too large", "Very large image (corrupt?)"); - if (s->img_x > STBI_MAX_DIMENSIONS) - return stbi__err("too large", "Very large image (corrupt?)"); - c = stbi__get8(s); - if (c != 3 && c != 1 && c != 4) - return stbi__err("bad component count", "Corrupt JPEG"); - s->img_n = c; - for (i = 0; i < c; ++i) - { - z->img_comp[i].data = NULL; - z->img_comp[i].linebuf = NULL; - } - - if (Lf != 8 + 3 * s->img_n) - return stbi__err("bad SOF len", "Corrupt JPEG"); - - z->rgb = 0; - for (i = 0; i < s->img_n; ++i) - { - static const unsigned char rgb[3] = {'R', 'G', 'B'}; - z->img_comp[i].id = stbi__get8(s); - if (s->img_n == 3 && z->img_comp[i].id == rgb[i]) - ++z->rgb; - q = stbi__get8(s); - z->img_comp[i].h = (q >> 4); - if (!z->img_comp[i].h || z->img_comp[i].h > 4) - return stbi__err("bad H", "Corrupt JPEG"); - z->img_comp[i].v = q & 15; - if (!z->img_comp[i].v || z->img_comp[i].v > 4) - return stbi__err("bad V", "Corrupt JPEG"); - z->img_comp[i].tq = stbi__get8(s); - if (z->img_comp[i].tq > 3) - return stbi__err("bad TQ", "Corrupt JPEG"); - } - - if (scan != STBI__SCAN_load) - return 1; - - if (!stbi__mad3sizes_valid(s->img_x, s->img_y, s->img_n, 0)) - return stbi__err("too large", "Image too large to decode"); - - for (i = 0; i < s->img_n; ++i) - { - if (z->img_comp[i].h > h_max) - h_max = z->img_comp[i].h; - if (z->img_comp[i].v > v_max) - v_max = z->img_comp[i].v; - } - - // check that plane subsampling factors are integer ratios; our resamplers can't deal with fractional ratios - // and I've never seen a non-corrupted JPEG file actually use them - for (i = 0; i < s->img_n; ++i) - { - if (h_max % z->img_comp[i].h != 0) - return stbi__err("bad H", "Corrupt JPEG"); - if (v_max % z->img_comp[i].v != 0) - return stbi__err("bad V", "Corrupt JPEG"); - } - - // compute interleaved mcu info - z->img_h_max = h_max; - z->img_v_max = v_max; - z->img_mcu_w = h_max * 8; - z->img_mcu_h = v_max * 8; - // these sizes can't be more than 17 bits - z->img_mcu_x = (s->img_x + z->img_mcu_w - 1) / z->img_mcu_w; - z->img_mcu_y = (s->img_y + z->img_mcu_h - 1) / z->img_mcu_h; - - for (i = 0; i < s->img_n; ++i) - { - // number of effective pixels (e.g. for non-interleaved MCU) - z->img_comp[i].x = (s->img_x * z->img_comp[i].h + h_max - 1) / h_max; - z->img_comp[i].y = (s->img_y * z->img_comp[i].v + v_max - 1) / v_max; - // to simplify generation, we'll allocate enough memory to decode - // the bogus oversized data from using interleaved MCUs and their - // big blocks (e.g. a 16x16 iMCU on an image of width 33); we won't - // discard the extra data until colorspace conversion - // - // img_mcu_x, img_mcu_y: <=17 bits; comp[i].h and .v are <=4 (checked earlier) - // so these muls can't overflow with 32-bit ints (which we require) - z->img_comp[i].w2 = z->img_mcu_x * z->img_comp[i].h * 8; - z->img_comp[i].h2 = z->img_mcu_y * z->img_comp[i].v * 8; - z->img_comp[i].coeff = 0; - z->img_comp[i].raw_coeff = 0; - z->img_comp[i].linebuf = NULL; - z->img_comp[i].raw_data = stbi__malloc_mad2(z->img_comp[i].w2, z->img_comp[i].h2, 15); - if (z->img_comp[i].raw_data == NULL) - return stbi__free_jpeg_components(z, i + 1, stbi__err("outofmem", "Out of memory")); - // align blocks for idct using mmx/sse - z->img_comp[i].data = (stbi_uc*)(((size_t)z->img_comp[i].raw_data + 15) & ~15); - if (z->progressive) - { - // w2, h2 are multiples of 8 (see above) - z->img_comp[i].coeff_w = z->img_comp[i].w2 / 8; - z->img_comp[i].coeff_h = z->img_comp[i].h2 / 8; - z->img_comp[i].raw_coeff = stbi__malloc_mad3(z->img_comp[i].w2, z->img_comp[i].h2, sizeof(short), 15); - if (z->img_comp[i].raw_coeff == NULL) - return stbi__free_jpeg_components(z, i + 1, stbi__err("outofmem", "Out of memory")); - z->img_comp[i].coeff = (short*)(((size_t)z->img_comp[i].raw_coeff + 15) & ~15); - } - } - - return 1; -} - - // use comparisons since in some cases we handle more than one case (e.g. SOF) - #define stbi__DNL(x) ((x) == 0xdc) - #define stbi__SOI(x) ((x) == 0xd8) - #define stbi__EOI(x) ((x) == 0xd9) - #define stbi__SOF(x) ((x) == 0xc0 || (x) == 0xc1 || (x) == 0xc2) - #define stbi__SOS(x) ((x) == 0xda) - - #define stbi__SOF_progressive(x) ((x) == 0xc2) - -static int stbi__decode_jpeg_header(stbi__jpeg* z, int scan) -{ - int m; - z->jfif = 0; - z->app14_color_transform = -1; // valid values are 0,1,2 - z->marker = STBI__MARKER_none; // initialize cached marker to empty - m = stbi__get_marker(z); - if (!stbi__SOI(m)) - return stbi__err("no SOI", "Corrupt JPEG"); - if (scan == STBI__SCAN_type) - return 1; - m = stbi__get_marker(z); - while (!stbi__SOF(m)) - { - if (!stbi__process_marker(z, m)) - return 0; - m = stbi__get_marker(z); - while (m == STBI__MARKER_none) - { - // some files have extra padding after their blocks, so ok, we'll scan - if (stbi__at_eof(z->s)) - return stbi__err("no SOF", "Corrupt JPEG"); - m = stbi__get_marker(z); - } - } - z->progressive = stbi__SOF_progressive(m); - if (!stbi__process_frame_header(z, scan)) - return 0; - return 1; -} - -static stbi_uc stbi__skip_jpeg_junk_at_end(stbi__jpeg* j) -{ - // some JPEGs have junk at end, skip over it but if we find what looks - // like a valid marker, resume there - while (!stbi__at_eof(j->s)) - { - stbi_uc x = stbi__get8(j->s); - while (x == 0xff) - { // might be a marker - if (stbi__at_eof(j->s)) - return STBI__MARKER_none; - x = stbi__get8(j->s); - if (x != 0x00 && x != 0xff) - { - // not a stuffed zero or lead-in to another marker, looks - // like an actual marker, return it - return x; - } - // stuffed zero has x=0 now which ends the loop, meaning we go - // back to regular scan loop. - // repeated 0xff keeps trying to read the next byte of the marker. - } - } - return STBI__MARKER_none; -} - -// decode image to YCbCr format -static int stbi__decode_jpeg_image(stbi__jpeg* j) -{ - int m; - for (m = 0; m < 4; m++) - { - j->img_comp[m].raw_data = NULL; - j->img_comp[m].raw_coeff = NULL; - } - j->restart_interval = 0; - if (!stbi__decode_jpeg_header(j, STBI__SCAN_load)) - return 0; - m = stbi__get_marker(j); - while (!stbi__EOI(m)) - { - if (stbi__SOS(m)) - { - if (!stbi__process_scan_header(j)) - return 0; - if (!stbi__parse_entropy_coded_data(j)) - return 0; - if (j->marker == STBI__MARKER_none) - { - j->marker = stbi__skip_jpeg_junk_at_end(j); - // if we reach eof without hitting a marker, stbi__get_marker() below will fail and we'll eventually return 0 - } - m = stbi__get_marker(j); - if (STBI__RESTART(m)) - m = stbi__get_marker(j); - } - else if (stbi__DNL(m)) - { - int Ld = stbi__get16be(j->s); - stbi__uint32 NL = stbi__get16be(j->s); - if (Ld != 4) - return stbi__err("bad DNL len", "Corrupt JPEG"); - if (NL != j->s->img_y) - return stbi__err("bad DNL height", "Corrupt JPEG"); - m = stbi__get_marker(j); - } - else - { - if (!stbi__process_marker(j, m)) - return 1; - m = stbi__get_marker(j); - } - } - if (j->progressive) - stbi__jpeg_finish(j); - return 1; -} - -// static jfif-centered resampling (across block boundaries) - -typedef stbi_uc* (*resample_row_func)(stbi_uc* out, stbi_uc* in0, stbi_uc* in1, int w, int hs); - - #define stbi__div4(x) ((stbi_uc)((x) >> 2)) - -static stbi_uc* resample_row_1(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) -{ - STBI_NOTUSED(out); - STBI_NOTUSED(in_far); - STBI_NOTUSED(w); - STBI_NOTUSED(hs); - return in_near; -} - -static stbi_uc* stbi__resample_row_v_2(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) -{ - // need to generate two samples vertically for every one in input - int i; - STBI_NOTUSED(hs); - for (i = 0; i < w; ++i) - out[i] = stbi__div4(3 * in_near[i] + in_far[i] + 2); - return out; -} - -static stbi_uc* stbi__resample_row_h_2(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) -{ - // need to generate two samples horizontally for every one in input - int i; - stbi_uc* input = in_near; - - if (w == 1) - { - // if only one sample, can't do any interpolation - out[0] = out[1] = input[0]; - return out; - } - - out[0] = input[0]; - out[1] = stbi__div4(input[0] * 3 + input[1] + 2); - for (i = 1; i < w - 1; ++i) - { - int n = 3 * input[i] + 2; - out[i * 2 + 0] = stbi__div4(n + input[i - 1]); - out[i * 2 + 1] = stbi__div4(n + input[i + 1]); - } - out[i * 2 + 0] = stbi__div4(input[w - 2] * 3 + input[w - 1] + 2); - out[i * 2 + 1] = input[w - 1]; - - STBI_NOTUSED(in_far); - STBI_NOTUSED(hs); - - return out; -} - - #define stbi__div16(x) ((stbi_uc)((x) >> 4)) - -static stbi_uc* stbi__resample_row_hv_2(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) -{ - // need to generate 2x2 samples for every one in input - int i, t0, t1; - if (w == 1) - { - out[0] = out[1] = stbi__div4(3 * in_near[0] + in_far[0] + 2); - return out; - } - - t1 = 3 * in_near[0] + in_far[0]; - out[0] = stbi__div4(t1 + 2); - for (i = 1; i < w; ++i) - { - t0 = t1; - t1 = 3 * in_near[i] + in_far[i]; - out[i * 2 - 1] = stbi__div16(3 * t0 + t1 + 8); - out[i * 2] = stbi__div16(3 * t1 + t0 + 8); - } - out[w * 2 - 1] = stbi__div4(t1 + 2); - - STBI_NOTUSED(hs); - - return out; -} - - #if defined(STBI_SSE2) || defined(STBI_NEON) -static stbi_uc* stbi__resample_row_hv_2_simd(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) -{ - // need to generate 2x2 samples for every one in input - int i = 0, t0, t1; - - if (w == 1) - { - out[0] = out[1] = stbi__div4(3 * in_near[0] + in_far[0] + 2); - return out; - } - - t1 = 3 * in_near[0] + in_far[0]; - // process groups of 8 pixels for as long as we can. - // note we can't handle the last pixel in a row in this loop - // because we need to handle the filter boundary conditions. - for (; i < ((w - 1) & ~7); i += 8) - { - #if defined(STBI_SSE2) - // load and perform the vertical filtering pass - // this uses 3*x + y = 4*x + (y - x) - __m128i zero = _mm_setzero_si128(); - __m128i farb = _mm_loadl_epi64((__m128i*)(in_far + i)); - __m128i nearb = _mm_loadl_epi64((__m128i*)(in_near + i)); - __m128i farw = _mm_unpacklo_epi8(farb, zero); - __m128i nearw = _mm_unpacklo_epi8(nearb, zero); - __m128i diff = _mm_sub_epi16(farw, nearw); - __m128i nears = _mm_slli_epi16(nearw, 2); - __m128i curr = _mm_add_epi16(nears, diff); // current row - - // horizontal filter works the same based on shifted vers of current - // row. "prev" is current row shifted right by 1 pixel; we need to - // insert the previous pixel value (from t1). - // "next" is current row shifted left by 1 pixel, with first pixel - // of next block of 8 pixels added in. - __m128i prv0 = _mm_slli_si128(curr, 2); - __m128i nxt0 = _mm_srli_si128(curr, 2); - __m128i prev = _mm_insert_epi16(prv0, t1, 0); - __m128i next = _mm_insert_epi16(nxt0, 3 * in_near[i + 8] + in_far[i + 8], 7); - - // horizontal filter, polyphase implementation since it's convenient: - // even pixels = 3*cur + prev = cur*4 + (prev - cur) - // odd pixels = 3*cur + next = cur*4 + (next - cur) - // note the shared term. - __m128i bias = _mm_set1_epi16(8); - __m128i curs = _mm_slli_epi16(curr, 2); - __m128i prvd = _mm_sub_epi16(prev, curr); - __m128i nxtd = _mm_sub_epi16(next, curr); - __m128i curb = _mm_add_epi16(curs, bias); - __m128i even = _mm_add_epi16(prvd, curb); - __m128i odd = _mm_add_epi16(nxtd, curb); - - // interleave even and odd pixels, then undo scaling. - __m128i int0 = _mm_unpacklo_epi16(even, odd); - __m128i int1 = _mm_unpackhi_epi16(even, odd); - __m128i de0 = _mm_srli_epi16(int0, 4); - __m128i de1 = _mm_srli_epi16(int1, 4); - - // pack and write output - __m128i outv = _mm_packus_epi16(de0, de1); - _mm_storeu_si128((__m128i*)(out + i * 2), outv); - #elif defined(STBI_NEON) - // load and perform the vertical filtering pass - // this uses 3*x + y = 4*x + (y - x) - uint8x8_t farb = vld1_u8(in_far + i); - uint8x8_t nearb = vld1_u8(in_near + i); - int16x8_t diff = vreinterpretq_s16_u16(vsubl_u8(farb, nearb)); - int16x8_t nears = vreinterpretq_s16_u16(vshll_n_u8(nearb, 2)); - int16x8_t curr = vaddq_s16(nears, diff); // current row - - // horizontal filter works the same based on shifted vers of current - // row. "prev" is current row shifted right by 1 pixel; we need to - // insert the previous pixel value (from t1). - // "next" is current row shifted left by 1 pixel, with first pixel - // of next block of 8 pixels added in. - int16x8_t prv0 = vextq_s16(curr, curr, 7); - int16x8_t nxt0 = vextq_s16(curr, curr, 1); - int16x8_t prev = vsetq_lane_s16(t1, prv0, 0); - int16x8_t next = vsetq_lane_s16(3 * in_near[i + 8] + in_far[i + 8], nxt0, 7); - - // horizontal filter, polyphase implementation since it's convenient: - // even pixels = 3*cur + prev = cur*4 + (prev - cur) - // odd pixels = 3*cur + next = cur*4 + (next - cur) - // note the shared term. - int16x8_t curs = vshlq_n_s16(curr, 2); - int16x8_t prvd = vsubq_s16(prev, curr); - int16x8_t nxtd = vsubq_s16(next, curr); - int16x8_t even = vaddq_s16(curs, prvd); - int16x8_t odd = vaddq_s16(curs, nxtd); - - // undo scaling and round, then store with even/odd phases interleaved - uint8x8x2_t o; - o.val[0] = vqrshrun_n_s16(even, 4); - o.val[1] = vqrshrun_n_s16(odd, 4); - vst2_u8(out + i * 2, o); - #endif - - // "previous" value for next iter - t1 = 3 * in_near[i + 7] + in_far[i + 7]; - } - - t0 = t1; - t1 = 3 * in_near[i] + in_far[i]; - out[i * 2] = stbi__div16(3 * t1 + t0 + 8); - - for (++i; i < w; ++i) - { - t0 = t1; - t1 = 3 * in_near[i] + in_far[i]; - out[i * 2 - 1] = stbi__div16(3 * t0 + t1 + 8); - out[i * 2] = stbi__div16(3 * t1 + t0 + 8); - } - out[w * 2 - 1] = stbi__div4(t1 + 2); - - STBI_NOTUSED(hs); - - return out; -} - #endif - -static stbi_uc* stbi__resample_row_generic(stbi_uc* out, stbi_uc* in_near, stbi_uc* in_far, int w, int hs) -{ - // resample with nearest-neighbor - int i, j; - STBI_NOTUSED(in_far); - for (i = 0; i < w; ++i) - for (j = 0; j < hs; ++j) - out[i * hs + j] = in_near[i]; - return out; -} - - // this is a reduced-precision calculation of YCbCr-to-RGB introduced - // to make sure the code produces the same results in both SIMD and scalar - #define stbi__float2fixed(x) (((int)((x) * 4096.0f + 0.5f)) << 8) -static void stbi__YCbCr_to_RGB_row(stbi_uc* out, const stbi_uc* y, const stbi_uc* pcb, const stbi_uc* pcr, int count, int step) -{ - int i; - for (i = 0; i < count; ++i) - { - int y_fixed = (y[i] << 20) + (1 << 19); // rounding - int r, g, b; - int cr = pcr[i] - 128; - int cb = pcb[i] - 128; - r = y_fixed + cr * stbi__float2fixed(1.40200f); - g = y_fixed + (cr * -stbi__float2fixed(0.71414f)) + ((cb * -stbi__float2fixed(0.34414f)) & 0xffff0000); - b = y_fixed + cb * stbi__float2fixed(1.77200f); - r >>= 20; - g >>= 20; - b >>= 20; - if ((unsigned)r > 255) - { - if (r < 0) - r = 0; - else - r = 255; - } - if ((unsigned)g > 255) - { - if (g < 0) - g = 0; - else - g = 255; - } - if ((unsigned)b > 255) - { - if (b < 0) - b = 0; - else - b = 255; - } - out[0] = (stbi_uc)r; - out[1] = (stbi_uc)g; - out[2] = (stbi_uc)b; - out[3] = 255; - out += step; - } -} - - #if defined(STBI_SSE2) || defined(STBI_NEON) -static void stbi__YCbCr_to_RGB_simd(stbi_uc* out, const stbi_uc* y, const stbi_uc* pcb, const stbi_uc* pcr, int count, int step) -{ - int i = 0; - - #ifdef STBI_SSE2 - // step == 3 is pretty ugly on the final interleave, and i'm not convinced - // it's useful in practice (you wouldn't use it for textures, for example). - // so just accelerate step == 4 case. - if (step == 4) - { - // this is a fairly straightforward implementation and not super-optimized. - __m128i signflip = _mm_set1_epi8(-0x80); - __m128i cr_const0 = _mm_set1_epi16((short)(1.40200f * 4096.0f + 0.5f)); - __m128i cr_const1 = _mm_set1_epi16(-(short)(0.71414f * 4096.0f + 0.5f)); - __m128i cb_const0 = _mm_set1_epi16(-(short)(0.34414f * 4096.0f + 0.5f)); - __m128i cb_const1 = _mm_set1_epi16((short)(1.77200f * 4096.0f + 0.5f)); - __m128i y_bias = _mm_set1_epi8((char)(unsigned char)128); - __m128i xw = _mm_set1_epi16(255); // alpha channel - - for (; i + 7 < count; i += 8) - { - // load - __m128i y_bytes = _mm_loadl_epi64((__m128i*)(y + i)); - __m128i cr_bytes = _mm_loadl_epi64((__m128i*)(pcr + i)); - __m128i cb_bytes = _mm_loadl_epi64((__m128i*)(pcb + i)); - __m128i cr_biased = _mm_xor_si128(cr_bytes, signflip); // -128 - __m128i cb_biased = _mm_xor_si128(cb_bytes, signflip); // -128 - - // unpack to short (and left-shift cr, cb by 8) - __m128i yw = _mm_unpacklo_epi8(y_bias, y_bytes); - __m128i crw = _mm_unpacklo_epi8(_mm_setzero_si128(), cr_biased); - __m128i cbw = _mm_unpacklo_epi8(_mm_setzero_si128(), cb_biased); - - // color transform - __m128i yws = _mm_srli_epi16(yw, 4); - __m128i cr0 = _mm_mulhi_epi16(cr_const0, crw); - __m128i cb0 = _mm_mulhi_epi16(cb_const0, cbw); - __m128i cb1 = _mm_mulhi_epi16(cbw, cb_const1); - __m128i cr1 = _mm_mulhi_epi16(crw, cr_const1); - __m128i rws = _mm_add_epi16(cr0, yws); - __m128i gwt = _mm_add_epi16(cb0, yws); - __m128i bws = _mm_add_epi16(yws, cb1); - __m128i gws = _mm_add_epi16(gwt, cr1); - - // descale - __m128i rw = _mm_srai_epi16(rws, 4); - __m128i bw = _mm_srai_epi16(bws, 4); - __m128i gw = _mm_srai_epi16(gws, 4); - - // back to byte, set up for transpose - __m128i brb = _mm_packus_epi16(rw, bw); - __m128i gxb = _mm_packus_epi16(gw, xw); - - // transpose to interleave channels - __m128i t0 = _mm_unpacklo_epi8(brb, gxb); - __m128i t1 = _mm_unpackhi_epi8(brb, gxb); - __m128i o0 = _mm_unpacklo_epi16(t0, t1); - __m128i o1 = _mm_unpackhi_epi16(t0, t1); - - // store - _mm_storeu_si128((__m128i*)(out + 0), o0); - _mm_storeu_si128((__m128i*)(out + 16), o1); - out += 32; - } - } - #endif - - #ifdef STBI_NEON - // in this version, step=3 support would be easy to add. but is there demand? - if (step == 4) - { - // this is a fairly straightforward implementation and not super-optimized. - uint8x8_t signflip = vdup_n_u8(0x80); - int16x8_t cr_const0 = vdupq_n_s16((short)(1.40200f * 4096.0f + 0.5f)); - int16x8_t cr_const1 = vdupq_n_s16(-(short)(0.71414f * 4096.0f + 0.5f)); - int16x8_t cb_const0 = vdupq_n_s16(-(short)(0.34414f * 4096.0f + 0.5f)); - int16x8_t cb_const1 = vdupq_n_s16((short)(1.77200f * 4096.0f + 0.5f)); - - for (; i + 7 < count; i += 8) - { - // load - uint8x8_t y_bytes = vld1_u8(y + i); - uint8x8_t cr_bytes = vld1_u8(pcr + i); - uint8x8_t cb_bytes = vld1_u8(pcb + i); - int8x8_t cr_biased = vreinterpret_s8_u8(vsub_u8(cr_bytes, signflip)); - int8x8_t cb_biased = vreinterpret_s8_u8(vsub_u8(cb_bytes, signflip)); - - // expand to s16 - int16x8_t yws = vreinterpretq_s16_u16(vshll_n_u8(y_bytes, 4)); - int16x8_t crw = vshll_n_s8(cr_biased, 7); - int16x8_t cbw = vshll_n_s8(cb_biased, 7); - - // color transform - int16x8_t cr0 = vqdmulhq_s16(crw, cr_const0); - int16x8_t cb0 = vqdmulhq_s16(cbw, cb_const0); - int16x8_t cr1 = vqdmulhq_s16(crw, cr_const1); - int16x8_t cb1 = vqdmulhq_s16(cbw, cb_const1); - int16x8_t rws = vaddq_s16(yws, cr0); - int16x8_t gws = vaddq_s16(vaddq_s16(yws, cb0), cr1); - int16x8_t bws = vaddq_s16(yws, cb1); - - // undo scaling, round, convert to byte - uint8x8x4_t o; - o.val[0] = vqrshrun_n_s16(rws, 4); - o.val[1] = vqrshrun_n_s16(gws, 4); - o.val[2] = vqrshrun_n_s16(bws, 4); - o.val[3] = vdup_n_u8(255); - - // store, interleaving r/g/b/a - vst4_u8(out, o); - out += 8 * 4; - } - } - #endif - - for (; i < count; ++i) - { - int y_fixed = (y[i] << 20) + (1 << 19); // rounding - int r, g, b; - int cr = pcr[i] - 128; - int cb = pcb[i] - 128; - r = y_fixed + cr * stbi__float2fixed(1.40200f); - g = y_fixed + cr * -stbi__float2fixed(0.71414f) + ((cb * -stbi__float2fixed(0.34414f)) & 0xffff0000); - b = y_fixed + cb * stbi__float2fixed(1.77200f); - r >>= 20; - g >>= 20; - b >>= 20; - if ((unsigned)r > 255) - { - if (r < 0) - r = 0; - else - r = 255; - } - if ((unsigned)g > 255) - { - if (g < 0) - g = 0; - else - g = 255; - } - if ((unsigned)b > 255) - { - if (b < 0) - b = 0; - else - b = 255; - } - out[0] = (stbi_uc)r; - out[1] = (stbi_uc)g; - out[2] = (stbi_uc)b; - out[3] = 255; - out += step; - } -} - #endif - -// set up the kernels -static void stbi__setup_jpeg(stbi__jpeg* j) -{ - j->idct_block_kernel = stbi__idct_block; - j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_row; - j->resample_row_hv_2_kernel = stbi__resample_row_hv_2; - - #ifdef STBI_SSE2 - if (stbi__sse2_available()) - { - j->idct_block_kernel = stbi__idct_simd; - j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; - j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; - } - #endif - - #ifdef STBI_NEON - j->idct_block_kernel = stbi__idct_simd; - j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; - j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; - #endif -} - -// clean up the temporary component buffers -static void stbi__cleanup_jpeg(stbi__jpeg* j) -{ - stbi__free_jpeg_components(j, j->s->img_n, 0); -} - -typedef struct { - resample_row_func resample; - stbi_uc *line0, *line1; - int hs, vs; // expansion factor in each axis - int w_lores; // horizontal pixels pre-expansion - int ystep; // how far through vertical expansion we are - int ypos; // which pre-expansion row we're on -} stbi__resample; - -// fast 0..255 * 0..255 => 0..255 rounded multiplication -static stbi_uc stbi__blinn_8x8(stbi_uc x, stbi_uc y) -{ - unsigned int t = x * y + 128; - return (stbi_uc)((t + (t >> 8)) >> 8); -} - -static stbi_uc* load_jpeg_image(stbi__jpeg* z, int* out_x, int* out_y, int* comp, int req_comp) -{ - int n, decode_n, is_rgb; - z->s->img_n = 0; // make stbi__cleanup_jpeg safe - - // validate req_comp - if (req_comp < 0 || req_comp > 4) - return stbi__errpuc("bad req_comp", "Internal error"); - - // load a jpeg image from whichever source, but leave in YCbCr format - if (!stbi__decode_jpeg_image(z)) - { - stbi__cleanup_jpeg(z); - return NULL; - } - - // determine actual number of components to generate - n = req_comp ? req_comp : z->s->img_n >= 3 ? 3 : 1; - - is_rgb = z->s->img_n == 3 && (z->rgb == 3 || (z->app14_color_transform == 0 && !z->jfif)); - - if (z->s->img_n == 3 && n < 3 && !is_rgb) - decode_n = 1; - else - decode_n = z->s->img_n; - - // nothing to do if no components requested; check this now to avoid - // accessing uninitialized coutput[0] later - if (decode_n <= 0) - { - stbi__cleanup_jpeg(z); - return NULL; - } - - // resample and color-convert - { - int k; - unsigned int i, j; - stbi_uc* output; - stbi_uc* coutput[4] = {NULL, NULL, NULL, NULL}; - - stbi__resample res_comp[4]; - - for (k = 0; k < decode_n; ++k) - { - stbi__resample* r = &res_comp[k]; - - // allocate line buffer big enough for upsampling off the edges - // with upsample factor of 4 - z->img_comp[k].linebuf = (stbi_uc*)stbi__malloc(z->s->img_x + 3); - if (!z->img_comp[k].linebuf) - { - stbi__cleanup_jpeg(z); - return stbi__errpuc("outofmem", "Out of memory"); - } - - r->hs = z->img_h_max / z->img_comp[k].h; - r->vs = z->img_v_max / z->img_comp[k].v; - r->ystep = r->vs >> 1; - r->w_lores = (z->s->img_x + r->hs - 1) / r->hs; - r->ypos = 0; - r->line0 = r->line1 = z->img_comp[k].data; - - if (r->hs == 1 && r->vs == 1) - r->resample = resample_row_1; - else if (r->hs == 1 && r->vs == 2) - r->resample = stbi__resample_row_v_2; - else if (r->hs == 2 && r->vs == 1) - r->resample = stbi__resample_row_h_2; - else if (r->hs == 2 && r->vs == 2) - r->resample = z->resample_row_hv_2_kernel; - else - r->resample = stbi__resample_row_generic; - } - - // can't error after this so, this is safe - output = (stbi_uc*)stbi__malloc_mad3(n, z->s->img_x, z->s->img_y, 1); - if (!output) - { - stbi__cleanup_jpeg(z); - return stbi__errpuc("outofmem", "Out of memory"); - } - - // now go ahead and resample - for (j = 0; j < z->s->img_y; ++j) - { - stbi_uc* out = output + n * z->s->img_x * j; - for (k = 0; k < decode_n; ++k) - { - stbi__resample* r = &res_comp[k]; - int y_bot = r->ystep >= (r->vs >> 1); - coutput[k] = r->resample(z->img_comp[k].linebuf, y_bot ? r->line1 : r->line0, y_bot ? r->line0 : r->line1, r->w_lores, r->hs); - if (++r->ystep >= r->vs) - { - r->ystep = 0; - r->line0 = r->line1; - if (++r->ypos < z->img_comp[k].y) - r->line1 += z->img_comp[k].w2; - } - } - if (n >= 3) - { - stbi_uc* y = coutput[0]; - if (z->s->img_n == 3) - { - if (is_rgb) - { - for (i = 0; i < z->s->img_x; ++i) - { - out[0] = y[i]; - out[1] = coutput[1][i]; - out[2] = coutput[2][i]; - out[3] = 255; - out += n; - } - } - else - { - z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); - } - } - else if (z->s->img_n == 4) - { - if (z->app14_color_transform == 0) - { // CMYK - for (i = 0; i < z->s->img_x; ++i) - { - stbi_uc m = coutput[3][i]; - out[0] = stbi__blinn_8x8(coutput[0][i], m); - out[1] = stbi__blinn_8x8(coutput[1][i], m); - out[2] = stbi__blinn_8x8(coutput[2][i], m); - out[3] = 255; - out += n; - } - } - else if (z->app14_color_transform == 2) - { // YCCK - z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); - for (i = 0; i < z->s->img_x; ++i) - { - stbi_uc m = coutput[3][i]; - out[0] = stbi__blinn_8x8(255 - out[0], m); - out[1] = stbi__blinn_8x8(255 - out[1], m); - out[2] = stbi__blinn_8x8(255 - out[2], m); - out += n; - } - } - else - { // YCbCr + alpha? Ignore the fourth channel for now - z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); - } - } - else - for (i = 0; i < z->s->img_x; ++i) - { - out[0] = out[1] = out[2] = y[i]; - out[3] = 255; // not used if n==3 - out += n; - } - } - else - { - if (is_rgb) - { - if (n == 1) - for (i = 0; i < z->s->img_x; ++i) - *out++ = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]); - else - { - for (i = 0; i < z->s->img_x; ++i, out += 2) - { - out[0] = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]); - out[1] = 255; - } - } - } - else if (z->s->img_n == 4 && z->app14_color_transform == 0) - { - for (i = 0; i < z->s->img_x; ++i) - { - stbi_uc m = coutput[3][i]; - stbi_uc r = stbi__blinn_8x8(coutput[0][i], m); - stbi_uc g = stbi__blinn_8x8(coutput[1][i], m); - stbi_uc b = stbi__blinn_8x8(coutput[2][i], m); - out[0] = stbi__compute_y(r, g, b); - out[1] = 255; - out += n; - } - } - else if (z->s->img_n == 4 && z->app14_color_transform == 2) - { - for (i = 0; i < z->s->img_x; ++i) - { - out[0] = stbi__blinn_8x8(255 - coutput[0][i], coutput[3][i]); - out[1] = 255; - out += n; - } - } - else - { - stbi_uc* y = coutput[0]; - if (n == 1) - for (i = 0; i < z->s->img_x; ++i) - out[i] = y[i]; - else - for (i = 0; i < z->s->img_x; ++i) - { - *out++ = y[i]; - *out++ = 255; - } - } - } - } - stbi__cleanup_jpeg(z); - *out_x = z->s->img_x; - *out_y = z->s->img_y; - if (comp) - *comp = z->s->img_n >= 3 ? 3 : 1; // report original components, not output - return output; - } -} - -static void* stbi__jpeg_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) -{ - unsigned char* result; - stbi__jpeg* j = (stbi__jpeg*)stbi__malloc(sizeof(stbi__jpeg)); - if (!j) - return stbi__errpuc("outofmem", "Out of memory"); - memset(j, 0, sizeof(stbi__jpeg)); - STBI_NOTUSED(ri); - j->s = s; - stbi__setup_jpeg(j); - result = load_jpeg_image(j, x, y, comp, req_comp); - STBI_FREE(j); - return result; -} - -static int stbi__jpeg_test(stbi__context* s) -{ - int r; - stbi__jpeg* j = (stbi__jpeg*)stbi__malloc(sizeof(stbi__jpeg)); - if (!j) - return stbi__err("outofmem", "Out of memory"); - memset(j, 0, sizeof(stbi__jpeg)); - j->s = s; - stbi__setup_jpeg(j); - r = stbi__decode_jpeg_header(j, STBI__SCAN_type); - stbi__rewind(s); - STBI_FREE(j); - return r; -} - -static int stbi__jpeg_info_raw(stbi__jpeg* j, int* x, int* y, int* comp) -{ - if (!stbi__decode_jpeg_header(j, STBI__SCAN_header)) - { - stbi__rewind(j->s); - return 0; - } - if (x) - *x = j->s->img_x; - if (y) - *y = j->s->img_y; - if (comp) - *comp = j->s->img_n >= 3 ? 3 : 1; - return 1; -} - -static int stbi__jpeg_info(stbi__context* s, int* x, int* y, int* comp) -{ - int result; - stbi__jpeg* j = (stbi__jpeg*)(stbi__malloc(sizeof(stbi__jpeg))); - if (!j) - return stbi__err("outofmem", "Out of memory"); - memset(j, 0, sizeof(stbi__jpeg)); - j->s = s; - result = stbi__jpeg_info_raw(j, x, y, comp); - STBI_FREE(j); - return result; -} - #endif - -// public domain zlib decode v0.2 Sean Barrett 2006-11-18 -// simple implementation -// - all input must be provided in an upfront buffer -// - all output is written to a single output buffer (can malloc/realloc) -// performance -// - fast huffman - - #ifndef STBI_NO_ZLIB - - // fast-way is faster to check than jpeg huffman, but slow way is slower - #define STBI__ZFAST_BITS 9 // accelerate all cases in default tables - #define STBI__ZFAST_MASK ((1 << STBI__ZFAST_BITS) - 1) - #define STBI__ZNSYMS 288 // number of symbols in literal/length alphabet - -// zlib-style huffman encoding -// (jpegs packs from left, zlib from right, so can't share code) -typedef struct { - stbi__uint16 fast[1 << STBI__ZFAST_BITS]; - stbi__uint16 firstcode[16]; - int maxcode[17]; - stbi__uint16 firstsymbol[16]; - stbi_uc size[STBI__ZNSYMS]; - stbi__uint16 value[STBI__ZNSYMS]; -} stbi__zhuffman; - -stbi_inline static int stbi__bitreverse16(int n) -{ - n = ((n & 0xAAAA) >> 1) | ((n & 0x5555) << 1); - n = ((n & 0xCCCC) >> 2) | ((n & 0x3333) << 2); - n = ((n & 0xF0F0) >> 4) | ((n & 0x0F0F) << 4); - n = ((n & 0xFF00) >> 8) | ((n & 0x00FF) << 8); - return n; -} - -stbi_inline static int stbi__bit_reverse(int v, int bits) -{ - STBI_ASSERT(bits <= 16); - // to bit reverse n bits, reverse 16 and shift - // e.g. 11 bits, bit reverse and shift away 5 - return stbi__bitreverse16(v) >> (16 - bits); -} - -static int stbi__zbuild_huffman(stbi__zhuffman* z, const stbi_uc* sizelist, int num) -{ - int i, k = 0; - int code, next_code[16], sizes[17]; - - // DEFLATE spec for generating codes - memset(sizes, 0, sizeof(sizes)); - memset(z->fast, 0, sizeof(z->fast)); - for (i = 0; i < num; ++i) - ++sizes[sizelist[i]]; - sizes[0] = 0; - for (i = 1; i < 16; ++i) - if (sizes[i] > (1 << i)) - return stbi__err("bad sizes", "Corrupt PNG"); - code = 0; - for (i = 1; i < 16; ++i) - { - next_code[i] = code; - z->firstcode[i] = (stbi__uint16)code; - z->firstsymbol[i] = (stbi__uint16)k; - code = (code + sizes[i]); - if (sizes[i]) - if (code - 1 >= (1 << i)) - return stbi__err("bad codelengths", "Corrupt PNG"); - z->maxcode[i] = code << (16 - i); // preshift for inner loop - code <<= 1; - k += sizes[i]; - } - z->maxcode[16] = 0x10000; // sentinel - for (i = 0; i < num; ++i) - { - int s = sizelist[i]; - if (s) - { - int c = next_code[s] - z->firstcode[s] + z->firstsymbol[s]; - stbi__uint16 fastv = (stbi__uint16)((s << 9) | i); - z->size[c] = (stbi_uc)s; - z->value[c] = (stbi__uint16)i; - if (s <= STBI__ZFAST_BITS) - { - int j = stbi__bit_reverse(next_code[s], s); - while (j < (1 << STBI__ZFAST_BITS)) - { - z->fast[j] = fastv; - j += (1 << s); - } - } - ++next_code[s]; - } - } - return 1; -} - -// zlib-from-memory implementation for PNG reading -// because PNG allows splitting the zlib stream arbitrarily, -// and it's annoying structurally to have PNG call ZLIB call PNG, -// we require PNG read all the IDATs and combine them into a single -// memory buffer - -typedef struct { - stbi_uc *zbuffer, *zbuffer_end; - int num_bits; - int hit_zeof_once; - stbi__uint32 code_buffer; - - char* zout; - char* zout_start; - char* zout_end; - int z_expandable; - - stbi__zhuffman z_length, z_distance; -} stbi__zbuf; - -stbi_inline static int stbi__zeof(stbi__zbuf* z) -{ - return (z->zbuffer >= z->zbuffer_end); -} - -stbi_inline static stbi_uc stbi__zget8(stbi__zbuf* z) -{ - return stbi__zeof(z) ? 0 : *z->zbuffer++; -} - -static void stbi__fill_bits(stbi__zbuf* z) -{ - do - { - if (z->code_buffer >= (1U << z->num_bits)) - { - z->zbuffer = z->zbuffer_end; /* treat this as EOF so we fail. */ - return; - } - z->code_buffer |= (unsigned int)stbi__zget8(z) << z->num_bits; - z->num_bits += 8; - } while (z->num_bits <= 24); -} - -stbi_inline static unsigned int stbi__zreceive(stbi__zbuf* z, int n) -{ - unsigned int k; - if (z->num_bits < n) - stbi__fill_bits(z); - k = z->code_buffer & ((1 << n) - 1); - z->code_buffer >>= n; - z->num_bits -= n; - return k; -} - -static int stbi__zhuffman_decode_slowpath(stbi__zbuf* a, stbi__zhuffman* z) -{ - int b, s, k; - // not resolved by fast table, so compute it the slow way - // use jpeg approach, which requires MSbits at top - k = stbi__bit_reverse(a->code_buffer, 16); - for (s = STBI__ZFAST_BITS + 1;; ++s) - if (k < z->maxcode[s]) - break; - if (s >= 16) - return -1; // invalid code! - // code size is s, so: - b = (k >> (16 - s)) - z->firstcode[s] + z->firstsymbol[s]; - if (b >= STBI__ZNSYMS) - return -1; // some data was corrupt somewhere! - if (z->size[b] != s) - return -1; // was originally an assert, but report failure instead. - a->code_buffer >>= s; - a->num_bits -= s; - return z->value[b]; -} - -stbi_inline static int stbi__zhuffman_decode(stbi__zbuf* a, stbi__zhuffman* z) -{ - int b, s; - if (a->num_bits < 16) - { - if (stbi__zeof(a)) - { - if (!a->hit_zeof_once) - { - // This is the first time we hit eof, insert 16 extra padding btis - // to allow us to keep going; if we actually consume any of them - // though, that is invalid data. This is caught later. - a->hit_zeof_once = 1; - a->num_bits += 16; // add 16 implicit zero bits - } - else - { - // We already inserted our extra 16 padding bits and are again - // out, this stream is actually prematurely terminated. - return -1; - } - } - else - { - stbi__fill_bits(a); - } - } - b = z->fast[a->code_buffer & STBI__ZFAST_MASK]; - if (b) - { - s = b >> 9; - a->code_buffer >>= s; - a->num_bits -= s; - return b & 511; - } - return stbi__zhuffman_decode_slowpath(a, z); -} - -static int stbi__zexpand(stbi__zbuf* z, char* zout, int n) // need to make room for n bytes -{ - char* q; - unsigned int cur, limit, old_limit; - z->zout = zout; - if (!z->z_expandable) - return stbi__err("output buffer limit", "Corrupt PNG"); - cur = (unsigned int)(z->zout - z->zout_start); - limit = old_limit = (unsigned)(z->zout_end - z->zout_start); - if (UINT_MAX - cur < (unsigned)n) - return stbi__err("outofmem", "Out of memory"); - while (cur + n > limit) - { - if (limit > UINT_MAX / 2) - return stbi__err("outofmem", "Out of memory"); - limit *= 2; - } - q = (char*)STBI_REALLOC_SIZED(z->zout_start, old_limit, limit); - STBI_NOTUSED(old_limit); - if (q == NULL) - return stbi__err("outofmem", "Out of memory"); - z->zout_start = q; - z->zout = q + cur; - z->zout_end = q + limit; - return 1; -} - -static const int stbi__zlength_base[31] = {3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31, 35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, - 227, 258, 0, 0}; - -static const int stbi__zlength_extra[31] = {0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 0, 0}; - -static const int stbi__zdist_base[32] = {1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193, 257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, - 6145, 8193, 12289, 16385, 24577, 0, 0}; - -static const int stbi__zdist_extra[32] = {0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13}; - -static int stbi__parse_huffman_block(stbi__zbuf* a) -{ - char* zout = a->zout; - for (;;) - { - int z = stbi__zhuffman_decode(a, &a->z_length); - if (z < 256) - { - if (z < 0) - return stbi__err("bad huffman code", "Corrupt PNG"); // error in huffman codes - if (zout >= a->zout_end) - { - if (!stbi__zexpand(a, zout, 1)) - return 0; - zout = a->zout; - } - *zout++ = (char)z; - } - else - { - stbi_uc* p; - int len, dist; - if (z == 256) - { - a->zout = zout; - if (a->hit_zeof_once && a->num_bits < 16) - { - // The first time we hit zeof, we inserted 16 extra zero bits into our bit - // buffer so the decoder can just do its speculative decoding. But if we - // actually consumed any of those bits (which is the case when num_bits < 16), - // the stream actually read past the end so it is malformed. - return stbi__err("unexpected end", "Corrupt PNG"); - } - return 1; - } - if (z >= 286) - return stbi__err("bad huffman code", "Corrupt PNG"); // per DEFLATE, length codes 286 and 287 must not appear in compressed data - z -= 257; - len = stbi__zlength_base[z]; - if (stbi__zlength_extra[z]) - len += stbi__zreceive(a, stbi__zlength_extra[z]); - z = stbi__zhuffman_decode(a, &a->z_distance); - if (z < 0 || z >= 30) - return stbi__err("bad huffman code", "Corrupt PNG"); // per DEFLATE, distance codes 30 and 31 must not appear in compressed data - dist = stbi__zdist_base[z]; - if (stbi__zdist_extra[z]) - dist += stbi__zreceive(a, stbi__zdist_extra[z]); - if (zout - a->zout_start < dist) - return stbi__err("bad dist", "Corrupt PNG"); - if (len > a->zout_end - zout) - { - if (!stbi__zexpand(a, zout, len)) - return 0; - zout = a->zout; - } - p = (stbi_uc*)(zout - dist); - if (dist == 1) - { // run of one byte; common in images. - stbi_uc v = *p; - if (len) - { - do - *zout++ = v; - while (--len); - } - } - else - { - if (len) - { - do - *zout++ = *p++; - while (--len); - } - } - } - } -} - -static int stbi__compute_huffman_codes(stbi__zbuf* a) -{ - static const stbi_uc length_dezigzag[19] = {16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15}; - stbi__zhuffman z_codelength; - stbi_uc lencodes[286 + 32 + 137]; // padding for maximum single op - stbi_uc codelength_sizes[19]; - int i, n; - - int hlit = stbi__zreceive(a, 5) + 257; - int hdist = stbi__zreceive(a, 5) + 1; - int hclen = stbi__zreceive(a, 4) + 4; - int ntot = hlit + hdist; - - memset(codelength_sizes, 0, sizeof(codelength_sizes)); - for (i = 0; i < hclen; ++i) - { - int s = stbi__zreceive(a, 3); - codelength_sizes[length_dezigzag[i]] = (stbi_uc)s; - } - if (!stbi__zbuild_huffman(&z_codelength, codelength_sizes, 19)) - return 0; - - n = 0; - while (n < ntot) - { - int c = stbi__zhuffman_decode(a, &z_codelength); - if (c < 0 || c >= 19) - return stbi__err("bad codelengths", "Corrupt PNG"); - if (c < 16) - lencodes[n++] = (stbi_uc)c; - else - { - stbi_uc fill = 0; - if (c == 16) - { - c = stbi__zreceive(a, 2) + 3; - if (n == 0) - return stbi__err("bad codelengths", "Corrupt PNG"); - fill = lencodes[n - 1]; - } - else if (c == 17) - { - c = stbi__zreceive(a, 3) + 3; - } - else if (c == 18) - { - c = stbi__zreceive(a, 7) + 11; - } - else - { - return stbi__err("bad codelengths", "Corrupt PNG"); - } - if (ntot - n < c) - return stbi__err("bad codelengths", "Corrupt PNG"); - memset(lencodes + n, fill, c); - n += c; - } - } - if (n != ntot) - return stbi__err("bad codelengths", "Corrupt PNG"); - if (!stbi__zbuild_huffman(&a->z_length, lencodes, hlit)) - return 0; - if (!stbi__zbuild_huffman(&a->z_distance, lencodes + hlit, hdist)) - return 0; - return 1; -} - -static int stbi__parse_uncompressed_block(stbi__zbuf* a) -{ - stbi_uc header[4]; - int len, nlen, k; - if (a->num_bits & 7) - stbi__zreceive(a, a->num_bits & 7); // discard - // drain the bit-packed data into header - k = 0; - while (a->num_bits > 0) - { - header[k++] = (stbi_uc)(a->code_buffer & 255); // suppress MSVC run-time check - a->code_buffer >>= 8; - a->num_bits -= 8; - } - if (a->num_bits < 0) - return stbi__err("zlib corrupt", "Corrupt PNG"); - // now fill header the normal way - while (k < 4) - header[k++] = stbi__zget8(a); - len = header[1] * 256 + header[0]; - nlen = header[3] * 256 + header[2]; - if (nlen != (len ^ 0xffff)) - return stbi__err("zlib corrupt", "Corrupt PNG"); - if (a->zbuffer + len > a->zbuffer_end) - return stbi__err("read past buffer", "Corrupt PNG"); - if (a->zout + len > a->zout_end) - if (!stbi__zexpand(a, a->zout, len)) - return 0; - memcpy(a->zout, a->zbuffer, len); - a->zbuffer += len; - a->zout += len; - return 1; -} - -static int stbi__parse_zlib_header(stbi__zbuf* a) -{ - int cmf = stbi__zget8(a); - int cm = cmf & 15; - /* int cinfo = cmf >> 4; */ - int flg = stbi__zget8(a); - if (stbi__zeof(a)) - return stbi__err("bad zlib header", "Corrupt PNG"); // zlib spec - if ((cmf * 256 + flg) % 31 != 0) - return stbi__err("bad zlib header", "Corrupt PNG"); // zlib spec - if (flg & 32) - return stbi__err("no preset dict", "Corrupt PNG"); // preset dictionary not allowed in png - if (cm != 8) - return stbi__err("bad compression", "Corrupt PNG"); // DEFLATE required for png - // window = 1 << (8 + cinfo)... but who cares, we fully buffer output - return 1; -} - -static const stbi_uc stbi__zdefault_length[STBI__ZNSYMS] = {8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, - 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, - 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, - 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, - 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, - 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, - 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8}; -static const stbi_uc stbi__zdefault_distance[32] = {5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5}; -/* -Init algorithm: -{ - int i; // use <= to match clearly with spec - for (i=0; i <= 143; ++i) stbi__zdefault_length[i] = 8; - for ( ; i <= 255; ++i) stbi__zdefault_length[i] = 9; - for ( ; i <= 279; ++i) stbi__zdefault_length[i] = 7; - for ( ; i <= 287; ++i) stbi__zdefault_length[i] = 8; - - for (i=0; i <= 31; ++i) stbi__zdefault_distance[i] = 5; -} -*/ - -static int stbi__parse_zlib(stbi__zbuf* a, int parse_header) -{ - int final, type; - if (parse_header) - if (!stbi__parse_zlib_header(a)) - return 0; - a->num_bits = 0; - a->code_buffer = 0; - a->hit_zeof_once = 0; - do - { - final = stbi__zreceive(a, 1); - type = stbi__zreceive(a, 2); - if (type == 0) - { - if (!stbi__parse_uncompressed_block(a)) - return 0; - } - else if (type == 3) - { - return 0; - } - else - { - if (type == 1) - { - // use fixed code lengths - if (!stbi__zbuild_huffman(&a->z_length, stbi__zdefault_length, STBI__ZNSYMS)) - return 0; - if (!stbi__zbuild_huffman(&a->z_distance, stbi__zdefault_distance, 32)) - return 0; - } - else - { - if (!stbi__compute_huffman_codes(a)) - return 0; - } - if (!stbi__parse_huffman_block(a)) - return 0; - } - } while (!final); - return 1; -} - -static int stbi__do_zlib(stbi__zbuf* a, char* obuf, int olen, int exp, int parse_header) -{ - a->zout_start = obuf; - a->zout = obuf; - a->zout_end = obuf + olen; - a->z_expandable = exp; - - return stbi__parse_zlib(a, parse_header); -} - -STBIDEF char* stbi_zlib_decode_malloc_guesssize(const char* buffer, int len, int initial_size, int* outlen) -{ - stbi__zbuf a; - char* p = (char*)stbi__malloc(initial_size); - if (p == NULL) - return NULL; - a.zbuffer = (stbi_uc*)buffer; - a.zbuffer_end = (stbi_uc*)buffer + len; - if (stbi__do_zlib(&a, p, initial_size, 1, 1)) - { - if (outlen) - *outlen = (int)(a.zout - a.zout_start); - return a.zout_start; - } - else - { - STBI_FREE(a.zout_start); - return NULL; - } -} - -STBIDEF char* stbi_zlib_decode_malloc(const char* buffer, int len, int* outlen) -{ - return stbi_zlib_decode_malloc_guesssize(buffer, len, 16384, outlen); -} - -STBIDEF char* stbi_zlib_decode_malloc_guesssize_headerflag(const char* buffer, int len, int initial_size, int* outlen, int parse_header) -{ - stbi__zbuf a; - char* p = (char*)stbi__malloc(initial_size); - if (p == NULL) - return NULL; - a.zbuffer = (stbi_uc*)buffer; - a.zbuffer_end = (stbi_uc*)buffer + len; - if (stbi__do_zlib(&a, p, initial_size, 1, parse_header)) - { - if (outlen) - *outlen = (int)(a.zout - a.zout_start); - return a.zout_start; - } - else - { - STBI_FREE(a.zout_start); - return NULL; - } -} - -STBIDEF int stbi_zlib_decode_buffer(char* obuffer, int olen, const char* ibuffer, int ilen) -{ - stbi__zbuf a; - a.zbuffer = (stbi_uc*)ibuffer; - a.zbuffer_end = (stbi_uc*)ibuffer + ilen; - if (stbi__do_zlib(&a, obuffer, olen, 0, 1)) - return (int)(a.zout - a.zout_start); - else - return -1; -} - -STBIDEF char* stbi_zlib_decode_noheader_malloc(const char* buffer, int len, int* outlen) -{ - stbi__zbuf a; - char* p = (char*)stbi__malloc(16384); - if (p == NULL) - return NULL; - a.zbuffer = (stbi_uc*)buffer; - a.zbuffer_end = (stbi_uc*)buffer + len; - if (stbi__do_zlib(&a, p, 16384, 1, 0)) - { - if (outlen) - *outlen = (int)(a.zout - a.zout_start); - return a.zout_start; - } - else - { - STBI_FREE(a.zout_start); - return NULL; - } -} - -STBIDEF int stbi_zlib_decode_noheader_buffer(char* obuffer, int olen, const char* ibuffer, int ilen) -{ - stbi__zbuf a; - a.zbuffer = (stbi_uc*)ibuffer; - a.zbuffer_end = (stbi_uc*)ibuffer + ilen; - if (stbi__do_zlib(&a, obuffer, olen, 0, 0)) - return (int)(a.zout - a.zout_start); - else - return -1; -} - #endif - -// public domain "baseline" PNG decoder v0.10 Sean Barrett 2006-11-18 -// simple implementation -// - only 8-bit samples -// - no CRC checking -// - allocates lots of intermediate memory -// - avoids problem of streaming data between subsystems -// - avoids explicit window management -// performance -// - uses stb_zlib, a PD zlib implementation with fast huffman decoding - - #ifndef STBI_NO_PNG -typedef struct { - stbi__uint32 length; - stbi__uint32 type; -} stbi__pngchunk; - -static stbi__pngchunk stbi__get_chunk_header(stbi__context* s) -{ - stbi__pngchunk c; - c.length = stbi__get32be(s); - c.type = stbi__get32be(s); - return c; -} - -static int stbi__check_png_header(stbi__context* s) -{ - static const stbi_uc png_sig[8] = {137, 80, 78, 71, 13, 10, 26, 10}; - int i; - for (i = 0; i < 8; ++i) - if (stbi__get8(s) != png_sig[i]) - return stbi__err("bad png sig", "Not a PNG"); - return 1; -} - -typedef struct { - stbi__context* s; - stbi_uc *idata, *expanded, *out; - int depth; -} stbi__png; - -enum { - STBI__F_none = 0, - STBI__F_sub = 1, - STBI__F_up = 2, - STBI__F_avg = 3, - STBI__F_paeth = 4, - // synthetic filter used for first scanline to avoid needing a dummy row of 0s - STBI__F_avg_first -}; - -static stbi_uc first_row_filter[5] = { - STBI__F_none, STBI__F_sub, STBI__F_none, STBI__F_avg_first, - STBI__F_sub // Paeth with b=c=0 turns out to be equivalent to sub -}; - -static int stbi__paeth(int a, int b, int c) -{ - // This formulation looks very different from the reference in the PNG spec, but is - // actually equivalent and has favorable data dependencies and admits straightforward - // generation of branch-free code, which helps performance significantly. - int thresh = c * 3 - (a + b); - int lo = a < b ? a : b; - int hi = a < b ? b : a; - int t0 = (hi <= thresh) ? lo : c; - int t1 = (thresh <= lo) ? hi : t0; - return t1; -} - -static const stbi_uc stbi__depth_scale_table[9] = {0, 0xff, 0x55, 0, 0x11, 0, 0, 0, 0x01}; - -// adds an extra all-255 alpha channel -// dest == src is legal -// img_n must be 1 or 3 -static void stbi__create_png_alpha_expand8(stbi_uc* dest, stbi_uc* src, stbi__uint32 x, int img_n) -{ - int i; - // must process data backwards since we allow dest==src - if (img_n == 1) - { - for (i = x - 1; i >= 0; --i) - { - dest[i * 2 + 1] = 255; - dest[i * 2 + 0] = src[i]; - } - } - else - { - STBI_ASSERT(img_n == 3); - for (i = x - 1; i >= 0; --i) - { - dest[i * 4 + 3] = 255; - dest[i * 4 + 2] = src[i * 3 + 2]; - dest[i * 4 + 1] = src[i * 3 + 1]; - dest[i * 4 + 0] = src[i * 3 + 0]; - } - } -} - -// create the png data from post-deflated data -static int stbi__create_png_image_raw(stbi__png* a, stbi_uc* raw, stbi__uint32 raw_len, int out_n, stbi__uint32 x, stbi__uint32 y, int depth, - int color) -{ - int bytes = (depth == 16 ? 2 : 1); - stbi__context* s = a->s; - stbi__uint32 i, j, stride = x * out_n * bytes; - stbi__uint32 img_len, img_width_bytes; - stbi_uc* filter_buf; - int all_ok = 1; - int k; - int img_n = s->img_n; // copy it into a local for later - - int output_bytes = out_n * bytes; - int filter_bytes = img_n * bytes; - int width = x; - - STBI_ASSERT(out_n == s->img_n || out_n == s->img_n + 1); - a->out = (stbi_uc*)stbi__malloc_mad3(x, y, output_bytes, 0); // extra bytes to write off the end into - if (!a->out) - return stbi__err("outofmem", "Out of memory"); - - // note: error exits here don't need to clean up a->out individually, - // stbi__do_png always does on error. - if (!stbi__mad3sizes_valid(img_n, x, depth, 7)) - return stbi__err("too large", "Corrupt PNG"); - img_width_bytes = (((img_n * x * depth) + 7) >> 3); - if (!stbi__mad2sizes_valid(img_width_bytes, y, img_width_bytes)) - return stbi__err("too large", "Corrupt PNG"); - img_len = (img_width_bytes + 1) * y; - - // we used to check for exact match between raw_len and img_len on non-interlaced PNGs, - // but issue #276 reported a PNG in the wild that had extra data at the end (all zeros), - // so just check for raw_len < img_len always. - if (raw_len < img_len) - return stbi__err("not enough pixels", "Corrupt PNG"); - - // Allocate two scan lines worth of filter workspace buffer. - filter_buf = (stbi_uc*)stbi__malloc_mad2(img_width_bytes, 2, 0); - if (!filter_buf) - return stbi__err("outofmem", "Out of memory"); - - // Filtering for low-bit-depth images - if (depth < 8) - { - filter_bytes = 1; - width = img_width_bytes; - } - - for (j = 0; j < y; ++j) - { - // cur/prior filter buffers alternate - stbi_uc* cur = filter_buf + (j & 1) * img_width_bytes; - stbi_uc* prior = filter_buf + (~j & 1) * img_width_bytes; - stbi_uc* dest = a->out + stride * j; - int nk = width * filter_bytes; - int filter = *raw++; - - // check filter type - if (filter > 4) - { - all_ok = stbi__err("invalid filter", "Corrupt PNG"); - break; - } - - // if first row, use special filter that doesn't sample previous row - if (j == 0) - filter = first_row_filter[filter]; - - // perform actual filtering - switch (filter) - { - case STBI__F_none: memcpy(cur, raw, nk); break; - case STBI__F_sub: - memcpy(cur, raw, filter_bytes); - for (k = filter_bytes; k < nk; ++k) - cur[k] = STBI__BYTECAST(raw[k] + cur[k - filter_bytes]); - break; - case STBI__F_up: - for (k = 0; k < nk; ++k) - cur[k] = STBI__BYTECAST(raw[k] + prior[k]); - break; - case STBI__F_avg: - for (k = 0; k < filter_bytes; ++k) - cur[k] = STBI__BYTECAST(raw[k] + (prior[k] >> 1)); - for (k = filter_bytes; k < nk; ++k) - cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k - filter_bytes]) >> 1)); - break; - case STBI__F_paeth: - for (k = 0; k < filter_bytes; ++k) - cur[k] = STBI__BYTECAST(raw[k] + prior[k]); // prior[k] == stbi__paeth(0,prior[k],0) - for (k = filter_bytes; k < nk; ++k) - cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k - filter_bytes], prior[k], prior[k - filter_bytes])); - break; - case STBI__F_avg_first: - memcpy(cur, raw, filter_bytes); - for (k = filter_bytes; k < nk; ++k) - cur[k] = STBI__BYTECAST(raw[k] + (cur[k - filter_bytes] >> 1)); - break; - } - - raw += nk; - - // expand decoded bits in cur to dest, also adding an extra alpha channel if desired - if (depth < 8) - { - stbi_uc scale = (color == 0) ? stbi__depth_scale_table[depth] : 1; // scale grayscale values to 0..255 range - stbi_uc* in = cur; - stbi_uc* out = dest; - stbi_uc inb = 0; - stbi__uint32 nsmp = x * img_n; - - // expand bits to bytes first - if (depth == 4) - { - for (i = 0; i < nsmp; ++i) - { - if ((i & 1) == 0) - inb = *in++; - *out++ = scale * (inb >> 4); - inb <<= 4; - } - } - else if (depth == 2) - { - for (i = 0; i < nsmp; ++i) - { - if ((i & 3) == 0) - inb = *in++; - *out++ = scale * (inb >> 6); - inb <<= 2; - } - } - else - { - STBI_ASSERT(depth == 1); - for (i = 0; i < nsmp; ++i) - { - if ((i & 7) == 0) - inb = *in++; - *out++ = scale * (inb >> 7); - inb <<= 1; - } - } - - // insert alpha=255 values if desired - if (img_n != out_n) - stbi__create_png_alpha_expand8(dest, dest, x, img_n); - } - else if (depth == 8) - { - if (img_n == out_n) - memcpy(dest, cur, x * img_n); - else - stbi__create_png_alpha_expand8(dest, cur, x, img_n); - } - else if (depth == 16) - { - // convert the image data from big-endian to platform-native - stbi__uint16* dest16 = (stbi__uint16*)dest; - stbi__uint32 nsmp = x * img_n; - - if (img_n == out_n) - { - for (i = 0; i < nsmp; ++i, ++dest16, cur += 2) - *dest16 = (cur[0] << 8) | cur[1]; - } - else - { - STBI_ASSERT(img_n + 1 == out_n); - if (img_n == 1) - { - for (i = 0; i < x; ++i, dest16 += 2, cur += 2) - { - dest16[0] = (cur[0] << 8) | cur[1]; - dest16[1] = 0xffff; - } - } - else - { - STBI_ASSERT(img_n == 3); - for (i = 0; i < x; ++i, dest16 += 4, cur += 6) - { - dest16[0] = (cur[0] << 8) | cur[1]; - dest16[1] = (cur[2] << 8) | cur[3]; - dest16[2] = (cur[4] << 8) | cur[5]; - dest16[3] = 0xffff; - } - } - } - } - } - - STBI_FREE(filter_buf); - if (!all_ok) - return 0; - - return 1; -} - -static int stbi__create_png_image(stbi__png* a, stbi_uc* image_data, stbi__uint32 image_data_len, int out_n, int depth, int color, int interlaced) -{ - int bytes = (depth == 16 ? 2 : 1); - int out_bytes = out_n * bytes; - stbi_uc* final; - int p; - if (!interlaced) - return stbi__create_png_image_raw(a, image_data, image_data_len, out_n, a->s->img_x, a->s->img_y, depth, color); - - // de-interlacing - final = (stbi_uc*)stbi__malloc_mad3(a->s->img_x, a->s->img_y, out_bytes, 0); - if (!final) - return stbi__err("outofmem", "Out of memory"); - for (p = 0; p < 7; ++p) - { - int xorig[] = {0, 4, 0, 2, 0, 1, 0}; - int yorig[] = {0, 0, 4, 0, 2, 0, 1}; - int xspc[] = {8, 8, 4, 4, 2, 2, 1}; - int yspc[] = {8, 8, 8, 4, 4, 2, 2}; - int i, j, x, y; - // pass1_x[4] = 0, pass1_x[5] = 1, pass1_x[12] = 1 - x = (a->s->img_x - xorig[p] + xspc[p] - 1) / xspc[p]; - y = (a->s->img_y - yorig[p] + yspc[p] - 1) / yspc[p]; - if (x && y) - { - stbi__uint32 img_len = ((((a->s->img_n * x * depth) + 7) >> 3) + 1) * y; - if (!stbi__create_png_image_raw(a, image_data, image_data_len, out_n, x, y, depth, color)) - { - STBI_FREE(final); - return 0; - } - for (j = 0; j < y; ++j) - { - for (i = 0; i < x; ++i) - { - int out_y = j * yspc[p] + yorig[p]; - int out_x = i * xspc[p] + xorig[p]; - memcpy(final + out_y * a->s->img_x * out_bytes + out_x * out_bytes, a->out + (j * x + i) * out_bytes, out_bytes); - } - } - STBI_FREE(a->out); - image_data += img_len; - image_data_len -= img_len; - } - } - a->out = final; - - return 1; -} - -static int stbi__compute_transparency(stbi__png* z, stbi_uc tc[3], int out_n) -{ - stbi__context* s = z->s; - stbi__uint32 i, pixel_count = s->img_x * s->img_y; - stbi_uc* p = z->out; - - // compute color-based transparency, assuming we've - // already got 255 as the alpha value in the output - STBI_ASSERT(out_n == 2 || out_n == 4); - - if (out_n == 2) - { - for (i = 0; i < pixel_count; ++i) - { - p[1] = (p[0] == tc[0] ? 0 : 255); - p += 2; - } - } - else - { - for (i = 0; i < pixel_count; ++i) - { - if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2]) - p[3] = 0; - p += 4; - } - } - return 1; -} - -static int stbi__compute_transparency16(stbi__png* z, stbi__uint16 tc[3], int out_n) -{ - stbi__context* s = z->s; - stbi__uint32 i, pixel_count = s->img_x * s->img_y; - stbi__uint16* p = (stbi__uint16*)z->out; - - // compute color-based transparency, assuming we've - // already got 65535 as the alpha value in the output - STBI_ASSERT(out_n == 2 || out_n == 4); - - if (out_n == 2) - { - for (i = 0; i < pixel_count; ++i) - { - p[1] = (p[0] == tc[0] ? 0 : 65535); - p += 2; - } - } - else - { - for (i = 0; i < pixel_count; ++i) - { - if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2]) - p[3] = 0; - p += 4; - } - } - return 1; -} - -static int stbi__expand_png_palette(stbi__png* a, stbi_uc* palette, int len, int pal_img_n) -{ - stbi__uint32 i, pixel_count = a->s->img_x * a->s->img_y; - stbi_uc *p, *temp_out, *orig = a->out; - - p = (stbi_uc*)stbi__malloc_mad2(pixel_count, pal_img_n, 0); - if (p == NULL) - return stbi__err("outofmem", "Out of memory"); - - // between here and free(out) below, exitting would leak - temp_out = p; - - if (pal_img_n == 3) - { - for (i = 0; i < pixel_count; ++i) - { - int n = orig[i] * 4; - p[0] = palette[n]; - p[1] = palette[n + 1]; - p[2] = palette[n + 2]; - p += 3; - } - } - else - { - for (i = 0; i < pixel_count; ++i) - { - int n = orig[i] * 4; - p[0] = palette[n]; - p[1] = palette[n + 1]; - p[2] = palette[n + 2]; - p[3] = palette[n + 3]; - p += 4; - } - } - STBI_FREE(a->out); - a->out = temp_out; - - STBI_NOTUSED(len); - - return 1; -} - -static int stbi__unpremultiply_on_load_global = 0; -static int stbi__de_iphone_flag_global = 0; - -STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply) -{ - stbi__unpremultiply_on_load_global = flag_true_if_should_unpremultiply; -} - -STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert) -{ - stbi__de_iphone_flag_global = flag_true_if_should_convert; -} - - #ifndef STBI_THREAD_LOCAL - #define stbi__unpremultiply_on_load stbi__unpremultiply_on_load_global - #define stbi__de_iphone_flag stbi__de_iphone_flag_global - #else -static STBI_THREAD_LOCAL int stbi__unpremultiply_on_load_local, stbi__unpremultiply_on_load_set; -static STBI_THREAD_LOCAL int stbi__de_iphone_flag_local, stbi__de_iphone_flag_set; - -STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply) -{ - stbi__unpremultiply_on_load_local = flag_true_if_should_unpremultiply; - stbi__unpremultiply_on_load_set = 1; -} - -STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert) -{ - stbi__de_iphone_flag_local = flag_true_if_should_convert; - stbi__de_iphone_flag_set = 1; -} - - #define stbi__unpremultiply_on_load \ - (stbi__unpremultiply_on_load_set ? stbi__unpremultiply_on_load_local : stbi__unpremultiply_on_load_global) - #define stbi__de_iphone_flag (stbi__de_iphone_flag_set ? stbi__de_iphone_flag_local : stbi__de_iphone_flag_global) - #endif // STBI_THREAD_LOCAL - -static void stbi__de_iphone(stbi__png* z) -{ - stbi__context* s = z->s; - stbi__uint32 i, pixel_count = s->img_x * s->img_y; - stbi_uc* p = z->out; - - if (s->img_out_n == 3) - { // convert bgr to rgb - for (i = 0; i < pixel_count; ++i) - { - stbi_uc t = p[0]; - p[0] = p[2]; - p[2] = t; - p += 3; - } - } - else - { - STBI_ASSERT(s->img_out_n == 4); - if (stbi__unpremultiply_on_load) - { - // convert bgr to rgb and unpremultiply - for (i = 0; i < pixel_count; ++i) - { - stbi_uc a = p[3]; - stbi_uc t = p[0]; - if (a) - { - stbi_uc half = a / 2; - p[0] = (p[2] * 255 + half) / a; - p[1] = (p[1] * 255 + half) / a; - p[2] = (t * 255 + half) / a; - } - else - { - p[0] = p[2]; - p[2] = t; - } - p += 4; - } - } - else - { - // convert bgr to rgb - for (i = 0; i < pixel_count; ++i) - { - stbi_uc t = p[0]; - p[0] = p[2]; - p[2] = t; - p += 4; - } - } - } -} - - #define STBI__PNG_TYPE(a, b, c, d) (((unsigned)(a) << 24) + ((unsigned)(b) << 16) + ((unsigned)(c) << 8) + (unsigned)(d)) - -static int stbi__parse_png_file(stbi__png* z, int scan, int req_comp) -{ - stbi_uc palette[1024], pal_img_n = 0; - stbi_uc has_trans = 0, tc[3] = {0}; - stbi__uint16 tc16[3]; - stbi__uint32 ioff = 0, idata_limit = 0, i, pal_len = 0; - int first = 1, k, interlace = 0, color = 0, is_iphone = 0; - stbi__context* s = z->s; - - z->expanded = NULL; - z->idata = NULL; - z->out = NULL; - - if (!stbi__check_png_header(s)) - return 0; - - if (scan == STBI__SCAN_type) - return 1; - - for (;;) - { - stbi__pngchunk c = stbi__get_chunk_header(s); - switch (c.type) - { - case STBI__PNG_TYPE('C', 'g', 'B', 'I'): - is_iphone = 1; - stbi__skip(s, c.length); - break; - case STBI__PNG_TYPE('I', 'H', 'D', 'R'): - { - int comp, filter; - if (!first) - return stbi__err("multiple IHDR", "Corrupt PNG"); - first = 0; - if (c.length != 13) - return stbi__err("bad IHDR len", "Corrupt PNG"); - s->img_x = stbi__get32be(s); - s->img_y = stbi__get32be(s); - if (s->img_y > STBI_MAX_DIMENSIONS) - return stbi__err("too large", "Very large image (corrupt?)"); - if (s->img_x > STBI_MAX_DIMENSIONS) - return stbi__err("too large", "Very large image (corrupt?)"); - z->depth = stbi__get8(s); - if (z->depth != 1 && z->depth != 2 && z->depth != 4 && z->depth != 8 && z->depth != 16) - return stbi__err("1/2/4/8/16-bit only", "PNG not supported: 1/2/4/8/16-bit only"); - color = stbi__get8(s); - if (color > 6) - return stbi__err("bad ctype", "Corrupt PNG"); - if (color == 3 && z->depth == 16) - return stbi__err("bad ctype", "Corrupt PNG"); - if (color == 3) - pal_img_n = 3; - else if (color & 1) - return stbi__err("bad ctype", "Corrupt PNG"); - comp = stbi__get8(s); - if (comp) - return stbi__err("bad comp method", "Corrupt PNG"); - filter = stbi__get8(s); - if (filter) - return stbi__err("bad filter method", "Corrupt PNG"); - interlace = stbi__get8(s); - if (interlace > 1) - return stbi__err("bad interlace method", "Corrupt PNG"); - if (!s->img_x || !s->img_y) - return stbi__err("0-pixel image", "Corrupt PNG"); - if (!pal_img_n) - { - s->img_n = (color & 2 ? 3 : 1) + (color & 4 ? 1 : 0); - if ((1 << 30) / s->img_x / s->img_n < s->img_y) - return stbi__err("too large", "Image too large to decode"); - } - else - { - // if paletted, then pal_n is our final components, and - // img_n is # components to decompress/filter. - s->img_n = 1; - if ((1 << 30) / s->img_x / 4 < s->img_y) - return stbi__err("too large", "Corrupt PNG"); - } - // even with SCAN_header, have to scan to see if we have a tRNS - break; - } - - case STBI__PNG_TYPE('P', 'L', 'T', 'E'): - { - if (first) - return stbi__err("first not IHDR", "Corrupt PNG"); - if (c.length > 256 * 3) - return stbi__err("invalid PLTE", "Corrupt PNG"); - pal_len = c.length / 3; - if (pal_len * 3 != c.length) - return stbi__err("invalid PLTE", "Corrupt PNG"); - for (i = 0; i < pal_len; ++i) - { - palette[i * 4 + 0] = stbi__get8(s); - palette[i * 4 + 1] = stbi__get8(s); - palette[i * 4 + 2] = stbi__get8(s); - palette[i * 4 + 3] = 255; - } - break; - } - - case STBI__PNG_TYPE('t', 'R', 'N', 'S'): - { - if (first) - return stbi__err("first not IHDR", "Corrupt PNG"); - if (z->idata) - return stbi__err("tRNS after IDAT", "Corrupt PNG"); - if (pal_img_n) - { - if (scan == STBI__SCAN_header) - { - s->img_n = 4; - return 1; - } - if (pal_len == 0) - return stbi__err("tRNS before PLTE", "Corrupt PNG"); - if (c.length > pal_len) - return stbi__err("bad tRNS len", "Corrupt PNG"); - pal_img_n = 4; - for (i = 0; i < c.length; ++i) - palette[i * 4 + 3] = stbi__get8(s); - } - else - { - if (!(s->img_n & 1)) - return stbi__err("tRNS with alpha", "Corrupt PNG"); - if (c.length != (stbi__uint32)s->img_n * 2) - return stbi__err("bad tRNS len", "Corrupt PNG"); - has_trans = 1; - // non-paletted with tRNS = constant alpha. if header-scanning, we can stop now. - if (scan == STBI__SCAN_header) - { - ++s->img_n; - return 1; - } - if (z->depth == 16) - { - for (k = 0; k < s->img_n && k < 3; ++k) // extra loop test to suppress false GCC warning - tc16[k] = (stbi__uint16)stbi__get16be(s); // copy the values as-is - } - else - { - for (k = 0; k < s->img_n && k < 3; ++k) - tc[k] = (stbi_uc)(stbi__get16be(s) & 255) * stbi__depth_scale_table[z->depth]; // non 8-bit images will be larger - } - } - break; - } - - case STBI__PNG_TYPE('I', 'D', 'A', 'T'): - { - if (first) - return stbi__err("first not IHDR", "Corrupt PNG"); - if (pal_img_n && !pal_len) - return stbi__err("no PLTE", "Corrupt PNG"); - if (scan == STBI__SCAN_header) - { - // header scan definitely stops at first IDAT - if (pal_img_n) - s->img_n = pal_img_n; - return 1; - } - if (c.length > (1u << 30)) - return stbi__err("IDAT size limit", "IDAT section larger than 2^30 bytes"); - if ((int)(ioff + c.length) < (int)ioff) - return 0; - if (ioff + c.length > idata_limit) - { - stbi__uint32 idata_limit_old = idata_limit; - stbi_uc* p; - if (idata_limit == 0) - idata_limit = c.length > 4096 ? c.length : 4096; - while (ioff + c.length > idata_limit) - idata_limit *= 2; - STBI_NOTUSED(idata_limit_old); - p = (stbi_uc*)STBI_REALLOC_SIZED(z->idata, idata_limit_old, idata_limit); - if (p == NULL) - return stbi__err("outofmem", "Out of memory"); - z->idata = p; - } - if (!stbi__getn(s, z->idata + ioff, c.length)) - return stbi__err("outofdata", "Corrupt PNG"); - ioff += c.length; - break; - } - - case STBI__PNG_TYPE('I', 'E', 'N', 'D'): - { - stbi__uint32 raw_len, bpl; - if (first) - return stbi__err("first not IHDR", "Corrupt PNG"); - if (scan != STBI__SCAN_load) - return 1; - if (z->idata == NULL) - return stbi__err("no IDAT", "Corrupt PNG"); - // initial guess for decoded data size to avoid unnecessary reallocs - bpl = (s->img_x * z->depth + 7) / 8; // bytes per line, per component - raw_len = bpl * s->img_y * s->img_n /* pixels */ + s->img_y /* filter mode per row */; - z->expanded = (stbi_uc*)stbi_zlib_decode_malloc_guesssize_headerflag((char*)z->idata, ioff, raw_len, (int*)&raw_len, !is_iphone); - if (z->expanded == NULL) - return 0; // zlib should set error - STBI_FREE(z->idata); - z->idata = NULL; - if ((req_comp == s->img_n + 1 && req_comp != 3 && !pal_img_n) || has_trans) - s->img_out_n = s->img_n + 1; - else - s->img_out_n = s->img_n; - if (!stbi__create_png_image(z, z->expanded, raw_len, s->img_out_n, z->depth, color, interlace)) - return 0; - if (has_trans) - { - if (z->depth == 16) - { - if (!stbi__compute_transparency16(z, tc16, s->img_out_n)) - return 0; - } - else - { - if (!stbi__compute_transparency(z, tc, s->img_out_n)) - return 0; - } - } - if (is_iphone && stbi__de_iphone_flag && s->img_out_n > 2) - stbi__de_iphone(z); - if (pal_img_n) - { - // pal_img_n == 3 or 4 - s->img_n = pal_img_n; // record the actual colors we had - s->img_out_n = pal_img_n; - if (req_comp >= 3) - s->img_out_n = req_comp; - if (!stbi__expand_png_palette(z, palette, pal_len, s->img_out_n)) - return 0; - } - else if (has_trans) - { - // non-paletted image with tRNS -> source image has (constant) alpha - ++s->img_n; - } - STBI_FREE(z->expanded); - z->expanded = NULL; - // end of PNG chunk, read and skip CRC - stbi__get32be(s); - return 1; - } - - default: - // if critical, fail - if (first) - return stbi__err("first not IHDR", "Corrupt PNG"); - if ((c.type & (1 << 29)) == 0) - { - #ifndef STBI_NO_FAILURE_STRINGS - // not threadsafe - static char invalid_chunk[] = "XXXX PNG chunk not known"; - invalid_chunk[0] = STBI__BYTECAST(c.type >> 24); - invalid_chunk[1] = STBI__BYTECAST(c.type >> 16); - invalid_chunk[2] = STBI__BYTECAST(c.type >> 8); - invalid_chunk[3] = STBI__BYTECAST(c.type >> 0); - #endif - return stbi__err(invalid_chunk, "PNG not supported: unknown PNG chunk type"); - } - stbi__skip(s, c.length); - break; - } - // end of PNG chunk, read and skip CRC - stbi__get32be(s); - } -} - -static void* stbi__do_png(stbi__png* p, int* x, int* y, int* n, int req_comp, stbi__result_info* ri) -{ - void* result = NULL; - if (req_comp < 0 || req_comp > 4) - return stbi__errpuc("bad req_comp", "Internal error"); - if (stbi__parse_png_file(p, STBI__SCAN_load, req_comp)) - { - if (p->depth <= 8) - ri->bits_per_channel = 8; - else if (p->depth == 16) - ri->bits_per_channel = 16; - else - return stbi__errpuc("bad bits_per_channel", "PNG not supported: unsupported color depth"); - result = p->out; - p->out = NULL; - if (req_comp && req_comp != p->s->img_out_n) - { - if (ri->bits_per_channel == 8) - result = stbi__convert_format((unsigned char*)result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y); - else - result = stbi__convert_format16((stbi__uint16*)result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y); - p->s->img_out_n = req_comp; - if (result == NULL) - return result; - } - *x = p->s->img_x; - *y = p->s->img_y; - if (n) - *n = p->s->img_n; - } - STBI_FREE(p->out); - p->out = NULL; - STBI_FREE(p->expanded); - p->expanded = NULL; - STBI_FREE(p->idata); - p->idata = NULL; - - return result; -} - -static void* stbi__png_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) -{ - stbi__png p; - p.s = s; - return stbi__do_png(&p, x, y, comp, req_comp, ri); -} - -static int stbi__png_test(stbi__context* s) -{ - int r; - r = stbi__check_png_header(s); - stbi__rewind(s); - return r; -} - -static int stbi__png_info_raw(stbi__png* p, int* x, int* y, int* comp) -{ - if (!stbi__parse_png_file(p, STBI__SCAN_header, 0)) - { - stbi__rewind(p->s); - return 0; - } - if (x) - *x = p->s->img_x; - if (y) - *y = p->s->img_y; - if (comp) - *comp = p->s->img_n; - return 1; -} - -static int stbi__png_info(stbi__context* s, int* x, int* y, int* comp) -{ - stbi__png p; - p.s = s; - return stbi__png_info_raw(&p, x, y, comp); -} - -static int stbi__png_is16(stbi__context* s) -{ - stbi__png p; - p.s = s; - if (!stbi__png_info_raw(&p, NULL, NULL, NULL)) - return 0; - if (p.depth != 16) - { - stbi__rewind(p.s); - return 0; - } - return 1; -} - #endif - -// Microsoft/Windows BMP image - - #ifndef STBI_NO_BMP -static int stbi__bmp_test_raw(stbi__context* s) -{ - int r; - int sz; - if (stbi__get8(s) != 'B') - return 0; - if (stbi__get8(s) != 'M') - return 0; - stbi__get32le(s); // discard filesize - stbi__get16le(s); // discard reserved - stbi__get16le(s); // discard reserved - stbi__get32le(s); // discard data offset - sz = stbi__get32le(s); - r = (sz == 12 || sz == 40 || sz == 56 || sz == 108 || sz == 124); - return r; -} - -static int stbi__bmp_test(stbi__context* s) -{ - int r = stbi__bmp_test_raw(s); - stbi__rewind(s); - return r; -} - -// returns 0..31 for the highest set bit -static int stbi__high_bit(unsigned int z) -{ - int n = 0; - if (z == 0) - return -1; - if (z >= 0x10000) - { - n += 16; - z >>= 16; - } - if (z >= 0x00100) - { - n += 8; - z >>= 8; - } - if (z >= 0x00010) - { - n += 4; - z >>= 4; - } - if (z >= 0x00004) - { - n += 2; - z >>= 2; - } - if (z >= 0x00002) - { - n += 1; /* >>= 1;*/ - } - return n; -} - -static int stbi__bitcount(unsigned int a) -{ - a = (a & 0x55555555) + ((a >> 1) & 0x55555555); // max 2 - a = (a & 0x33333333) + ((a >> 2) & 0x33333333); // max 4 - a = (a + (a >> 4)) & 0x0f0f0f0f; // max 8 per 4, now 8 bits - a = (a + (a >> 8)); // max 16 per 8 bits - a = (a + (a >> 16)); // max 32 per 8 bits - return a & 0xff; -} - -// extract an arbitrarily-aligned N-bit value (N=bits) -// from v, and then make it 8-bits long and fractionally -// extend it to full full range. -static int stbi__shiftsigned(unsigned int v, int shift, int bits) -{ - static unsigned int mul_table[9] = { - 0, - 0xff /*0b11111111*/, - 0x55 /*0b01010101*/, - 0x49 /*0b01001001*/, - 0x11 /*0b00010001*/, - 0x21 /*0b00100001*/, - 0x41 /*0b01000001*/, - 0x81 /*0b10000001*/, - 0x01 /*0b00000001*/, - }; - static unsigned int shift_table[9] = { - 0, - 0, - 0, - 1, - 0, - 2, - 4, - 6, - 0, - }; - if (shift < 0) - v <<= -shift; - else - v >>= shift; - STBI_ASSERT(v < 256); - v >>= (8 - bits); - STBI_ASSERT(bits >= 0 && bits <= 8); - return (int)((unsigned)v * mul_table[bits]) >> shift_table[bits]; -} - -typedef struct { - int bpp, offset, hsz; - unsigned int mr, mg, mb, ma, all_a; - int extra_read; -} stbi__bmp_data; - -static int stbi__bmp_set_mask_defaults(stbi__bmp_data* info, int compress) -{ - // BI_BITFIELDS specifies masks explicitly, don't override - if (compress == 3) - return 1; - - if (compress == 0) - { - if (info->bpp == 16) - { - info->mr = 31u << 10; - info->mg = 31u << 5; - info->mb = 31u << 0; - } - else if (info->bpp == 32) - { - info->mr = 0xffu << 16; - info->mg = 0xffu << 8; - info->mb = 0xffu << 0; - info->ma = 0xffu << 24; - info->all_a = 0; // if all_a is 0 at end, then we loaded alpha channel but it was all 0 - } - else - { - // otherwise, use defaults, which is all-0 - info->mr = info->mg = info->mb = info->ma = 0; - } - return 1; - } - return 0; // error -} - -static void* stbi__bmp_parse_header(stbi__context* s, stbi__bmp_data* info) -{ - int hsz; - if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') - return stbi__errpuc("not BMP", "Corrupt BMP"); - stbi__get32le(s); // discard filesize - stbi__get16le(s); // discard reserved - stbi__get16le(s); // discard reserved - info->offset = stbi__get32le(s); - info->hsz = hsz = stbi__get32le(s); - info->mr = info->mg = info->mb = info->ma = 0; - info->extra_read = 14; - - if (info->offset < 0) - return stbi__errpuc("bad BMP", "bad BMP"); - - if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) - return stbi__errpuc("unknown BMP", "BMP type not supported: unknown"); - if (hsz == 12) - { - s->img_x = stbi__get16le(s); - s->img_y = stbi__get16le(s); - } - else - { - s->img_x = stbi__get32le(s); - s->img_y = stbi__get32le(s); - } - if (stbi__get16le(s) != 1) - return stbi__errpuc("bad BMP", "bad BMP"); - info->bpp = stbi__get16le(s); - if (hsz != 12) - { - int compress = stbi__get32le(s); - if (compress == 1 || compress == 2) - return stbi__errpuc("BMP RLE", "BMP type not supported: RLE"); - if (compress >= 4) - return stbi__errpuc("BMP JPEG/PNG", "BMP type not supported: unsupported compression"); // this includes PNG/JPEG modes - if (compress == 3 && info->bpp != 16 && info->bpp != 32) - return stbi__errpuc("bad BMP", "bad BMP"); // bitfields requires 16 or 32 bits/pixel - stbi__get32le(s); // discard sizeof - stbi__get32le(s); // discard hres - stbi__get32le(s); // discard vres - stbi__get32le(s); // discard colorsused - stbi__get32le(s); // discard max important - if (hsz == 40 || hsz == 56) - { - if (hsz == 56) - { - stbi__get32le(s); - stbi__get32le(s); - stbi__get32le(s); - stbi__get32le(s); - } - if (info->bpp == 16 || info->bpp == 32) - { - if (compress == 0) - { - stbi__bmp_set_mask_defaults(info, compress); - } - else if (compress == 3) - { - info->mr = stbi__get32le(s); - info->mg = stbi__get32le(s); - info->mb = stbi__get32le(s); - info->extra_read += 12; - // not documented, but generated by photoshop and handled by mspaint - if (info->mr == info->mg && info->mg == info->mb) - { - // ?!?!? - return stbi__errpuc("bad BMP", "bad BMP"); - } - } - else - return stbi__errpuc("bad BMP", "bad BMP"); - } - } - else - { - // V4/V5 header - int i; - if (hsz != 108 && hsz != 124) - return stbi__errpuc("bad BMP", "bad BMP"); - info->mr = stbi__get32le(s); - info->mg = stbi__get32le(s); - info->mb = stbi__get32le(s); - info->ma = stbi__get32le(s); - if (compress != 3) // override mr/mg/mb unless in BI_BITFIELDS mode, as per docs - stbi__bmp_set_mask_defaults(info, compress); - stbi__get32le(s); // discard color space - for (i = 0; i < 12; ++i) - stbi__get32le(s); // discard color space parameters - if (hsz == 124) - { - stbi__get32le(s); // discard rendering intent - stbi__get32le(s); // discard offset of profile data - stbi__get32le(s); // discard size of profile data - stbi__get32le(s); // discard reserved - } - } - } - return (void*)1; -} - -static void* stbi__bmp_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) -{ - stbi_uc* out; - unsigned int mr = 0, mg = 0, mb = 0, ma = 0, all_a; - stbi_uc pal[256][4]; - int psize = 0, i, j, width; - int flip_vertically, pad, target; - stbi__bmp_data info; - STBI_NOTUSED(ri); - - info.all_a = 255; - if (stbi__bmp_parse_header(s, &info) == NULL) - return NULL; // error code already set - - flip_vertically = ((int)s->img_y) > 0; - s->img_y = abs((int)s->img_y); - - if (s->img_y > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - if (s->img_x > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - - mr = info.mr; - mg = info.mg; - mb = info.mb; - ma = info.ma; - all_a = info.all_a; - - if (info.hsz == 12) - { - if (info.bpp < 24) - psize = (info.offset - info.extra_read - 24) / 3; - } - else - { - if (info.bpp < 16) - psize = (info.offset - info.extra_read - info.hsz) >> 2; - } - if (psize == 0) - { - // accept some number of extra bytes after the header, but if the offset points either to before - // the header ends or implies a large amount of extra data, reject the file as malformed - int bytes_read_so_far = s->callback_already_read + (int)(s->img_buffer - s->img_buffer_original); - int header_limit = 1024; // max we actually read is below 256 bytes currently. - int extra_data_limit = 256 * 4; // what ordinarily goes here is a palette; 256 entries*4 bytes is its max size. - if (bytes_read_so_far <= 0 || bytes_read_so_far > header_limit) - { - return stbi__errpuc("bad header", "Corrupt BMP"); - } - // we established that bytes_read_so_far is positive and sensible. - // the first half of this test rejects offsets that are either too small positives, or - // negative, and guarantees that info.offset >= bytes_read_so_far > 0. this in turn - // ensures the number computed in the second half of the test can't overflow. - if (info.offset < bytes_read_so_far || info.offset - bytes_read_so_far > extra_data_limit) - { - return stbi__errpuc("bad offset", "Corrupt BMP"); - } - else - { - stbi__skip(s, info.offset - bytes_read_so_far); - } - } - - if (info.bpp == 24 && ma == 0xff000000) - s->img_n = 3; - else - s->img_n = ma ? 4 : 3; - if (req_comp && req_comp >= 3) // we can directly decode 3 or 4 - target = req_comp; - else - target = s->img_n; // if they want monochrome, we'll post-convert - - // sanity-check size - if (!stbi__mad3sizes_valid(target, s->img_x, s->img_y, 0)) - return stbi__errpuc("too large", "Corrupt BMP"); - - out = (stbi_uc*)stbi__malloc_mad3(target, s->img_x, s->img_y, 0); - if (!out) - return stbi__errpuc("outofmem", "Out of memory"); - if (info.bpp < 16) - { - int z = 0; - if (psize == 0 || psize > 256) - { - STBI_FREE(out); - return stbi__errpuc("invalid", "Corrupt BMP"); - } - for (i = 0; i < psize; ++i) - { - pal[i][2] = stbi__get8(s); - pal[i][1] = stbi__get8(s); - pal[i][0] = stbi__get8(s); - if (info.hsz != 12) - stbi__get8(s); - pal[i][3] = 255; - } - stbi__skip(s, info.offset - info.extra_read - info.hsz - psize * (info.hsz == 12 ? 3 : 4)); - if (info.bpp == 1) - width = (s->img_x + 7) >> 3; - else if (info.bpp == 4) - width = (s->img_x + 1) >> 1; - else if (info.bpp == 8) - width = s->img_x; - else - { - STBI_FREE(out); - return stbi__errpuc("bad bpp", "Corrupt BMP"); - } - pad = (-width) & 3; - if (info.bpp == 1) - { - for (j = 0; j < (int)s->img_y; ++j) - { - int bit_offset = 7, v = stbi__get8(s); - for (i = 0; i < (int)s->img_x; ++i) - { - int color = (v >> bit_offset) & 0x1; - out[z++] = pal[color][0]; - out[z++] = pal[color][1]; - out[z++] = pal[color][2]; - if (target == 4) - out[z++] = 255; - if (i + 1 == (int)s->img_x) - break; - if ((--bit_offset) < 0) - { - bit_offset = 7; - v = stbi__get8(s); - } - } - stbi__skip(s, pad); - } - } - else - { - for (j = 0; j < (int)s->img_y; ++j) - { - for (i = 0; i < (int)s->img_x; i += 2) - { - int v = stbi__get8(s), v2 = 0; - if (info.bpp == 4) - { - v2 = v & 15; - v >>= 4; - } - out[z++] = pal[v][0]; - out[z++] = pal[v][1]; - out[z++] = pal[v][2]; - if (target == 4) - out[z++] = 255; - if (i + 1 == (int)s->img_x) - break; - v = (info.bpp == 8) ? stbi__get8(s) : v2; - out[z++] = pal[v][0]; - out[z++] = pal[v][1]; - out[z++] = pal[v][2]; - if (target == 4) - out[z++] = 255; - } - stbi__skip(s, pad); - } - } - } - else - { - int rshift = 0, gshift = 0, bshift = 0, ashift = 0, rcount = 0, gcount = 0, bcount = 0, acount = 0; - int z = 0; - int easy = 0; - stbi__skip(s, info.offset - info.extra_read - info.hsz); - if (info.bpp == 24) - width = 3 * s->img_x; - else if (info.bpp == 16) - width = 2 * s->img_x; - else /* bpp = 32 and pad = 0 */ - width = 0; - pad = (-width) & 3; - if (info.bpp == 24) - { - easy = 1; - } - else if (info.bpp == 32) - { - if (mb == 0xff && mg == 0xff00 && mr == 0x00ff0000 && ma == 0xff000000) - easy = 2; - } - if (!easy) - { - if (!mr || !mg || !mb) - { - STBI_FREE(out); - return stbi__errpuc("bad masks", "Corrupt BMP"); - } - // right shift amt to put high bit in position #7 - rshift = stbi__high_bit(mr) - 7; - rcount = stbi__bitcount(mr); - gshift = stbi__high_bit(mg) - 7; - gcount = stbi__bitcount(mg); - bshift = stbi__high_bit(mb) - 7; - bcount = stbi__bitcount(mb); - ashift = stbi__high_bit(ma) - 7; - acount = stbi__bitcount(ma); - if (rcount > 8 || gcount > 8 || bcount > 8 || acount > 8) - { - STBI_FREE(out); - return stbi__errpuc("bad masks", "Corrupt BMP"); - } - } - for (j = 0; j < (int)s->img_y; ++j) - { - if (easy) - { - for (i = 0; i < (int)s->img_x; ++i) - { - unsigned char a; - out[z + 2] = stbi__get8(s); - out[z + 1] = stbi__get8(s); - out[z + 0] = stbi__get8(s); - z += 3; - a = (easy == 2 ? stbi__get8(s) : 255); - all_a |= a; - if (target == 4) - out[z++] = a; - } - } - else - { - int bpp = info.bpp; - for (i = 0; i < (int)s->img_x; ++i) - { - stbi__uint32 v = (bpp == 16 ? (stbi__uint32)stbi__get16le(s) : stbi__get32le(s)); - unsigned int a; - out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mr, rshift, rcount)); - out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mg, gshift, gcount)); - out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mb, bshift, bcount)); - a = (ma ? stbi__shiftsigned(v & ma, ashift, acount) : 255); - all_a |= a; - if (target == 4) - out[z++] = STBI__BYTECAST(a); - } - } - stbi__skip(s, pad); - } - } - - // if alpha channel is all 0s, replace with all 255s - if (target == 4 && all_a == 0) - for (i = 4 * s->img_x * s->img_y - 1; i >= 0; i -= 4) - out[i] = 255; - - if (flip_vertically) - { - stbi_uc t; - for (j = 0; j < (int)s->img_y >> 1; ++j) - { - stbi_uc* p1 = out + j * s->img_x * target; - stbi_uc* p2 = out + (s->img_y - 1 - j) * s->img_x * target; - for (i = 0; i < (int)s->img_x * target; ++i) - { - t = p1[i]; - p1[i] = p2[i]; - p2[i] = t; - } - } - } - - if (req_comp && req_comp != target) - { - out = stbi__convert_format(out, target, req_comp, s->img_x, s->img_y); - if (out == NULL) - return out; // stbi__convert_format frees input on failure - } - - *x = s->img_x; - *y = s->img_y; - if (comp) - *comp = s->img_n; - return out; -} - #endif - - // Targa Truevision - TGA - // by Jonathan Dummer - #ifndef STBI_NO_TGA -// returns STBI_rgb or whatever, 0 on error -static int stbi__tga_get_comp(int bits_per_pixel, int is_grey, int* is_rgb16) -{ - // only RGB or RGBA (incl. 16bit) or grey allowed - if (is_rgb16) - *is_rgb16 = 0; - switch (bits_per_pixel) - { - case 8: return STBI_grey; - case 16: - if (is_grey) - return STBI_grey_alpha; - // fallthrough - case 15: - if (is_rgb16) - *is_rgb16 = 1; - return STBI_rgb; - case 24: // fallthrough - case 32: return bits_per_pixel / 8; - default: return 0; - } -} - -static int stbi__tga_info(stbi__context* s, int* x, int* y, int* comp) -{ - int tga_w, tga_h, tga_comp, tga_image_type, tga_bits_per_pixel, tga_colormap_bpp; - int sz, tga_colormap_type; - stbi__get8(s); // discard Offset - tga_colormap_type = stbi__get8(s); // colormap type - if (tga_colormap_type > 1) - { - stbi__rewind(s); - return 0; // only RGB or indexed allowed - } - tga_image_type = stbi__get8(s); // image type - if (tga_colormap_type == 1) - { // colormapped (paletted) image - if (tga_image_type != 1 && tga_image_type != 9) - { - stbi__rewind(s); - return 0; - } - stbi__skip(s, 4); // skip index of first colormap entry and number of entries - sz = stbi__get8(s); // check bits per palette color entry - if ((sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32)) - { - stbi__rewind(s); - return 0; - } - stbi__skip(s, 4); // skip image x and y origin - tga_colormap_bpp = sz; - } - else - { // "normal" image w/o colormap - only RGB or grey allowed, +/- RLE - if ((tga_image_type != 2) && (tga_image_type != 3) && (tga_image_type != 10) && (tga_image_type != 11)) - { - stbi__rewind(s); - return 0; // only RGB or grey allowed, +/- RLE - } - stbi__skip(s, 9); // skip colormap specification and image x/y origin - tga_colormap_bpp = 0; - } - tga_w = stbi__get16le(s); - if (tga_w < 1) - { - stbi__rewind(s); - return 0; // test width - } - tga_h = stbi__get16le(s); - if (tga_h < 1) - { - stbi__rewind(s); - return 0; // test height - } - tga_bits_per_pixel = stbi__get8(s); // bits per pixel - stbi__get8(s); // ignore alpha bits - if (tga_colormap_bpp != 0) - { - if ((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16)) - { - // when using a colormap, tga_bits_per_pixel is the size of the indexes - // I don't think anything but 8 or 16bit indexes makes sense - stbi__rewind(s); - return 0; - } - tga_comp = stbi__tga_get_comp(tga_colormap_bpp, 0, NULL); - } - else - { - tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3) || (tga_image_type == 11), NULL); - } - if (!tga_comp) - { - stbi__rewind(s); - return 0; - } - if (x) - *x = tga_w; - if (y) - *y = tga_h; - if (comp) - *comp = tga_comp; - return 1; // seems to have passed everything -} - -static int stbi__tga_test(stbi__context* s) -{ - int res = 0; - int sz, tga_color_type; - stbi__get8(s); // discard Offset - tga_color_type = stbi__get8(s); // color type - if (tga_color_type > 1) - goto errorEnd; // only RGB or indexed allowed - sz = stbi__get8(s); // image type - if (tga_color_type == 1) - { // colormapped (paletted) image - if (sz != 1 && sz != 9) - goto errorEnd; // colortype 1 demands image type 1 or 9 - stbi__skip(s, 4); // skip index of first colormap entry and number of entries - sz = stbi__get8(s); // check bits per palette color entry - if ((sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32)) - goto errorEnd; - stbi__skip(s, 4); // skip image x and y origin - } - else - { // "normal" image w/o colormap - if ((sz != 2) && (sz != 3) && (sz != 10) && (sz != 11)) - goto errorEnd; // only RGB or grey allowed, +/- RLE - stbi__skip(s, 9); // skip colormap specification and image x/y origin - } - if (stbi__get16le(s) < 1) - goto errorEnd; // test width - if (stbi__get16le(s) < 1) - goto errorEnd; // test height - sz = stbi__get8(s); // bits per pixel - if ((tga_color_type == 1) && (sz != 8) && (sz != 16)) - goto errorEnd; // for colormapped images, bpp is size of an index - if ((sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32)) - goto errorEnd; - - res = 1; // if we got this far, everything's good and we can return 1 instead of 0 - -errorEnd: - stbi__rewind(s); - return res; -} - -// read 16bit value and convert to 24bit RGB -static void stbi__tga_read_rgb16(stbi__context* s, stbi_uc* out) -{ - stbi__uint16 px = (stbi__uint16)stbi__get16le(s); - stbi__uint16 fiveBitMask = 31; - // we have 3 channels with 5bits each - int r = (px >> 10) & fiveBitMask; - int g = (px >> 5) & fiveBitMask; - int b = px & fiveBitMask; - // Note that this saves the data in RGB(A) order, so it doesn't need to be swapped later - out[0] = (stbi_uc)((r * 255) / 31); - out[1] = (stbi_uc)((g * 255) / 31); - out[2] = (stbi_uc)((b * 255) / 31); - - // some people claim that the most significant bit might be used for alpha - // (possibly if an alpha-bit is set in the "image descriptor byte") - // but that only made 16bit test images completely translucent.. - // so let's treat all 15 and 16bit TGAs as RGB with no alpha. -} - -static void* stbi__tga_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) -{ - // read in the TGA header stuff - int tga_offset = stbi__get8(s); - int tga_indexed = stbi__get8(s); - int tga_image_type = stbi__get8(s); - int tga_is_RLE = 0; - int tga_palette_start = stbi__get16le(s); - int tga_palette_len = stbi__get16le(s); - int tga_palette_bits = stbi__get8(s); - int tga_x_origin = stbi__get16le(s); - int tga_y_origin = stbi__get16le(s); - int tga_width = stbi__get16le(s); - int tga_height = stbi__get16le(s); - int tga_bits_per_pixel = stbi__get8(s); - int tga_comp, tga_rgb16 = 0; - int tga_inverted = stbi__get8(s); - // int tga_alpha_bits = tga_inverted & 15; // the 4 lowest bits - unused (useless?) - // image data - unsigned char* tga_data; - unsigned char* tga_palette = NULL; - int i, j; - unsigned char raw_data[4] = {0}; - int RLE_count = 0; - int RLE_repeating = 0; - int read_next_pixel = 1; - STBI_NOTUSED(ri); - STBI_NOTUSED(tga_x_origin); // @TODO - STBI_NOTUSED(tga_y_origin); // @TODO - - if (tga_height > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - if (tga_width > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - - // do a tiny bit of precessing - if (tga_image_type >= 8) - { - tga_image_type -= 8; - tga_is_RLE = 1; - } - tga_inverted = 1 - ((tga_inverted >> 5) & 1); - - // If I'm paletted, then I'll use the number of bits from the palette - if (tga_indexed) - tga_comp = stbi__tga_get_comp(tga_palette_bits, 0, &tga_rgb16); - else - tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3), &tga_rgb16); - - if (!tga_comp) // shouldn't really happen, stbi__tga_test() should have ensured basic consistency - return stbi__errpuc("bad format", "Can't find out TGA pixelformat"); - - // tga info - *x = tga_width; - *y = tga_height; - if (comp) - *comp = tga_comp; - - if (!stbi__mad3sizes_valid(tga_width, tga_height, tga_comp, 0)) - return stbi__errpuc("too large", "Corrupt TGA"); - - tga_data = (unsigned char*)stbi__malloc_mad3(tga_width, tga_height, tga_comp, 0); - if (!tga_data) - return stbi__errpuc("outofmem", "Out of memory"); - - // skip to the data's starting position (offset usually = 0) - stbi__skip(s, tga_offset); - - if (!tga_indexed && !tga_is_RLE && !tga_rgb16) - { - for (i = 0; i < tga_height; ++i) - { - int row = tga_inverted ? tga_height - i - 1 : i; - stbi_uc* tga_row = tga_data + row * tga_width * tga_comp; - stbi__getn(s, tga_row, tga_width * tga_comp); - } - } - else - { - // do I need to load a palette? - if (tga_indexed) - { - if (tga_palette_len == 0) - { /* you have to have at least one entry! */ - STBI_FREE(tga_data); - return stbi__errpuc("bad palette", "Corrupt TGA"); - } - - // any data to skip? (offset usually = 0) - stbi__skip(s, tga_palette_start); - // load the palette - tga_palette = (unsigned char*)stbi__malloc_mad2(tga_palette_len, tga_comp, 0); - if (!tga_palette) - { - STBI_FREE(tga_data); - return stbi__errpuc("outofmem", "Out of memory"); - } - if (tga_rgb16) - { - stbi_uc* pal_entry = tga_palette; - STBI_ASSERT(tga_comp == STBI_rgb); - for (i = 0; i < tga_palette_len; ++i) - { - stbi__tga_read_rgb16(s, pal_entry); - pal_entry += tga_comp; - } - } - else if (!stbi__getn(s, tga_palette, tga_palette_len * tga_comp)) - { - STBI_FREE(tga_data); - STBI_FREE(tga_palette); - return stbi__errpuc("bad palette", "Corrupt TGA"); - } - } - // load the data - for (i = 0; i < tga_width * tga_height; ++i) - { - // if I'm in RLE mode, do I need to get a RLE stbi__pngchunk? - if (tga_is_RLE) - { - if (RLE_count == 0) - { - // yep, get the next byte as a RLE command - int RLE_cmd = stbi__get8(s); - RLE_count = 1 + (RLE_cmd & 127); - RLE_repeating = RLE_cmd >> 7; - read_next_pixel = 1; - } - else if (!RLE_repeating) - { - read_next_pixel = 1; - } - } - else - { - read_next_pixel = 1; - } - // OK, if I need to read a pixel, do it now - if (read_next_pixel) - { - // load however much data we did have - if (tga_indexed) - { - // read in index, then perform the lookup - int pal_idx = (tga_bits_per_pixel == 8) ? stbi__get8(s) : stbi__get16le(s); - if (pal_idx >= tga_palette_len) - { - // invalid index - pal_idx = 0; - } - pal_idx *= tga_comp; - for (j = 0; j < tga_comp; ++j) - { - raw_data[j] = tga_palette[pal_idx + j]; - } - } - else if (tga_rgb16) - { - STBI_ASSERT(tga_comp == STBI_rgb); - stbi__tga_read_rgb16(s, raw_data); - } - else - { - // read in the data raw - for (j = 0; j < tga_comp; ++j) - { - raw_data[j] = stbi__get8(s); - } - } - // clear the reading flag for the next pixel - read_next_pixel = 0; - } // end of reading a pixel - - // copy data - for (j = 0; j < tga_comp; ++j) - tga_data[i * tga_comp + j] = raw_data[j]; - - // in case we're in RLE mode, keep counting down - --RLE_count; - } - // do I need to invert the image? - if (tga_inverted) - { - for (j = 0; j * 2 < tga_height; ++j) - { - int index1 = j * tga_width * tga_comp; - int index2 = (tga_height - 1 - j) * tga_width * tga_comp; - for (i = tga_width * tga_comp; i > 0; --i) - { - unsigned char temp = tga_data[index1]; - tga_data[index1] = tga_data[index2]; - tga_data[index2] = temp; - ++index1; - ++index2; - } - } - } - // clear my palette, if I had one - if (tga_palette != NULL) - { - STBI_FREE(tga_palette); - } - } - - // swap RGB - if the source data was RGB16, it already is in the right order - if (tga_comp >= 3 && !tga_rgb16) - { - unsigned char* tga_pixel = tga_data; - for (i = 0; i < tga_width * tga_height; ++i) - { - unsigned char temp = tga_pixel[0]; - tga_pixel[0] = tga_pixel[2]; - tga_pixel[2] = temp; - tga_pixel += tga_comp; - } - } - - // convert to target component count - if (req_comp && req_comp != tga_comp) - tga_data = stbi__convert_format(tga_data, tga_comp, req_comp, tga_width, tga_height); - - // the things I do to get rid of an error message, and yet keep - // Microsoft's C compilers happy... [8^( - tga_palette_start = tga_palette_len = tga_palette_bits = tga_x_origin = tga_y_origin = 0; - STBI_NOTUSED(tga_palette_start); - // OK, done - return tga_data; -} - #endif - -// ************************************************************************************************* -// Photoshop PSD loader -- PD by Thatcher Ulrich, integration by Nicolas Schulz, tweaked by STB - - #ifndef STBI_NO_PSD -static int stbi__psd_test(stbi__context* s) -{ - int r = (stbi__get32be(s) == 0x38425053); - stbi__rewind(s); - return r; -} - -static int stbi__psd_decode_rle(stbi__context* s, stbi_uc* p, int pixelCount) -{ - int count, nleft, len; - - count = 0; - while ((nleft = pixelCount - count) > 0) - { - len = stbi__get8(s); - if (len == 128) - { - // No-op. - } - else if (len < 128) - { - // Copy next len+1 bytes literally. - len++; - if (len > nleft) - return 0; // corrupt data - count += len; - while (len) - { - *p = stbi__get8(s); - p += 4; - len--; - } - } - else if (len > 128) - { - stbi_uc val; - // Next -len+1 bytes in the dest are replicated from next source byte. - // (Interpret len as a negative 8-bit int.) - len = 257 - len; - if (len > nleft) - return 0; // corrupt data - val = stbi__get8(s); - count += len; - while (len) - { - *p = val; - p += 4; - len--; - } - } - } - - return 1; -} - -static void* stbi__psd_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri, int bpc) -{ - int pixelCount; - int channelCount, compression; - int channel, i; - int bitdepth; - int w, h; - stbi_uc* out; - STBI_NOTUSED(ri); - - // Check identifier - if (stbi__get32be(s) != 0x38425053) // "8BPS" - return stbi__errpuc("not PSD", "Corrupt PSD image"); - - // Check file type version. - if (stbi__get16be(s) != 1) - return stbi__errpuc("wrong version", "Unsupported version of PSD image"); - - // Skip 6 reserved bytes. - stbi__skip(s, 6); - - // Read the number of channels (R, G, B, A, etc). - channelCount = stbi__get16be(s); - if (channelCount < 0 || channelCount > 16) - return stbi__errpuc("wrong channel count", "Unsupported number of channels in PSD image"); - - // Read the rows and columns of the image. - h = stbi__get32be(s); - w = stbi__get32be(s); - - if (h > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - if (w > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - - // Make sure the depth is 8 bits. - bitdepth = stbi__get16be(s); - if (bitdepth != 8 && bitdepth != 16) - return stbi__errpuc("unsupported bit depth", "PSD bit depth is not 8 or 16 bit"); - - // Make sure the color mode is RGB. - // Valid options are: - // 0: Bitmap - // 1: Grayscale - // 2: Indexed color - // 3: RGB color - // 4: CMYK color - // 7: Multichannel - // 8: Duotone - // 9: Lab color - if (stbi__get16be(s) != 3) - return stbi__errpuc("wrong color format", "PSD is not in RGB color format"); - - // Skip the Mode Data. (It's the palette for indexed color; other info for other modes.) - stbi__skip(s, stbi__get32be(s)); - - // Skip the image resources. (resolution, pen tool paths, etc) - stbi__skip(s, stbi__get32be(s)); - - // Skip the reserved data. - stbi__skip(s, stbi__get32be(s)); - - // Find out if the data is compressed. - // Known values: - // 0: no compression - // 1: RLE compressed - compression = stbi__get16be(s); - if (compression > 1) - return stbi__errpuc("bad compression", "PSD has an unknown compression format"); - - // Check size - if (!stbi__mad3sizes_valid(4, w, h, 0)) - return stbi__errpuc("too large", "Corrupt PSD"); - - // Create the destination image. - - if (!compression && bitdepth == 16 && bpc == 16) - { - out = (stbi_uc*)stbi__malloc_mad3(8, w, h, 0); - ri->bits_per_channel = 16; - } - else - out = (stbi_uc*)stbi__malloc(4 * w * h); - - if (!out) - return stbi__errpuc("outofmem", "Out of memory"); - pixelCount = w * h; - - // Initialize the data to zero. - // memset( out, 0, pixelCount * 4 ); - - // Finally, the image data. - if (compression) - { - // RLE as used by .PSD and .TIFF - // Loop until you get the number of unpacked bytes you are expecting: - // Read the next source byte into n. - // If n is between 0 and 127 inclusive, copy the next n+1 bytes literally. - // Else if n is between -127 and -1 inclusive, copy the next byte -n+1 times. - // Else if n is 128, noop. - // Endloop - - // The RLE-compressed data is preceded by a 2-byte data count for each row in the data, - // which we're going to just skip. - stbi__skip(s, h * channelCount * 2); - - // Read the RLE data by channel. - for (channel = 0; channel < 4; channel++) - { - stbi_uc* p; - - p = out + channel; - if (channel >= channelCount) - { - // Fill this channel with default data. - for (i = 0; i < pixelCount; i++, p += 4) - *p = (channel == 3 ? 255 : 0); - } - else - { - // Read the RLE data. - if (!stbi__psd_decode_rle(s, p, pixelCount)) - { - STBI_FREE(out); - return stbi__errpuc("corrupt", "bad RLE data"); - } - } - } - } - else - { - // We're at the raw image data. It's each channel in order (Red, Green, Blue, Alpha, ...) - // where each channel consists of an 8-bit (or 16-bit) value for each pixel in the image. - - // Read the data by channel. - for (channel = 0; channel < 4; channel++) - { - if (channel >= channelCount) - { - // Fill this channel with default data. - if (bitdepth == 16 && bpc == 16) - { - stbi__uint16* q = ((stbi__uint16*)out) + channel; - stbi__uint16 val = channel == 3 ? 65535 : 0; - for (i = 0; i < pixelCount; i++, q += 4) - *q = val; - } - else - { - stbi_uc* p = out + channel; - stbi_uc val = channel == 3 ? 255 : 0; - for (i = 0; i < pixelCount; i++, p += 4) - *p = val; - } - } - else - { - if (ri->bits_per_channel == 16) - { // output bpc - stbi__uint16* q = ((stbi__uint16*)out) + channel; - for (i = 0; i < pixelCount; i++, q += 4) - *q = (stbi__uint16)stbi__get16be(s); - } - else - { - stbi_uc* p = out + channel; - if (bitdepth == 16) - { // input bpc - for (i = 0; i < pixelCount; i++, p += 4) - *p = (stbi_uc)(stbi__get16be(s) >> 8); - } - else - { - for (i = 0; i < pixelCount; i++, p += 4) - *p = stbi__get8(s); - } - } - } - } - } - - // remove weird white matte from PSD - if (channelCount >= 4) - { - if (ri->bits_per_channel == 16) - { - for (i = 0; i < w * h; ++i) - { - stbi__uint16* pixel = (stbi__uint16*)out + 4 * i; - if (pixel[3] != 0 && pixel[3] != 65535) - { - float a = pixel[3] / 65535.0f; - float ra = 1.0f / a; - float inv_a = 65535.0f * (1 - ra); - pixel[0] = (stbi__uint16)(pixel[0] * ra + inv_a); - pixel[1] = (stbi__uint16)(pixel[1] * ra + inv_a); - pixel[2] = (stbi__uint16)(pixel[2] * ra + inv_a); - } - } - } - else - { - for (i = 0; i < w * h; ++i) - { - unsigned char* pixel = out + 4 * i; - if (pixel[3] != 0 && pixel[3] != 255) - { - float a = pixel[3] / 255.0f; - float ra = 1.0f / a; - float inv_a = 255.0f * (1 - ra); - pixel[0] = (unsigned char)(pixel[0] * ra + inv_a); - pixel[1] = (unsigned char)(pixel[1] * ra + inv_a); - pixel[2] = (unsigned char)(pixel[2] * ra + inv_a); - } - } - } - } - - // convert to desired output format - if (req_comp && req_comp != 4) - { - if (ri->bits_per_channel == 16) - out = (stbi_uc*)stbi__convert_format16((stbi__uint16*)out, 4, req_comp, w, h); - else - out = stbi__convert_format(out, 4, req_comp, w, h); - if (out == NULL) - return out; // stbi__convert_format frees input on failure - } - - if (comp) - *comp = 4; - *y = h; - *x = w; - - return out; -} - #endif - -// ************************************************************************************************* -// Softimage PIC loader -// by Tom Seddon -// -// See http://softimage.wiki.softimage.com/index.php/INFO:_PIC_file_format -// See http://ozviz.wasp.uwa.edu.au/~pbourke/dataformats/softimagepic/ - - #ifndef STBI_NO_PIC -static int stbi__pic_is4(stbi__context* s, const char* str) -{ - int i; - for (i = 0; i < 4; ++i) - if (stbi__get8(s) != (stbi_uc)str[i]) - return 0; - - return 1; -} - -static int stbi__pic_test_core(stbi__context* s) -{ - int i; - - if (!stbi__pic_is4(s, "\x53\x80\xF6\x34")) - return 0; - - for (i = 0; i < 84; ++i) - stbi__get8(s); - - if (!stbi__pic_is4(s, "PICT")) - return 0; - - return 1; -} - -typedef struct { - stbi_uc size, type, channel; -} stbi__pic_packet; - -static stbi_uc* stbi__readval(stbi__context* s, int channel, stbi_uc* dest) -{ - int mask = 0x80, i; - - for (i = 0; i < 4; ++i, mask >>= 1) - { - if (channel & mask) - { - if (stbi__at_eof(s)) - return stbi__errpuc("bad file", "PIC file too short"); - dest[i] = stbi__get8(s); - } - } - - return dest; -} - -static void stbi__copyval(int channel, stbi_uc* dest, const stbi_uc* src) -{ - int mask = 0x80, i; - - for (i = 0; i < 4; ++i, mask >>= 1) - if (channel & mask) - dest[i] = src[i]; -} - -static stbi_uc* stbi__pic_load_core(stbi__context* s, int width, int height, int* comp, stbi_uc* result) -{ - int act_comp = 0, num_packets = 0, y, chained; - stbi__pic_packet packets[10]; - - // this will (should...) cater for even some bizarre stuff like having data - // for the same channel in multiple packets. - do - { - stbi__pic_packet* packet; - - if (num_packets == sizeof(packets) / sizeof(packets[0])) - return stbi__errpuc("bad format", "too many packets"); - - packet = &packets[num_packets++]; - - chained = stbi__get8(s); - packet->size = stbi__get8(s); - packet->type = stbi__get8(s); - packet->channel = stbi__get8(s); - - act_comp |= packet->channel; - - if (stbi__at_eof(s)) - return stbi__errpuc("bad file", "file too short (reading packets)"); - if (packet->size != 8) - return stbi__errpuc("bad format", "packet isn't 8bpp"); - } while (chained); - - *comp = (act_comp & 0x10 ? 4 : 3); // has alpha channel? - - for (y = 0; y < height; ++y) - { - int packet_idx; - - for (packet_idx = 0; packet_idx < num_packets; ++packet_idx) - { - stbi__pic_packet* packet = &packets[packet_idx]; - stbi_uc* dest = result + y * width * 4; - - switch (packet->type) - { - default: return stbi__errpuc("bad format", "packet has bad compression type"); - - case 0: - { // uncompressed - int x; - - for (x = 0; x < width; ++x, dest += 4) - if (!stbi__readval(s, packet->channel, dest)) - return 0; - break; - } - - case 1: // Pure RLE - { - int left = width, i; - - while (left > 0) - { - stbi_uc count, value[4]; - - count = stbi__get8(s); - if (stbi__at_eof(s)) - return stbi__errpuc("bad file", "file too short (pure read count)"); - - if (count > left) - count = (stbi_uc)left; - - if (!stbi__readval(s, packet->channel, value)) - return 0; - - for (i = 0; i < count; ++i, dest += 4) - stbi__copyval(packet->channel, dest, value); - left -= count; - } - } - break; - - case 2: - { // Mixed RLE - int left = width; - while (left > 0) - { - int count = stbi__get8(s), i; - if (stbi__at_eof(s)) - return stbi__errpuc("bad file", "file too short (mixed read count)"); - - if (count >= 128) - { // Repeated - stbi_uc value[4]; - - if (count == 128) - count = stbi__get16be(s); - else - count -= 127; - if (count > left) - return stbi__errpuc("bad file", "scanline overrun"); - - if (!stbi__readval(s, packet->channel, value)) - return 0; - - for (i = 0; i < count; ++i, dest += 4) - stbi__copyval(packet->channel, dest, value); - } - else - { // Raw - ++count; - if (count > left) - return stbi__errpuc("bad file", "scanline overrun"); - - for (i = 0; i < count; ++i, dest += 4) - if (!stbi__readval(s, packet->channel, dest)) - return 0; - } - left -= count; - } - break; - } - } - } - } - - return result; -} - -static void* stbi__pic_load(stbi__context* s, int* px, int* py, int* comp, int req_comp, stbi__result_info* ri) -{ - stbi_uc* result; - int i, x, y, internal_comp; - STBI_NOTUSED(ri); - - if (!comp) - comp = &internal_comp; - - for (i = 0; i < 92; ++i) - stbi__get8(s); - - x = stbi__get16be(s); - y = stbi__get16be(s); - - if (y > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - if (x > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - - if (stbi__at_eof(s)) - return stbi__errpuc("bad file", "file too short (pic header)"); - if (!stbi__mad3sizes_valid(x, y, 4, 0)) - return stbi__errpuc("too large", "PIC image too large to decode"); - - stbi__get32be(s); // skip `ratio' - stbi__get16be(s); // skip `fields' - stbi__get16be(s); // skip `pad' - - // intermediate buffer is RGBA - result = (stbi_uc*)stbi__malloc_mad3(x, y, 4, 0); - if (!result) - return stbi__errpuc("outofmem", "Out of memory"); - memset(result, 0xff, x * y * 4); - - if (!stbi__pic_load_core(s, x, y, comp, result)) - { - STBI_FREE(result); - result = 0; - } - *px = x; - *py = y; - if (req_comp == 0) - req_comp = *comp; - result = stbi__convert_format(result, 4, req_comp, x, y); - - return result; -} - -static int stbi__pic_test(stbi__context* s) -{ - int r = stbi__pic_test_core(s); - stbi__rewind(s); - return r; -} - #endif - -// ************************************************************************************************* -// GIF loader -- public domain by Jean-Marc Lienher -- simplified/shrunk by stb - - #ifndef STBI_NO_GIF -typedef struct { - stbi__int16 prefix; - stbi_uc first; - stbi_uc suffix; -} stbi__gif_lzw; - -typedef struct { - int w, h; - stbi_uc* out; // output buffer (always 4 components) - stbi_uc* background; // The current "background" as far as a gif is concerned - stbi_uc* history; - int flags, bgindex, ratio, transparent, eflags; - stbi_uc pal[256][4]; - stbi_uc lpal[256][4]; - stbi__gif_lzw codes[8192]; - stbi_uc* color_table; - int parse, step; - int lflags; - int start_x, start_y; - int max_x, max_y; - int cur_x, cur_y; - int line_size; - int delay; -} stbi__gif; - -static int stbi__gif_test_raw(stbi__context* s) -{ - int sz; - if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') - return 0; - sz = stbi__get8(s); - if (sz != '9' && sz != '7') - return 0; - if (stbi__get8(s) != 'a') - return 0; - return 1; -} - -static int stbi__gif_test(stbi__context* s) -{ - int r = stbi__gif_test_raw(s); - stbi__rewind(s); - return r; -} - -static void stbi__gif_parse_colortable(stbi__context* s, stbi_uc pal[256][4], int num_entries, int transp) -{ - int i; - for (i = 0; i < num_entries; ++i) - { - pal[i][2] = stbi__get8(s); - pal[i][1] = stbi__get8(s); - pal[i][0] = stbi__get8(s); - pal[i][3] = transp == i ? 0 : 255; - } -} - -static int stbi__gif_header(stbi__context* s, stbi__gif* g, int* comp, int is_info) -{ - stbi_uc version; - if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') - return stbi__err("not GIF", "Corrupt GIF"); - - version = stbi__get8(s); - if (version != '7' && version != '9') - return stbi__err("not GIF", "Corrupt GIF"); - if (stbi__get8(s) != 'a') - return stbi__err("not GIF", "Corrupt GIF"); - - stbi__g_failure_reason = ""; - g->w = stbi__get16le(s); - g->h = stbi__get16le(s); - g->flags = stbi__get8(s); - g->bgindex = stbi__get8(s); - g->ratio = stbi__get8(s); - g->transparent = -1; - - if (g->w > STBI_MAX_DIMENSIONS) - return stbi__err("too large", "Very large image (corrupt?)"); - if (g->h > STBI_MAX_DIMENSIONS) - return stbi__err("too large", "Very large image (corrupt?)"); - - if (comp != 0) - *comp = 4; // can't actually tell whether it's 3 or 4 until we parse the comments - - if (is_info) - return 1; - - if (g->flags & 0x80) - stbi__gif_parse_colortable(s, g->pal, 2 << (g->flags & 7), -1); - - return 1; -} - -static int stbi__gif_info_raw(stbi__context* s, int* x, int* y, int* comp) -{ - stbi__gif* g = (stbi__gif*)stbi__malloc(sizeof(stbi__gif)); - if (!g) - return stbi__err("outofmem", "Out of memory"); - if (!stbi__gif_header(s, g, comp, 1)) - { - STBI_FREE(g); - stbi__rewind(s); - return 0; - } - if (x) - *x = g->w; - if (y) - *y = g->h; - STBI_FREE(g); - return 1; -} - -static void stbi__out_gif_code(stbi__gif* g, stbi__uint16 code) -{ - stbi_uc *p, *c; - int idx; - - // recurse to decode the prefixes, since the linked-list is backwards, - // and working backwards through an interleaved image would be nasty - if (g->codes[code].prefix >= 0) - stbi__out_gif_code(g, g->codes[code].prefix); - - if (g->cur_y >= g->max_y) - return; - - idx = g->cur_x + g->cur_y; - p = &g->out[idx]; - g->history[idx / 4] = 1; - - c = &g->color_table[g->codes[code].suffix * 4]; - if (c[3] > 128) - { // don't render transparent pixels; - p[0] = c[2]; - p[1] = c[1]; - p[2] = c[0]; - p[3] = c[3]; - } - g->cur_x += 4; - - if (g->cur_x >= g->max_x) - { - g->cur_x = g->start_x; - g->cur_y += g->step; - - while (g->cur_y >= g->max_y && g->parse > 0) - { - g->step = (1 << g->parse) * g->line_size; - g->cur_y = g->start_y + (g->step >> 1); - --g->parse; - } - } -} - -static stbi_uc* stbi__process_gif_raster(stbi__context* s, stbi__gif* g) -{ - stbi_uc lzw_cs; - stbi__int32 len, init_code; - stbi__uint32 first; - stbi__int32 codesize, codemask, avail, oldcode, bits, valid_bits, clear; - stbi__gif_lzw* p; - - lzw_cs = stbi__get8(s); - if (lzw_cs > 12) - return NULL; - clear = 1 << lzw_cs; - first = 1; - codesize = lzw_cs + 1; - codemask = (1 << codesize) - 1; - bits = 0; - valid_bits = 0; - for (init_code = 0; init_code < clear; init_code++) - { - g->codes[init_code].prefix = -1; - g->codes[init_code].first = (stbi_uc)init_code; - g->codes[init_code].suffix = (stbi_uc)init_code; - } - - // support no starting clear code - avail = clear + 2; - oldcode = -1; - - len = 0; - for (;;) - { - if (valid_bits < codesize) - { - if (len == 0) - { - len = stbi__get8(s); // start new block - if (len == 0) - return g->out; - } - --len; - bits |= (stbi__int32)stbi__get8(s) << valid_bits; - valid_bits += 8; - } - else - { - stbi__int32 code = bits & codemask; - bits >>= codesize; - valid_bits -= codesize; - // @OPTIMIZE: is there some way we can accelerate the non-clear path? - if (code == clear) - { // clear code - codesize = lzw_cs + 1; - codemask = (1 << codesize) - 1; - avail = clear + 2; - oldcode = -1; - first = 0; - } - else if (code == clear + 1) - { // end of stream code - stbi__skip(s, len); - while ((len = stbi__get8(s)) > 0) - stbi__skip(s, len); - return g->out; - } - else if (code <= avail) - { - if (first) - { - return stbi__errpuc("no clear code", "Corrupt GIF"); - } - - if (oldcode >= 0) - { - p = &g->codes[avail++]; - if (avail > 8192) - { - return stbi__errpuc("too many codes", "Corrupt GIF"); - } - - p->prefix = (stbi__int16)oldcode; - p->first = g->codes[oldcode].first; - p->suffix = (code == avail) ? p->first : g->codes[code].first; - } - else if (code == avail) - return stbi__errpuc("illegal code in raster", "Corrupt GIF"); - - stbi__out_gif_code(g, (stbi__uint16)code); - - if ((avail & codemask) == 0 && avail <= 0x0FFF) - { - codesize++; - codemask = (1 << codesize) - 1; - } - - oldcode = code; - } - else - { - return stbi__errpuc("illegal code in raster", "Corrupt GIF"); - } - } - } -} - -// this function is designed to support animated gifs, although stb_image doesn't support it -// two back is the image from two frames ago, used for a very specific disposal format -static stbi_uc* stbi__gif_load_next(stbi__context* s, stbi__gif* g, int* comp, int req_comp, stbi_uc* two_back) -{ - int dispose; - int first_frame; - int pi; - int pcount; - STBI_NOTUSED(req_comp); - - // on first frame, any non-written pixels get the background colour (non-transparent) - first_frame = 0; - if (g->out == 0) - { - if (!stbi__gif_header(s, g, comp, 0)) - return 0; // stbi__g_failure_reason set by stbi__gif_header - if (!stbi__mad3sizes_valid(4, g->w, g->h, 0)) - return stbi__errpuc("too large", "GIF image is too large"); - pcount = g->w * g->h; - g->out = (stbi_uc*)stbi__malloc(4 * pcount); - g->background = (stbi_uc*)stbi__malloc(4 * pcount); - g->history = (stbi_uc*)stbi__malloc(pcount); - if (!g->out || !g->background || !g->history) - return stbi__errpuc("outofmem", "Out of memory"); - - // image is treated as "transparent" at the start - ie, nothing overwrites the current background; - // background colour is only used for pixels that are not rendered first frame, after that "background" - // color refers to the color that was there the previous frame. - memset(g->out, 0x00, 4 * pcount); - memset(g->background, 0x00, 4 * pcount); // state of the background (starts transparent) - memset(g->history, 0x00, pcount); // pixels that were affected previous frame - first_frame = 1; - } - else - { - // second frame - how do we dispose of the previous one? - dispose = (g->eflags & 0x1C) >> 2; - pcount = g->w * g->h; - - if ((dispose == 3) && (two_back == 0)) - { - dispose = 2; // if I don't have an image to revert back to, default to the old background - } - - if (dispose == 3) - { // use previous graphic - for (pi = 0; pi < pcount; ++pi) - { - if (g->history[pi]) - { - memcpy(&g->out[pi * 4], &two_back[pi * 4], 4); - } - } - } - else if (dispose == 2) - { - // restore what was changed last frame to background before that frame; - for (pi = 0; pi < pcount; ++pi) - { - if (g->history[pi]) - { - memcpy(&g->out[pi * 4], &g->background[pi * 4], 4); - } - } - } - else - { - // This is a non-disposal case eithe way, so just - // leave the pixels as is, and they will become the new background - // 1: do not dispose - // 0: not specified. - } - - // background is what out is after the undoing of the previou frame; - memcpy(g->background, g->out, 4 * g->w * g->h); - } - - // clear my history; - memset(g->history, 0x00, g->w * g->h); // pixels that were affected previous frame - - for (;;) - { - int tag = stbi__get8(s); - switch (tag) - { - case 0x2C: /* Image Descriptor */ - { - stbi__int32 x, y, w, h; - stbi_uc* o; - - x = stbi__get16le(s); - y = stbi__get16le(s); - w = stbi__get16le(s); - h = stbi__get16le(s); - if (((x + w) > (g->w)) || ((y + h) > (g->h))) - return stbi__errpuc("bad Image Descriptor", "Corrupt GIF"); - - g->line_size = g->w * 4; - g->start_x = x * 4; - g->start_y = y * g->line_size; - g->max_x = g->start_x + w * 4; - g->max_y = g->start_y + h * g->line_size; - g->cur_x = g->start_x; - g->cur_y = g->start_y; - - // if the width of the specified rectangle is 0, that means - // we may not see *any* pixels or the image is malformed; - // to make sure this is caught, move the current y down to - // max_y (which is what out_gif_code checks). - if (w == 0) - g->cur_y = g->max_y; - - g->lflags = stbi__get8(s); - - if (g->lflags & 0x40) - { - g->step = 8 * g->line_size; // first interlaced spacing - g->parse = 3; - } - else - { - g->step = g->line_size; - g->parse = 0; - } - - if (g->lflags & 0x80) - { - stbi__gif_parse_colortable(s, g->lpal, 2 << (g->lflags & 7), g->eflags & 0x01 ? g->transparent : -1); - g->color_table = (stbi_uc*)g->lpal; - } - else if (g->flags & 0x80) - { - g->color_table = (stbi_uc*)g->pal; - } - else - return stbi__errpuc("missing color table", "Corrupt GIF"); - - o = stbi__process_gif_raster(s, g); - if (!o) - return NULL; - - // if this was the first frame, - pcount = g->w * g->h; - if (first_frame && (g->bgindex > 0)) - { - // if first frame, any pixel not drawn to gets the background color - for (pi = 0; pi < pcount; ++pi) - { - if (g->history[pi] == 0) - { - g->pal[g->bgindex][3] = 255; // just in case it was made transparent, undo that; It will be reset next frame if need be; - memcpy(&g->out[pi * 4], &g->pal[g->bgindex], 4); - } - } - } - - return o; - } - - case 0x21: // Comment Extension. - { - int len; - int ext = stbi__get8(s); - if (ext == 0xF9) - { // Graphic Control Extension. - len = stbi__get8(s); - if (len == 4) - { - g->eflags = stbi__get8(s); - g->delay = 10 * stbi__get16le(s); // delay - 1/100th of a second, saving as 1/1000ths. - - // unset old transparent - if (g->transparent >= 0) - { - g->pal[g->transparent][3] = 255; - } - if (g->eflags & 0x01) - { - g->transparent = stbi__get8(s); - if (g->transparent >= 0) - { - g->pal[g->transparent][3] = 0; - } - } - else - { - // don't need transparent - stbi__skip(s, 1); - g->transparent = -1; - } - } - else - { - stbi__skip(s, len); - break; - } - } - while ((len = stbi__get8(s)) != 0) - { - stbi__skip(s, len); - } - break; - } - - case 0x3B: // gif stream termination code - return (stbi_uc*)s; // using '1' causes warning on some compilers - - default: return stbi__errpuc("unknown code", "Corrupt GIF"); - } - } -} - -static void* stbi__load_gif_main_outofmem(stbi__gif* g, stbi_uc* out, int** delays) -{ - STBI_FREE(g->out); - STBI_FREE(g->history); - STBI_FREE(g->background); - - if (out) - STBI_FREE(out); - if (delays && *delays) - STBI_FREE(*delays); - return stbi__errpuc("outofmem", "Out of memory"); -} - -static void* stbi__load_gif_main(stbi__context* s, int** delays, int* x, int* y, int* z, int* comp, int req_comp) -{ - if (stbi__gif_test(s)) - { - int layers = 0; - stbi_uc* u = 0; - stbi_uc* out = 0; - stbi_uc* two_back = 0; - stbi__gif g; - int stride; - int out_size = 0; - int delays_size = 0; - - STBI_NOTUSED(out_size); - STBI_NOTUSED(delays_size); - - memset(&g, 0, sizeof(g)); - if (delays) - { - *delays = 0; - } - - do - { - u = stbi__gif_load_next(s, &g, comp, req_comp, two_back); - if (u == (stbi_uc*)s) - u = 0; // end of animated gif marker - - if (u) - { - *x = g.w; - *y = g.h; - ++layers; - stride = g.w * g.h * 4; - - if (out) - { - void* tmp = (stbi_uc*)STBI_REALLOC_SIZED(out, out_size, layers * stride); - if (!tmp) - return stbi__load_gif_main_outofmem(&g, out, delays); - else - { - out = (stbi_uc*)tmp; - out_size = layers * stride; - } - - if (delays) - { - int* new_delays = (int*)STBI_REALLOC_SIZED(*delays, delays_size, sizeof(int) * layers); - if (!new_delays) - return stbi__load_gif_main_outofmem(&g, out, delays); - *delays = new_delays; - delays_size = layers * sizeof(int); - } - } - else - { - out = (stbi_uc*)stbi__malloc(layers * stride); - if (!out) - return stbi__load_gif_main_outofmem(&g, out, delays); - out_size = layers * stride; - if (delays) - { - *delays = (int*)stbi__malloc(layers * sizeof(int)); - if (!*delays) - return stbi__load_gif_main_outofmem(&g, out, delays); - delays_size = layers * sizeof(int); - } - } - memcpy(out + ((layers - 1) * stride), u, stride); - if (layers >= 2) - { - two_back = out - 2 * stride; - } - - if (delays) - { - (*delays)[layers - 1U] = g.delay; - } - } - } while (u != 0); - - // free temp buffer; - STBI_FREE(g.out); - STBI_FREE(g.history); - STBI_FREE(g.background); - - // do the final conversion after loading everything; - if (req_comp && req_comp != 4) - out = stbi__convert_format(out, 4, req_comp, layers * g.w, g.h); - - *z = layers; - return out; - } - else - { - return stbi__errpuc("not GIF", "Image was not as a gif type."); - } -} - -static void* stbi__gif_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) -{ - stbi_uc* u = 0; - stbi__gif g; - memset(&g, 0, sizeof(g)); - STBI_NOTUSED(ri); - - u = stbi__gif_load_next(s, &g, comp, req_comp, 0); - if (u == (stbi_uc*)s) - u = 0; // end of animated gif marker - if (u) - { - *x = g.w; - *y = g.h; - - // moved conversion to after successful load so that the same - // can be done for multiple frames. - if (req_comp && req_comp != 4) - u = stbi__convert_format(u, 4, req_comp, g.w, g.h); - } - else if (g.out) - { - // if there was an error and we allocated an image buffer, free it! - STBI_FREE(g.out); - } - - // free buffers needed for multiple frame loading; - STBI_FREE(g.history); - STBI_FREE(g.background); - - return u; -} - -static int stbi__gif_info(stbi__context* s, int* x, int* y, int* comp) -{ - return stbi__gif_info_raw(s, x, y, comp); -} - #endif - - // ************************************************************************************************* - // Radiance RGBE HDR loader - // originally by Nicolas Schulz - #ifndef STBI_NO_HDR -static int stbi__hdr_test_core(stbi__context* s, const char* signature) -{ - int i; - for (i = 0; signature[i]; ++i) - if (stbi__get8(s) != signature[i]) - return 0; - stbi__rewind(s); - return 1; -} - -static int stbi__hdr_test(stbi__context* s) -{ - int r = stbi__hdr_test_core(s, "#?RADIANCE\n"); - stbi__rewind(s); - if (!r) - { - r = stbi__hdr_test_core(s, "#?RGBE\n"); - stbi__rewind(s); - } - return r; -} - - #define STBI__HDR_BUFLEN 1024 -static char* stbi__hdr_gettoken(stbi__context* z, char* buffer) -{ - int len = 0; - char c = '\0'; - - c = (char)stbi__get8(z); - - while (!stbi__at_eof(z) && c != '\n') - { - buffer[len++] = c; - if (len == STBI__HDR_BUFLEN - 1) - { - // flush to end of line - while (!stbi__at_eof(z) && stbi__get8(z) != '\n') - ; - break; - } - c = (char)stbi__get8(z); - } - - buffer[len] = 0; - return buffer; -} - -static void stbi__hdr_convert(float* output, stbi_uc* input, int req_comp) -{ - if (input[3] != 0) - { - float f1; - // Exponent - f1 = (float)ldexp(1.0f, input[3] - (int)(128 + 8)); - if (req_comp <= 2) - output[0] = (input[0] + input[1] + input[2]) * f1 / 3; - else - { - output[0] = input[0] * f1; - output[1] = input[1] * f1; - output[2] = input[2] * f1; - } - if (req_comp == 2) - output[1] = 1; - if (req_comp == 4) - output[3] = 1; - } - else - { - switch (req_comp) - { - case 4: output[3] = 1; /* fallthrough */ - case 3: output[0] = output[1] = output[2] = 0; break; - case 2: output[1] = 1; /* fallthrough */ - case 1: output[0] = 0; break; - } - } -} - -static float* stbi__hdr_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) -{ - char buffer[STBI__HDR_BUFLEN]; - char* token; - int valid = 0; - int width, height; - stbi_uc* scanline; - float* hdr_data; - int len; - unsigned char count, value; - int i, j, k, c1, c2, z; - const char* headerToken; - STBI_NOTUSED(ri); - - // Check identifier - headerToken = stbi__hdr_gettoken(s, buffer); - if (strcmp(headerToken, "#?RADIANCE") != 0 && strcmp(headerToken, "#?RGBE") != 0) - return stbi__errpf("not HDR", "Corrupt HDR image"); - - // Parse header - for (;;) - { - token = stbi__hdr_gettoken(s, buffer); - if (token[0] == 0) - break; - if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) - valid = 1; - } - - if (!valid) - return stbi__errpf("unsupported format", "Unsupported HDR format"); - - // Parse width and height - // can't use sscanf() if we're not using stdio! - token = stbi__hdr_gettoken(s, buffer); - if (strncmp(token, "-Y ", 3)) - return stbi__errpf("unsupported data layout", "Unsupported HDR format"); - token += 3; - height = (int)strtol(token, &token, 10); - while (*token == ' ') - ++token; - if (strncmp(token, "+X ", 3)) - return stbi__errpf("unsupported data layout", "Unsupported HDR format"); - token += 3; - width = (int)strtol(token, NULL, 10); - - if (height > STBI_MAX_DIMENSIONS) - return stbi__errpf("too large", "Very large image (corrupt?)"); - if (width > STBI_MAX_DIMENSIONS) - return stbi__errpf("too large", "Very large image (corrupt?)"); - - *x = width; - *y = height; - - if (comp) - *comp = 3; - if (req_comp == 0) - req_comp = 3; - - if (!stbi__mad4sizes_valid(width, height, req_comp, sizeof(float), 0)) - return stbi__errpf("too large", "HDR image is too large"); - - // Read data - hdr_data = (float*)stbi__malloc_mad4(width, height, req_comp, sizeof(float), 0); - if (!hdr_data) - return stbi__errpf("outofmem", "Out of memory"); - - // Load image data - // image data is stored as some number of sca - if (width < 8 || width >= 32768) - { - // Read flat data - for (j = 0; j < height; ++j) - { - for (i = 0; i < width; ++i) - { - stbi_uc rgbe[4]; - main_decode_loop: - stbi__getn(s, rgbe, 4); - stbi__hdr_convert(hdr_data + j * width * req_comp + i * req_comp, rgbe, req_comp); - } - } - } - else - { - // Read RLE-encoded data - scanline = NULL; - - for (j = 0; j < height; ++j) - { - c1 = stbi__get8(s); - c2 = stbi__get8(s); - len = stbi__get8(s); - if (c1 != 2 || c2 != 2 || (len & 0x80)) - { - // not run-length encoded, so we have to actually use THIS data as a decoded - // pixel (note this can't be a valid pixel--one of RGB must be >= 128) - stbi_uc rgbe[4]; - rgbe[0] = (stbi_uc)c1; - rgbe[1] = (stbi_uc)c2; - rgbe[2] = (stbi_uc)len; - rgbe[3] = (stbi_uc)stbi__get8(s); - stbi__hdr_convert(hdr_data, rgbe, req_comp); - i = 1; - j = 0; - STBI_FREE(scanline); - goto main_decode_loop; // yes, this makes no sense - } - len <<= 8; - len |= stbi__get8(s); - if (len != width) - { - STBI_FREE(hdr_data); - STBI_FREE(scanline); - return stbi__errpf("invalid decoded scanline length", "corrupt HDR"); - } - if (scanline == NULL) - { - scanline = (stbi_uc*)stbi__malloc_mad2(width, 4, 0); - if (!scanline) - { - STBI_FREE(hdr_data); - return stbi__errpf("outofmem", "Out of memory"); - } - } - - for (k = 0; k < 4; ++k) - { - int nleft; - i = 0; - while ((nleft = width - i) > 0) - { - count = stbi__get8(s); - if (count > 128) - { - // Run - value = stbi__get8(s); - count -= 128; - if ((count == 0) || (count > nleft)) - { - STBI_FREE(hdr_data); - STBI_FREE(scanline); - return stbi__errpf("corrupt", "bad RLE data in HDR"); - } - for (z = 0; z < count; ++z) - scanline[i++ * 4 + k] = value; - } - else - { - // Dump - if ((count == 0) || (count > nleft)) - { - STBI_FREE(hdr_data); - STBI_FREE(scanline); - return stbi__errpf("corrupt", "bad RLE data in HDR"); - } - for (z = 0; z < count; ++z) - scanline[i++ * 4 + k] = stbi__get8(s); - } - } - } - for (i = 0; i < width; ++i) - stbi__hdr_convert(hdr_data + (j * width + i) * req_comp, scanline + i * 4, req_comp); - } - if (scanline) - STBI_FREE(scanline); - } - - return hdr_data; -} - -static int stbi__hdr_info(stbi__context* s, int* x, int* y, int* comp) -{ - char buffer[STBI__HDR_BUFLEN]; - char* token; - int valid = 0; - int dummy; - - if (!x) - x = &dummy; - if (!y) - y = &dummy; - if (!comp) - comp = &dummy; - - if (stbi__hdr_test(s) == 0) - { - stbi__rewind(s); - return 0; - } - - for (;;) - { - token = stbi__hdr_gettoken(s, buffer); - if (token[0] == 0) - break; - if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) - valid = 1; - } - - if (!valid) - { - stbi__rewind(s); - return 0; - } - token = stbi__hdr_gettoken(s, buffer); - if (strncmp(token, "-Y ", 3)) - { - stbi__rewind(s); - return 0; - } - token += 3; - *y = (int)strtol(token, &token, 10); - while (*token == ' ') - ++token; - if (strncmp(token, "+X ", 3)) - { - stbi__rewind(s); - return 0; - } - token += 3; - *x = (int)strtol(token, NULL, 10); - *comp = 3; - return 1; -} - #endif // STBI_NO_HDR - - #ifndef STBI_NO_BMP -static int stbi__bmp_info(stbi__context* s, int* x, int* y, int* comp) -{ - void* p; - stbi__bmp_data info; - - info.all_a = 255; - p = stbi__bmp_parse_header(s, &info); - if (p == NULL) - { - stbi__rewind(s); - return 0; - } - if (x) - *x = s->img_x; - if (y) - *y = s->img_y; - if (comp) - { - if (info.bpp == 24 && info.ma == 0xff000000) - *comp = 3; - else - *comp = info.ma ? 4 : 3; - } - return 1; -} - #endif - - #ifndef STBI_NO_PSD -static int stbi__psd_info(stbi__context* s, int* x, int* y, int* comp) -{ - int channelCount, dummy, depth; - if (!x) - x = &dummy; - if (!y) - y = &dummy; - if (!comp) - comp = &dummy; - if (stbi__get32be(s) != 0x38425053) - { - stbi__rewind(s); - return 0; - } - if (stbi__get16be(s) != 1) - { - stbi__rewind(s); - return 0; - } - stbi__skip(s, 6); - channelCount = stbi__get16be(s); - if (channelCount < 0 || channelCount > 16) - { - stbi__rewind(s); - return 0; - } - *y = stbi__get32be(s); - *x = stbi__get32be(s); - depth = stbi__get16be(s); - if (depth != 8 && depth != 16) - { - stbi__rewind(s); - return 0; - } - if (stbi__get16be(s) != 3) - { - stbi__rewind(s); - return 0; - } - *comp = 4; - return 1; -} - -static int stbi__psd_is16(stbi__context* s) -{ - int channelCount, depth; - if (stbi__get32be(s) != 0x38425053) - { - stbi__rewind(s); - return 0; - } - if (stbi__get16be(s) != 1) - { - stbi__rewind(s); - return 0; - } - stbi__skip(s, 6); - channelCount = stbi__get16be(s); - if (channelCount < 0 || channelCount > 16) - { - stbi__rewind(s); - return 0; - } - STBI_NOTUSED(stbi__get32be(s)); - STBI_NOTUSED(stbi__get32be(s)); - depth = stbi__get16be(s); - if (depth != 16) - { - stbi__rewind(s); - return 0; - } - return 1; -} - #endif - - #ifndef STBI_NO_PIC -static int stbi__pic_info(stbi__context* s, int* x, int* y, int* comp) -{ - int act_comp = 0, num_packets = 0, chained, dummy; - stbi__pic_packet packets[10]; - - if (!x) - x = &dummy; - if (!y) - y = &dummy; - if (!comp) - comp = &dummy; - - if (!stbi__pic_is4(s, "\x53\x80\xF6\x34")) - { - stbi__rewind(s); - return 0; - } - - stbi__skip(s, 88); - - *x = stbi__get16be(s); - *y = stbi__get16be(s); - if (stbi__at_eof(s)) - { - stbi__rewind(s); - return 0; - } - if ((*x) != 0 && (1 << 28) / (*x) < (*y)) - { - stbi__rewind(s); - return 0; - } - - stbi__skip(s, 8); - - do - { - stbi__pic_packet* packet; - - if (num_packets == sizeof(packets) / sizeof(packets[0])) - return 0; - - packet = &packets[num_packets++]; - chained = stbi__get8(s); - packet->size = stbi__get8(s); - packet->type = stbi__get8(s); - packet->channel = stbi__get8(s); - act_comp |= packet->channel; - - if (stbi__at_eof(s)) - { - stbi__rewind(s); - return 0; - } - if (packet->size != 8) - { - stbi__rewind(s); - return 0; - } - } while (chained); - - *comp = (act_comp & 0x10 ? 4 : 3); - - return 1; -} - #endif - -// ************************************************************************************************* -// Portable Gray Map and Portable Pixel Map loader -// by Ken Miller -// -// PGM: http://netpbm.sourceforge.net/doc/pgm.html -// PPM: http://netpbm.sourceforge.net/doc/ppm.html -// -// Known limitations: -// Does not support comments in the header section -// Does not support ASCII image data (formats P2 and P3) - - #ifndef STBI_NO_PNM - -static int stbi__pnm_test(stbi__context* s) -{ - char p, t; - p = (char)stbi__get8(s); - t = (char)stbi__get8(s); - if (p != 'P' || (t != '5' && t != '6')) - { - stbi__rewind(s); - return 0; - } - return 1; -} - -static void* stbi__pnm_load(stbi__context* s, int* x, int* y, int* comp, int req_comp, stbi__result_info* ri) -{ - stbi_uc* out; - STBI_NOTUSED(ri); - - ri->bits_per_channel = stbi__pnm_info(s, (int*)&s->img_x, (int*)&s->img_y, (int*)&s->img_n); - if (ri->bits_per_channel == 0) - return 0; - - if (s->img_y > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - if (s->img_x > STBI_MAX_DIMENSIONS) - return stbi__errpuc("too large", "Very large image (corrupt?)"); - - *x = s->img_x; - *y = s->img_y; - if (comp) - *comp = s->img_n; - - if (!stbi__mad4sizes_valid(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0)) - return stbi__errpuc("too large", "PNM too large"); - - out = (stbi_uc*)stbi__malloc_mad4(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0); - if (!out) - return stbi__errpuc("outofmem", "Out of memory"); - if (!stbi__getn(s, out, s->img_n * s->img_x * s->img_y * (ri->bits_per_channel / 8))) - { - STBI_FREE(out); - return stbi__errpuc("bad PNM", "PNM file truncated"); - } - - if (req_comp && req_comp != s->img_n) - { - if (ri->bits_per_channel == 16) - { - out = (stbi_uc*)stbi__convert_format16((stbi__uint16*)out, s->img_n, req_comp, s->img_x, s->img_y); - } - else - { - out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y); - } - if (out == NULL) - return out; // stbi__convert_format frees input on failure - } - return out; -} - -static int stbi__pnm_isspace(char c) -{ - return c == ' ' || c == '\t' || c == '\n' || c == '\v' || c == '\f' || c == '\r'; -} - -static void stbi__pnm_skip_whitespace(stbi__context* s, char* c) -{ - for (;;) - { - while (!stbi__at_eof(s) && stbi__pnm_isspace(*c)) - *c = (char)stbi__get8(s); - - if (stbi__at_eof(s) || *c != '#') - break; - - while (!stbi__at_eof(s) && *c != '\n' && *c != '\r') - *c = (char)stbi__get8(s); - } -} - -static int stbi__pnm_isdigit(char c) -{ - return c >= '0' && c <= '9'; -} - -static int stbi__pnm_getinteger(stbi__context* s, char* c) -{ - int value = 0; - - while (!stbi__at_eof(s) && stbi__pnm_isdigit(*c)) - { - value = value * 10 + (*c - '0'); - *c = (char)stbi__get8(s); - if ((value > 214748364) || (value == 214748364 && *c > '7')) - return stbi__err("integer parse overflow", "Parsing an integer in the PPM header overflowed a 32-bit int"); - } - - return value; -} - -static int stbi__pnm_info(stbi__context* s, int* x, int* y, int* comp) -{ - int maxv, dummy; - char c, p, t; - - if (!x) - x = &dummy; - if (!y) - y = &dummy; - if (!comp) - comp = &dummy; - - stbi__rewind(s); - - // Get identifier - p = (char)stbi__get8(s); - t = (char)stbi__get8(s); - if (p != 'P' || (t != '5' && t != '6')) - { - stbi__rewind(s); - return 0; - } - - *comp = (t == '6') ? 3 : 1; // '5' is 1-component .pgm; '6' is 3-component .ppm - - c = (char)stbi__get8(s); - stbi__pnm_skip_whitespace(s, &c); - - *x = stbi__pnm_getinteger(s, &c); // read width - if (*x == 0) - return stbi__err("invalid width", "PPM image header had zero or overflowing width"); - stbi__pnm_skip_whitespace(s, &c); - - *y = stbi__pnm_getinteger(s, &c); // read height - if (*y == 0) - return stbi__err("invalid width", "PPM image header had zero or overflowing width"); - stbi__pnm_skip_whitespace(s, &c); - - maxv = stbi__pnm_getinteger(s, &c); // read max value - if (maxv > 65535) - return stbi__err("max value > 65535", "PPM image supports only 8-bit and 16-bit images"); - else if (maxv > 255) - return 16; - else - return 8; -} - -static int stbi__pnm_is16(stbi__context* s) -{ - if (stbi__pnm_info(s, NULL, NULL, NULL) == 16) - return 1; - return 0; -} - #endif - -static int stbi__info_main(stbi__context* s, int* x, int* y, int* comp) -{ - #ifndef STBI_NO_JPEG - if (stbi__jpeg_info(s, x, y, comp)) - return 1; - #endif - - #ifndef STBI_NO_PNG - if (stbi__png_info(s, x, y, comp)) - return 1; - #endif - - #ifndef STBI_NO_GIF - if (stbi__gif_info(s, x, y, comp)) - return 1; - #endif - - #ifndef STBI_NO_BMP - if (stbi__bmp_info(s, x, y, comp)) - return 1; - #endif - - #ifndef STBI_NO_PSD - if (stbi__psd_info(s, x, y, comp)) - return 1; - #endif - - #ifndef STBI_NO_PIC - if (stbi__pic_info(s, x, y, comp)) - return 1; - #endif - - #ifndef STBI_NO_PNM - if (stbi__pnm_info(s, x, y, comp)) - return 1; - #endif - - #ifndef STBI_NO_HDR - if (stbi__hdr_info(s, x, y, comp)) - return 1; - #endif - - // test tga last because it's a crappy test! - #ifndef STBI_NO_TGA - if (stbi__tga_info(s, x, y, comp)) - return 1; - #endif - return stbi__err("unknown image type", "Image not of any known type, or corrupt"); -} - -static int stbi__is_16_main(stbi__context* s) -{ - #ifndef STBI_NO_PNG - if (stbi__png_is16(s)) - return 1; - #endif - - #ifndef STBI_NO_PSD - if (stbi__psd_is16(s)) - return 1; - #endif - - #ifndef STBI_NO_PNM - if (stbi__pnm_is16(s)) - return 1; - #endif - return 0; -} - - #ifndef STBI_NO_STDIO -STBIDEF int stbi_info(const char* filename, int* x, int* y, int* comp) -{ - FILE* f = stbi__fopen(filename, "rb"); - int result; - if (!f) - return stbi__err("can't fopen", "Unable to open file"); - result = stbi_info_from_file(f, x, y, comp); - fclose(f); - return result; -} - -STBIDEF int stbi_info_from_file(FILE* f, int* x, int* y, int* comp) -{ - int r; - stbi__context s; - long pos = ftell(f); - stbi__start_file(&s, f); - r = stbi__info_main(&s, x, y, comp); - fseek(f, pos, SEEK_SET); - return r; -} - -STBIDEF int stbi_is_16_bit(const char* filename) -{ - FILE* f = stbi__fopen(filename, "rb"); - int result; - if (!f) - return stbi__err("can't fopen", "Unable to open file"); - result = stbi_is_16_bit_from_file(f); - fclose(f); - return result; -} - -STBIDEF int stbi_is_16_bit_from_file(FILE* f) -{ - int r; - stbi__context s; - long pos = ftell(f); - stbi__start_file(&s, f); - r = stbi__is_16_main(&s); - fseek(f, pos, SEEK_SET); - return r; -} - #endif // !STBI_NO_STDIO - -STBIDEF int stbi_info_from_memory(const stbi_uc* buffer, int len, int* x, int* y, int* comp) -{ - stbi__context s; - stbi__start_mem(&s, buffer, len); - return stbi__info_main(&s, x, y, comp); -} - -STBIDEF int stbi_info_from_callbacks(const stbi_io_callbacks* c, void* user, int* x, int* y, int* comp) -{ - stbi__context s; - stbi__start_callbacks(&s, (stbi_io_callbacks*)c, user); - return stbi__info_main(&s, x, y, comp); -} - -STBIDEF int stbi_is_16_bit_from_memory(const stbi_uc* buffer, int len) -{ - stbi__context s; - stbi__start_mem(&s, buffer, len); - return stbi__is_16_main(&s); -} - -STBIDEF int stbi_is_16_bit_from_callbacks(const stbi_io_callbacks* c, void* user) -{ - stbi__context s; - stbi__start_callbacks(&s, (stbi_io_callbacks*)c, user); - return stbi__is_16_main(&s); -} - -#endif // STB_IMAGE_IMPLEMENTATION - - /* - revision history: - 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs - 2.19 (2018-02-11) fix warning - 2.18 (2018-01-30) fix warnings - 2.17 (2018-01-29) change sbti__shiftsigned to avoid clang -O2 bug - 1-bit BMP - *_is_16_bit api - avoid warnings - 2.16 (2017-07-23) all functions have 16-bit variants; - STBI_NO_STDIO works again; - compilation fixes; - fix rounding in unpremultiply; - optimize vertical flip; - disable raw_len validation; - documentation fixes - 2.15 (2017-03-18) fix png-1,2,4 bug; now all Imagenet JPGs decode; - warning fixes; disable run-time SSE detection on gcc; - uniform handling of optional "return" values; - thread-safe initialization of zlib tables - 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs - 2.13 (2016-11-29) add 16-bit API, only supported for PNG right now - 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes - 2.11 (2016-04-02) allocate large structures on the stack - remove white matting for transparent PSD - fix reported channel count for PNG & BMP - re-enable SSE2 in non-gcc 64-bit - support RGB-formatted JPEG - read 16-bit PNGs (only as 8-bit) - 2.10 (2016-01-22) avoid warning introduced in 2.09 by STBI_REALLOC_SIZED - 2.09 (2016-01-16) allow comments in PNM files - 16-bit-per-pixel TGA (not bit-per-component) - info() for TGA could break due to .hdr handling - info() for BMP to shares code instead of sloppy parse - can use STBI_REALLOC_SIZED if allocator doesn't support realloc - code cleanup - 2.08 (2015-09-13) fix to 2.07 cleanup, reading RGB PSD as RGBA - 2.07 (2015-09-13) fix compiler warnings - partial animated GIF support - limited 16-bpc PSD support - #ifdef unused functions - bug with < 92 byte PIC,PNM,HDR,TGA - 2.06 (2015-04-19) fix bug where PSD returns wrong '*comp' value - 2.05 (2015-04-19) fix bug in progressive JPEG handling, fix warning - 2.04 (2015-04-15) try to re-enable SIMD on MinGW 64-bit - 2.03 (2015-04-12) extra corruption checking (mmozeiko) - stbi_set_flip_vertically_on_load (nguillemot) - fix NEON support; fix mingw support - 2.02 (2015-01-19) fix incorrect assert, fix warning - 2.01 (2015-01-17) fix various warnings; suppress SIMD on gcc 32-bit without -msse2 - 2.00b (2014-12-25) fix STBI_MALLOC in progressive JPEG - 2.00 (2014-12-25) optimize JPG, including x86 SSE2 & NEON SIMD (ryg) - progressive JPEG (stb) - PGM/PPM support (Ken Miller) - STBI_MALLOC,STBI_REALLOC,STBI_FREE - GIF bugfix -- seemingly never worked - STBI_NO_*, STBI_ONLY_* - 1.48 (2014-12-14) fix incorrectly-named assert() - 1.47 (2014-12-14) 1/2/4-bit PNG support, both direct and paletted (Omar Cornut & stb) - optimize PNG (ryg) - fix bug in interlaced PNG with user-specified channel count (stb) - 1.46 (2014-08-26) - fix broken tRNS chunk (colorkey-style transparency) in non-paletted PNG - 1.45 (2014-08-16) - fix MSVC-ARM internal compiler error by wrapping malloc - 1.44 (2014-08-07) - various warning fixes from Ronny Chevalier - 1.43 (2014-07-15) - fix MSVC-only compiler problem in code changed in 1.42 - 1.42 (2014-07-09) - don't define _CRT_SECURE_NO_WARNINGS (affects user code) - fixes to stbi__cleanup_jpeg path - added STBI_ASSERT to avoid requiring assert.h - 1.41 (2014-06-25) - fix search&replace from 1.36 that messed up comments/error messages - 1.40 (2014-06-22) - fix gcc struct-initialization warning - 1.39 (2014-06-15) - fix to TGA optimization when req_comp != number of components in TGA; - fix to GIF loading because BMP wasn't rewinding (whoops, no GIFs in my test suite) - add support for BMP version 5 (more ignored fields) - 1.38 (2014-06-06) - suppress MSVC warnings on integer casts truncating values - fix accidental rename of 'skip' field of I/O - 1.37 (2014-06-04) - remove duplicate typedef - 1.36 (2014-06-03) - convert to header file single-file library - if de-iphone isn't set, load iphone images color-swapped instead of returning NULL - 1.35 (2014-05-27) - various warnings - fix broken STBI_SIMD path - fix bug where stbi_load_from_file no longer left file pointer in correct place - fix broken non-easy path for 32-bit BMP (possibly never used) - TGA optimization by Arseny Kapoulkine - 1.34 (unknown) - use STBI_NOTUSED in stbi__resample_row_generic(), fix one more leak in tga failure case - 1.33 (2011-07-14) - make stbi_is_hdr work in STBI_NO_HDR (as specified), minor compiler-friendly improvements - 1.32 (2011-07-13) - support for "info" function for all supported filetypes (SpartanJ) - 1.31 (2011-06-20) - a few more leak fixes, bug in PNG handling (SpartanJ) - 1.30 (2011-06-11) - added ability to load files via callbacks to accomidate custom input streams (Ben Wenger) - removed deprecated format-specific test/load functions - removed support for installable file formats (stbi_loader) -- would have been broken for IO callbacks anyway - error cases in bmp and tga give messages and don't leak (Raymond Barbiero, grisha) - fix inefficiency in decoding 32-bit BMP (David Woo) - 1.29 (2010-08-16) - various warning fixes from Aurelien Pocheville - 1.28 (2010-08-01) - fix bug in GIF palette transparency (SpartanJ) - 1.27 (2010-08-01) - cast-to-stbi_uc to fix warnings - 1.26 (2010-07-24) - fix bug in file buffering for PNG reported by SpartanJ - 1.25 (2010-07-17) - refix trans_data warning (Won Chun) - 1.24 (2010-07-12) - perf improvements reading from files on platforms with lock-heavy fgetc() - minor perf improvements for jpeg - deprecated type-specific functions so we'll get feedback if they're needed - attempt to fix trans_data warning (Won Chun) - 1.23 fixed bug in iPhone support - 1.22 (2010-07-10) - removed image *writing* support - stbi_info support from Jetro Lauha - GIF support from Jean-Marc Lienher - iPhone PNG-extensions from James Brown - warning-fixes from Nicolas Schulz and Janez Zemva (i.stbi__err. Janez (U+017D)emva) - 1.21 fix use of 'stbi_uc' in header (reported by jon blow) - 1.20 added support for Softimage PIC, by Tom Seddon - 1.19 bug in interlaced PNG corruption check (found by ryg) - 1.18 (2008-08-02) - fix a threading bug (local mutable static) - 1.17 support interlaced PNG - 1.16 major bugfix - stbi__convert_format converted one too many pixels - 1.15 initialize some fields for thread safety - 1.14 fix threadsafe conversion bug - header-file-only version (#define STBI_HEADER_FILE_ONLY before including) - 1.13 threadsafe - 1.12 const qualifiers in the API - 1.11 Support installable IDCT, colorspace conversion routines - 1.10 Fixes for 64-bit (don't use "unsigned long") - optimized upsampling by Fabian "ryg" Giesen - 1.09 Fix format-conversion for PSD code (bad global variables!) - 1.08 Thatcher Ulrich's PSD code integrated by Nicolas Schulz - 1.07 attempt to fix C++ warning/errors again - 1.06 attempt to fix C++ warning/errors again - 1.05 fix TGA loading to return correct *comp and use good luminance calc - 1.04 default float alpha is 1, not 255; use 'void *' for stbi_image_free - 1.03 bugfixes to STBI_NO_STDIO, STBI_NO_HDR - 1.02 support for (subset of) HDR files, float interface for preferred access to them - 1.01 fix bug: possible bug in handling right-side up bmps... not sure - fix bug: the stbi__bmp_load() and stbi__tga_load() functions didn't work at all - 1.00 interface to zlib that skips zlib header - 0.99 correct handling of alpha in palette - 0.98 TGA loader by lonesock; dynamically add loaders (untested) - 0.97 jpeg errors on too large a file; also catch another malloc failure - 0.96 fix detection of invalid v value - particleman@mollyrocket forum - 0.95 during header scan, seek to markers in case of padding - 0.94 STBI_NO_STDIO to disable stdio usage; rename all #defines the same - 0.93 handle jpegtran output; verbose errors - 0.92 read 4,8,16,24,32-bit BMP files of several formats - 0.91 output 24-bit Windows 3.0 BMP files - 0.90 fix a few more warnings; bump version number to approach 1.0 - 0.61 bugfixes due to Marc LeBlanc, Christopher Lloyd - 0.60 fix compiling as c++ - 0.59 fix warnings: merge Dave Moore's -Wall fixes - 0.58 fix bug: zlib uncompressed mode len/nlen was wrong endian - 0.57 fix bug: jpg last huffman symbol before marker was >9 bits but less than 16 available - 0.56 fix bug: zlib uncompressed mode len vs. nlen - 0.55 fix bug: restart_interval not initialized to 0 - 0.54 allow NULL for 'int *comp' - 0.53 fix bug in png 3->4; speedup png decoding - 0.52 png handles req_comp=3,4 directly; minor cleanup; jpeg comments - 0.51 obey req_comp requests, 1-component jpegs return as 1-component, - on 'test' only check type, not whether we support this variant - 0.50 (2006-11-19) - first released version - */ - -/* ------------------------------------------------------------------------------- -This software is available under 2 licenses -- choose whichever you prefer. ------------------------------------------------------------------------------- -ALTERNATIVE A - MIT License -Copyright (c) 2017 Sean Barrett -Permission is hereby granted, free of charge, to any person obtaining a copy of -this software and associated documentation files (the "Software"), to deal in -the Software without restriction, including without limitation the rights to -use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies -of the Software, and to permit persons to whom the Software is furnished to do -so, subject to the following conditions: -The above copyright notice and this permission notice shall be included in all -copies or substantial portions of the Software. -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE -SOFTWARE. ------------------------------------------------------------------------------- -ALTERNATIVE B - Public Domain (www.unlicense.org) -This is free and unencumbered software released into the public domain. -Anyone is free to copy, modify, publish, use, compile, sell, or distribute this -software, either in source code form or as a compiled binary, for any purpose, -commercial or non-commercial, and by any means. -In jurisdictions that recognize copyright laws, the author or authors of this -software dedicate any and all copyright interest in the software to the public -domain. We make this dedication for the benefit of the public at large and to -the detriment of our heirs and successors. We intend this dedication to be an -overt act of relinquishment in perpetuity of all present and future rights to -this software under copyright law. -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN -ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION -WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. ------------------------------------------------------------------------------- -*/ \ No newline at end of file From e5549169d8c9acd71a275ca08c4b5596b423f5b8 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 1 Oct 2024 12:26:28 +0200 Subject: [PATCH 089/125] (fix): Correct scissor being set incorrectly --- Backends/RmlUi_Renderer_DX11.cpp | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 01999a24a..002b867d8 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1750,14 +1750,16 @@ void RenderInterface_DX11::SetScissor(Rml::Rectanglei region, bool vertically_fl if (region.Valid() && region != m_scissor_state) { // Some render APIs don't like offscreen positions (WebGL in particular), so clamp them to the viewport. - const int x = Rml::Math::Clamp(region.Left(), 0, m_viewport_width); - const int y = Rml::Math::Clamp(m_viewport_height - region.Bottom(), 0, m_viewport_height); + const int x_min = Rml::Math::Clamp(region.Left(), 0, m_viewport_width); + const int y_min = Rml::Math::Clamp(region.Top(), 0, m_viewport_height); + const int x_max = Rml::Math::Clamp(region.Right(), 0, m_viewport_width); + const int y_max = Rml::Math::Clamp(region.Bottom(), 0, m_viewport_height); D3D11_RECT rect_scissor = {}; - rect_scissor.left = x; - rect_scissor.top = y; - rect_scissor.right = x + region.Width(); - rect_scissor.bottom = y + region.Height(); + rect_scissor.left = x_min; + rect_scissor.top = y_min; + rect_scissor.right = x_max; + rect_scissor.bottom = y_max; m_d3d_context->RSSetScissorRects(1, &rect_scissor); } From 8188086d925cdf20cfa8ebd0d4983dadf55b36f9 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 1 Oct 2024 13:29:30 +0200 Subject: [PATCH 090/125] (fix): Correct blit shader having incorrect maths --- Backends/RmlUi_Renderer_DX11.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 002b867d8..07c4b983b 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -504,7 +504,7 @@ PS_Input VSMain(const VS_Input IN) result.position.xy = result.position.xy * 0.5f + 0.5f; // Remap to 0-+1 result.position.x = lerp(m_destination_min.x, m_destination_max.x, result.position.x); result.position.y = lerp(m_destination_min.y, m_destination_max.y, result.position.y); - result.position.xy = (result.position.xy - 0.5f) * 2.0f; // Remap to -1-+1 + result.position.xy = result.position.xy * 2.0f - 1.0; // Remap to -1-+1 return result; }; From 3104f37025db5fc6fc9a83d9856055afcd0f9fe6 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 1 Oct 2024 13:31:23 +0200 Subject: [PATCH 091/125] (chore): Simplify binding render targets --- Backends/RmlUi_Renderer_DX11.cpp | 51 ++++++++++++++++---------------- 1 file changed, 26 insertions(+), 25 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 07c4b983b..45e593dca 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -638,7 +638,7 @@ struct ProgramData { }; // Create the shader, 'shader_type' is either GL_VERTEX_SHADER or GL_FRAGMENT_SHADER. -static bool CreateShader(ID3D11Device* p_device, ID3DBlob*& out_shader_dxil, ShaderType shader_type, const char* code_string, const size_t code_length) +static bool CreateShader(ID3DBlob*& out_shader_dxil, ShaderType shader_type, const char* code_string, const size_t code_length) { RMLUI_ASSERT(shader_type == ShaderType::Vertex || shader_type == ShaderType::Fragment); @@ -668,7 +668,7 @@ static bool CreateShader(ID3D11Device* p_device, ID3DBlob*& out_shader_dxil, Sha return true; } -static bool CreateProgram(ID3D11Device* p_device, ShaderProgram& out_program, ProgramId program_id, ID3DBlob* vertex_shader, ID3DBlob* fragment_shader) +static bool CreateProgram(ID3D11Device* p_device, ShaderProgram& out_program, ID3DBlob* vertex_shader, ID3DBlob* fragment_shader) { RMLUI_ASSERT(vertex_shader); RMLUI_ASSERT(fragment_shader); @@ -840,11 +840,15 @@ static void DestroyRenderTarget(RenderTargetData& rt) rt = {}; } -static void BindTexture(ID3D11DeviceContext* context, const RenderTargetData& rt, uint32_t slot = 0) { context->PSSetShaderResources(slot, 1, &rt.render_target_shader_resource_view); } +static inline void BindRenderTarget(ID3D11DeviceContext* context, const RenderTargetData& rt) +{ + context->OMSetRenderTargets(1, &rt.render_target_view, rt.depth_stencil_view); +} + static bool CreateShaders(ID3D11Device* p_device, ProgramData& data) { RMLUI_ASSERT(std::all_of(data.vert_shaders.begin(), data.vert_shaders.end(), [](auto&& value) { return value == nullptr; })); @@ -857,19 +861,18 @@ static bool CreateShaders(ID3D11Device* p_device, ProgramData& data) for (const VertShaderDefinition& def : vert_shader_definitions) { - if (!CreateShader(p_device, data.vert_shaders[def.id], ShaderType::Vertex, def.code_str, def.code_size)) + if (!CreateShader(data.vert_shaders[def.id], ShaderType::Vertex, def.code_str, def.code_size)) return ReportError("vertex shader", def.name_str); } for (const FragShaderDefinition& def : frag_shader_definitions) { - if (!CreateShader(p_device, data.frag_shaders[def.id], ShaderType::Fragment, def.code_str, def.code_size)) + if (!CreateShader(data.frag_shaders[def.id], ShaderType::Fragment, def.code_str, def.code_size)) return ReportError("fragment shader", def.name_str); } for (const ProgramDefinition& def : program_definitions) { - if (!CreateProgram(p_device, data.programs[def.id], def.id, data.vert_shaders[def.vert_shader], data.frag_shaders[def.frag_shader])) return ReportError("program", def.name_str); } @@ -1241,7 +1244,7 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT SetTransform(nullptr); m_render_layers.BeginFrame(m_viewport_width, m_viewport_height); - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); float clearColor[4] = {0.0f, 0.0f, 0.0f, 1.0f}; m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, clearColor); @@ -1918,7 +1921,7 @@ Rml::LayerHandle RenderInterface_DX11::PushLayer() m_debug->BeginEvent(L"BeginLayer"); const Rml::LayerHandle layer_handle = m_render_layers.PushLayer(); - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); float colors[4] = {0, 0, 0, 0}; m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, colors); @@ -1940,8 +1943,7 @@ void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml:: RenderFilters(filters); // Render to the destination layer. - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetLayer(destination_handle).render_target_view, - m_render_layers.GetLayer(destination_handle).depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetLayer(destination_handle)); Gfx::BindTexture(m_d3d_context, m_render_layers.GetPostprocessPrimary()); UseProgram(ProgramId::Passthrough); @@ -1955,14 +1957,14 @@ void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml:: SetBlendState(m_blend_state_enable); if (destination_handle != m_render_layers.GetTopLayerHandle()) - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); } void RenderInterface_DX11::PopLayer() { m_render_layers.PopLayer(); m_debug->EndEvent(); - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); } Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() @@ -2013,7 +2015,7 @@ Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() // Restore state and free memory SetScissor(bounds); - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); texture_resource->Release(); return render_texture; @@ -2045,7 +2047,7 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); const Gfx::RenderTargetData& destination = m_render_layers.GetBlendMask(); - m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, destination); Gfx::BindTexture(m_d3d_context, source); UseProgram(ProgramId::Passthrough); ID3D11BlendState* blend_state_backup = m_current_blend_state; @@ -2054,7 +2056,7 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() DrawFullscreenQuad(); SetBlendState(blend_state_backup); - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); CompiledFilter filter = {}; filter.type = FilterType::MaskImage; @@ -2461,8 +2463,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& scissor.p1 = Rml::Math::Max(scissor.p1 / 2, scissor.p0); const bool from_source = (i % 2 == 0); Gfx::BindTexture(m_d3d_context, from_source ? source_destination : temp); - m_d3d_context->OMSetRenderTargets(1, &(from_source ? temp : source_destination).render_target_view, - (from_source ? temp : source_destination).depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, (from_source ? temp : source_destination)); SetScissor(scissor, true); DrawFullscreenQuad({}, uv_scaling); @@ -2481,7 +2482,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& if (transfer_to_temp_buffer) { Gfx::BindTexture(m_d3d_context, source_destination); - m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, temp.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, temp); DrawFullscreenQuad(); } @@ -2518,14 +2519,14 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // Blur render pass - vertical. Gfx::BindTexture(m_d3d_context, temp); - m_d3d_context->OMSetRenderTargets(1, &source_destination.render_target_view, source_destination.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, source_destination); SetTexelOffset({0.f, 1.f}, temp.height); DrawFullscreenQuad(); // Blur render pass - horizontal. Gfx::BindTexture(m_d3d_context, source_destination); - m_d3d_context->OMSetRenderTargets(1, &temp.render_target_view, temp.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, temp); // Add a 1px transparent border around the blur region by first clearing with a padded scissor. This helps prevent // artifacts when upscaling the blur result in the later step. On Intel and AMD, we have observed that during @@ -2619,7 +2620,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetSamplers(0, 1, &m_sampler_state); - m_d3d_context->OMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, destination); DrawFullscreenQuad(); @@ -2655,7 +2656,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetRenderTargets(1, &secondary.render_target_view, secondary.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, secondary); D3D11_MAPPED_SUBRESOURCE mappedResource{}; // Lock the constant buffer so it can be written to. @@ -2740,7 +2741,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, destination); DrawFullscreenQuad(); @@ -2763,7 +2764,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetRenderTargets(1, &destination.render_target_view, destination.depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, destination); DrawFullscreenQuad(); @@ -2874,7 +2875,7 @@ void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rm float blendFactor[4] = {1.0f, 1.0f, 1.0f, 1.0f}; m_d3d_context->OMSetBlendState(m_blend_state_disable_color, blendFactor, 0xffffffff); - m_d3d_context->OMSetRenderTargets(1, &m_render_layers.GetTopLayer().render_target_view, m_render_layers.GetTopLayer().depth_stencil_view); + Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); m_d3d_context->OMSetDepthStencilState(stencil_state, 1); bool clear_stencil = (operation == ClipMaskOperation::Set || operation == ClipMaskOperation::SetInverse); From 8538211f92fd457ed92f092bad0c3f1e1dc82ddb Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 1 Oct 2024 13:32:01 +0200 Subject: [PATCH 092/125] (fix): DrawGeometry match behaviour of GL3 --- Backends/RmlUi_Renderer_DX11.cpp | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 45e593dca..9eb152720 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -840,6 +840,7 @@ static void DestroyRenderTarget(RenderTargetData& rt) rt = {}; } +static inline void BindTexture(ID3D11DeviceContext* context, const RenderTargetData& rt, uint32_t slot = 0) { context->PSSetShaderResources(slot, 1, &rt.render_target_shader_resource_view); } @@ -873,6 +874,7 @@ static bool CreateShaders(ID3D11Device* p_device, ProgramData& data) for (const ProgramDefinition& def : program_definitions) { + if (!CreateProgram(p_device, data.programs[def.id], data.vert_shaders[def.vert_shader], data.frag_shaders[def.frag_shader])) return ReportError("program", def.name_str); } @@ -1455,7 +1457,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm { DX11_GeometryData geometryData = m_geometry_cache[handle]; - if (texture != TexturePostprocess && m_translation != translation) + if (m_translation != translation) { m_translation = translation; m_cbuffer_dirty = true; @@ -1475,15 +1477,15 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm m_d3d_context->PSSetShaderResources(0, 1, &texture_view); } m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); - UpdateConstantBuffer(); } else { // No texture, use color UseProgram(ProgramId::Color); - UpdateConstantBuffer(); } + UpdateConstantBuffer(); + m_d3d_context->IASetInputLayout(m_vertex_layout); m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); @@ -1495,6 +1497,10 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm m_d3d_context->IASetIndexBuffer(geometryData.index_buffer, DXGI_FORMAT_R32_UINT, 0); m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); m_d3d_context->DrawIndexed(geometryData.index_count, 0, 0); + + // Unbind resources + ID3D11ShaderResourceView* const nullSRV = nullptr; + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); } else { From 254636ae85b88afaf01b9c7c1ffaa5483b6684b5 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 1 Oct 2024 13:33:00 +0200 Subject: [PATCH 093/125] (fix): Simplify clip mask debug scope --- Backends/RmlUi_Renderer_DX11.cpp | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 9eb152720..b2d163436 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2830,18 +2830,15 @@ void RenderInterface_DX11::UseProgram(ProgramId program_id) { void RenderInterface_DX11::EnableClipMask(bool enable) { - LPCWSTR clipMaskName = L"EnableClipMask"; - if (!enable) - clipMaskName = L"DisableClipMask"; - DebugScope scope(clipMaskName); - m_is_stencil_enabled = enable; if (enable) { + DebugScope scope(L"EnableClipMask"); m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_stencil_test, m_stencil_ref_value); } else { + DebugScope scope(L"DisableClipMask"); m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); } } From 064577c1f33abf231ded380d410d185981b46530 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 1 Oct 2024 13:33:20 +0200 Subject: [PATCH 094/125] (chore): Remove redundant code --- Backends/RmlUi_Renderer_DX11.cpp | 2 -- 1 file changed, 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index b2d163436..c17f45216 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1686,8 +1686,6 @@ Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::SpanGenerateMips(gpu_texture_view); } - uintptr_t handleId = HashPointer((uintptr_t)gpu_texture_view); - return Rml::TextureHandle(gpu_texture_view); } From dc6b3bc08a073ad2558ea40338bb36663e5c8dcc Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 1 Oct 2024 17:20:31 +0200 Subject: [PATCH 095/125] (feat): Better blitting --- Backends/RmlUi_Renderer_DX11.cpp | 172 +++++++++++++++++++++---------- Backends/RmlUi_Renderer_DX11.h | 3 +- 2 files changed, 119 insertions(+), 56 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index c17f45216..69568c4fb 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -940,7 +940,14 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* ZeroMemory(&blend_desc, sizeof(blend_desc)); blend_desc.AlphaToCoverageEnable = false; blend_desc.IndependentBlendEnable = false; - blend_desc.RenderTarget[0].BlendEnable = false; + blend_desc.RenderTarget[0].BlendEnable = true; + blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_ZERO; + blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_ONE; + blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ZERO; + blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_ONE; + blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; HRESULT result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_disable); RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); #ifdef RMLUI_DX_DEBUG @@ -1001,6 +1008,24 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* return; } #endif + + blend_desc.RenderTarget[0].BlendEnable = true; + blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_ZERO; + blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_ONE; + blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ZERO; + blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_ONE; + blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; + blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; + result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_replace); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); + #ifdef RMLUI_DX_DEBUG + if (FAILED(result)) + { + Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); + return; + } + #endif } // Scissor regions require a rasterizer state. Cache one for scissor on and off @@ -1176,6 +1201,7 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_color_filter); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable_color); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_replace); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_intersect); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_set); @@ -1841,73 +1867,109 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, RMLUI_ASSERTMSG(dest_width != 0, "Invalid destination rectangle (width)!"); RMLUI_ASSERTMSG(dest_height != 0, "Invalid destination rectangle (height)!"); - ID3D11RenderTargetView* original_render_target_view = nullptr; - ID3D11DepthStencilView* original_depth_stencil_view = nullptr; - m_d3d_context->OMGetRenderTargets(1, &original_render_target_view, &original_depth_stencil_view); + bool is_flipped = src_width < 0 || src_height < 0 || dest_width < 0 || dest_height < 0; + bool is_stretched = src_width != dest_width || src_height != dest_height; + bool is_full_copy = src_width == dest_width && src_height == dest_height && srcX0 == 0 && srcY0 == 0 && dstX0 == 0 && dstY0 == 0; - // Bind temporary as the render target - const Gfx::RenderTargetData& temporary_rt = m_render_layers.GetTemporary(); + if (is_flipped || is_stretched || !is_full_copy) + { + // Full draw call. Slow path because no equivalent in DX11 - // Unbind existing textures to prevent warning spam - ID3D11ShaderResourceView* null_shader_resource_views[2] = {nullptr, nullptr}; - m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); + // Cache current state + ID3D11RenderTargetView* original_render_target_view = nullptr; + ID3D11DepthStencilView* original_depth_stencil_view = nullptr; + m_d3d_context->OMGetRenderTargets(1, &original_render_target_view, &original_depth_stencil_view); + ID3D11BlendState* original_blend_state = nullptr; + float original_blend_factor[4]{}; + uint32_t original_blend_mask = 0; + m_d3d_context->OMGetBlendState(&original_blend_state, original_blend_factor, &original_blend_mask); - // Resolve from source to the temporary first - m_d3d_context->ResolveSubresource(temporary_rt.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); + // Bind temporary as the render target + const Gfx::RenderTargetData& temporary_rt = m_render_layers.GetTemporary(); - m_d3d_context->RSSetState(m_rasterizer_state_scissor_disabled); + // Unbind existing textures to prevent warning spam + ID3D11ShaderResourceView* null_shader_resource_views[2] = {nullptr, nullptr}; + m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); - // Bind destination as our final render texture - m_d3d_context->OMSetRenderTargets(1, &dest.render_target_view, dest.depth_stencil_view); + // Replace mode + float blendFactor[4] = {1.0f, 1.0f, 1.0f, 1.0f}; + m_d3d_context->OMSetBlendState(m_blend_state_replace, blendFactor, 0xFFFFFFFF); - // Draw a quad into temporary with source bound as the texture, and the UVs lerping to match the new dimensions - // We want to map UV 0 - 1 to src min-max - // We want to map pos -1 - +1 to dst min max - UseProgram(ProgramId::Blit); + // Resolve from source to the temporary first + m_d3d_context->ResolveSubresource(temporary_rt.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); - D3D11_MAPPED_SUBRESOURCE mappedResource{}; - // Lock the constant buffer so it can be written to. - HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); - if (FAILED(result)) - { - // Restore state before exiting - m_d3d_context->OMSetRenderTargets(1, &original_render_target_view, original_depth_stencil_view); - original_render_target_view->Release(); - original_depth_stencil_view->Release(); - return; - } + // Bind destination as our final render texture + Gfx::BindRenderTarget(m_d3d_context, dest); - // Get a pointer to the data in the constant buffer. - ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + D3D11_VIEWPORT d3dviewport{}; + d3dviewport.TopLeftX = 0; + d3dviewport.TopLeftY = 0; + d3dviewport.Width = (float) dest.width; + d3dviewport.Height = (float) dest.height; + d3dviewport.MinDepth = 0.0f; + d3dviewport.MaxDepth = 1.0f; + m_d3d_context->RSSetViewports(1, &d3dviewport); - dataPtr->transform = m_transform; + // Draw a quad into temporary with source bound as the texture, and the UVs lerping to match the new dimensions + // We want to map UV 0 - 1 to src min-max + // We want to map pos -1 - +1 to dst min max + UseProgram(ProgramId::Blit); - // To blit we need to map the UVs from a 0 to 1 range to a src0 to src1 range - dataPtr->blit.src_x_min = (float) srcX0 / (float) source.width; // Map to 0 - dataPtr->blit.src_y_min = (float) srcY0 / (float) source.height; // Map to 0 - dataPtr->blit.src_x_max = (float) srcX1 / (float) source.width; // Map to +1 - dataPtr->blit.src_y_max = (float) srcY1 / (float) source.height; // Map to +1 + ShaderCbuffer* dataPtr = nullptr; - // To blit we need to map the position from a -1 to 1 range to a dst0 dst1 range - dataPtr->blit.dst_x_min = (float) dstX0 / (float) dest.width; // Map to -1 - dataPtr->blit.dst_y_min = (float) dstY0 / (float) dest.height; // Map to -1 - dataPtr->blit.dst_x_max = (float) dstX1 / (float) dest.width; // Map to +1 - dataPtr->blit.dst_y_max = (float) dstY1 / (float) dest.height; // Map to +1 + D3D11_MAPPED_SUBRESOURCE mappedResource{}; + // Lock the constant buffer so it can be written to. + HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + // Restore state before exiting + goto blit_restore_state; + return; + } - // Upload to the GPU. - m_d3d_context->Unmap(m_shader_buffer, 0); + // Get a pointer to the data in the constant buffer. + dataPtr = (ShaderCbuffer*)mappedResource.pData; - // Bind texture - m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); - Gfx::BindTexture(m_d3d_context, temporary_rt); - - DrawFullscreenQuad(); + dataPtr->transform = m_transform; - // Restore state - m_d3d_context->OMSetRenderTargets(1, &original_render_target_view, original_depth_stencil_view); - m_d3d_context->RSSetState(m_scissor_enabled ? m_rasterizer_state_scissor_enabled : m_rasterizer_state_scissor_disabled); - original_render_target_view->Release(); - original_depth_stencil_view->Release(); + // To blit we need to map the UVs from a 0 to 1 range to a src0 to src1 range + dataPtr->blit.src_x_min = float(srcX0) / float(source.width); // Map to 0 + dataPtr->blit.src_y_min = float(srcY0) / float(source.height); // Map to 0 + dataPtr->blit.src_x_max = float(srcX1) / float(source.width); // Map to +1 + dataPtr->blit.src_y_max = float(srcY1) / float(source.height); // Map to +1 + + dataPtr->blit.dst_x_min = (dstX0 / float(dest.width)); + dataPtr->blit.dst_y_min = ((dest.height - dstY0 - dest_height) / float(dest.height)); + dataPtr->blit.dst_x_max = ((dstX0 + dest_width) / float(dest.width)); + dataPtr->blit.dst_y_max = ((dest.height - dstY0) / float(dest.height)); + + // Upload to the GPU. + m_d3d_context->Unmap(m_shader_buffer, 0); + + // Bind texture + m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); + + Gfx::BindTexture(m_d3d_context, temporary_rt); + + DrawFullscreenQuad(); + + // Restore state + blit_restore_state: + m_d3d_context->OMSetRenderTargets(1, &original_render_target_view, original_depth_stencil_view); + m_d3d_context->OMSetBlendState(original_blend_state, original_blend_factor, original_blend_mask); + original_render_target_view->Release(); + original_depth_stencil_view->Release(); + original_blend_state->Release(); + + d3dviewport.Width = m_viewport_width; + d3dviewport.Height = m_viewport_height; + m_d3d_context->RSSetViewports(1, &d3dviewport); + } + else + { + // Resolve and move on + m_d3d_context->ResolveSubresource(dest.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); + } } void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle) @@ -2055,7 +2117,7 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() Gfx::BindTexture(m_d3d_context, source); UseProgram(ProgramId::Passthrough); ID3D11BlendState* blend_state_backup = m_current_blend_state; - SetBlendState(m_blend_state_disable); + SetBlendState(m_blend_state_replace); DrawFullscreenQuad(); diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index e9a0597c6..7be4e8b18 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -197,6 +197,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11BlendState* m_blend_state_disable = nullptr; ID3D11BlendState* m_blend_state_disable_color = nullptr; ID3D11BlendState* m_blend_state_color_filter = nullptr; + ID3D11BlendState* m_blend_state_replace = nullptr; ProgramId active_program = {}; Rml::Rectanglei m_scissor_state = Rml::Rectanglei::MakeInvalid(); @@ -206,7 +207,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { public: ID3D11Buffer* vertex_buffer = nullptr; ID3D11Buffer* index_buffer = nullptr; - uint32_t index_count = 0; + size_t index_count = 0; }; std::unordered_map m_geometry_cache; From a8c9d36ecc83a7737ab11cf03559b2a0da9e5609 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 2 Oct 2024 01:13:27 +0200 Subject: [PATCH 096/125] (fix): Correct blitting --- Backends/RmlUi_Renderer_DX11.cpp | 205 +++++++++++-------------------- Backends/RmlUi_Renderer_DX11.h | 16 +-- 2 files changed, 74 insertions(+), 147 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 69568c4fb..1618ec4d5 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -463,53 +463,6 @@ float4 PSMain(const PS_Input IN) : SV_TARGET }; )"; -constexpr const char shader_vert_blit[] = RMLUI_SHADER_HEADER R"( -struct VS_Input -{ - float2 position : POSITION; - float4 color : COLOR; - float2 uv : TEXCOORD; -}; - -struct PS_Input -{ - float4 position : SV_Position; - float4 color : COLOR; - float2 uv : TEXCOORD; -}; - -cbuffer ConstantBuffer : register(b0) -{ - float4x4 m_transform; - float2 m_source_min; - float2 m_source_max; - float2 m_destination_min; - float2 m_destination_max; - float4 _padding2[20]; // Padding so that cbuffer aligns with the largest one (gradient) -}; - -PS_Input VSMain(const VS_Input IN) -{ - PS_Input result = (PS_Input)0; - - result.position = float4(IN.position.xy, 0.0f, 1.0f); - result.uv = float2(IN.uv.x, 1.0f - IN.uv.y); - - // Correct coordinates to match source and dest bounds - // For source we lerp the UVs - result.uv.x = lerp(m_source_min.x, m_source_max.x, result.uv.x); - result.uv.y = lerp(m_source_min.y, m_source_max.y, result.uv.y); - - // For destination we lerp the positions. Since the clip space box has a range of -1 to +1, we have to remap -1 to +1 to 0 to 1, lerp, then remap back to -1 to +1 - result.position.xy = result.position.xy * 0.5f + 0.5f; // Remap to 0-+1 - result.position.x = lerp(m_destination_min.x, m_destination_max.x, result.position.x); - result.position.y = lerp(m_destination_min.y, m_destination_max.y, result.position.y); - result.position.xy = result.position.xy * 2.0f - 1.0; // Remap to -1-+1 - - return result; -}; -)"; - enum class ProgramId { None, Color, @@ -521,14 +474,12 @@ enum class ProgramId { BlendMask, Blur, DropShadow, - Blit, Count, }; enum class VertShaderId { Main, Passthrough, Blur, - Blit, Count, }; enum class FragShaderId { @@ -574,7 +525,6 @@ static const VertShaderDefinition vert_shader_definitions[] = { {VertShaderId::Main, "main", shader_vert_main, sizeof(shader_vert_main)}, {VertShaderId::Passthrough, "passthrough", shader_vert_passthrough, sizeof(shader_vert_passthrough)}, {VertShaderId::Blur, "blur", shader_vert_blur, sizeof(shader_vert_blur)}, - {VertShaderId::Blit, "blit", shader_vert_blit, sizeof(shader_vert_blit)}, }; static const FragShaderDefinition frag_shader_definitions[] = { {FragShaderId::Color, "color", shader_frag_color, sizeof(shader_frag_color)}, @@ -597,7 +547,6 @@ static const ProgramDefinition program_definitions[] = { {ProgramId::BlendMask, "blend_mask", VertShaderId::Passthrough, FragShaderId::BlendMask}, {ProgramId::Blur, "blur", VertShaderId::Blur, FragShaderId::Blur}, {ProgramId::DropShadow, "drop_shadow", VertShaderId::Passthrough, FragShaderId::DropShadow}, - {ProgramId::Blit, "blur", VertShaderId::Blit, FragShaderId::Passthrough}, }; // clang-format on @@ -936,31 +885,13 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* if (!m_blend_state_enable) { + // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. + // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). D3D11_BLEND_DESC blend_desc{}; ZeroMemory(&blend_desc, sizeof(blend_desc)); blend_desc.AlphaToCoverageEnable = false; blend_desc.IndependentBlendEnable = false; blend_desc.RenderTarget[0].BlendEnable = true; - blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_ZERO; - blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_ONE; - blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; - blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ZERO; - blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_ONE; - blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; - blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; - HRESULT result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_disable); - RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); - #ifdef RMLUI_DX_DEBUG - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); - return; - } - #endif - - // RmlUi serves vertex colors and textures with premultiplied alpha, set the blend mode accordingly. - // Equivalent to glBlendFunc(GL_ONE, GL_ONE_MINUS_SRC_ALPHA). - blend_desc.RenderTarget[0].BlendEnable = true; blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_ONE; blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_INV_SRC_ALPHA; blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; @@ -968,7 +899,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_INV_SRC_ALPHA; blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; - result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_enable); + HRESULT result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_enable); RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); #ifdef RMLUI_DX_DEBUG if (FAILED(result)) @@ -1008,24 +939,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* return; } #endif - - blend_desc.RenderTarget[0].BlendEnable = true; - blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_ZERO; - blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_ONE; - blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; - blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_ZERO; - blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_ONE; - blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; - blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; - result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_replace); - RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); - #ifdef RMLUI_DX_DEBUG - if (FAILED(result)) - { - Rml::Log::Message(Rml::Log::LT_ERROR, "ID3D11Device::CreateBlendState (%d)", result); - return; - } - #endif } // Scissor regions require a rasterizer state. Cache one for scissor on and off @@ -1827,11 +1740,20 @@ void RenderInterface_DX11::Clear() void RenderInterface_DX11::SetBlendState(ID3D11BlendState* blend_state) { - if (blend_state != m_current_blend_state) + if (!m_current_blend_state_enabled || blend_state != m_current_blend_state) { const float blend_factor[4] = {1.0f, 1.0f, 1.0f, 1.0f}; m_d3d_context->OMSetBlendState(blend_state, blend_factor, 0xFFFFFFFF); m_current_blend_state = blend_state; + m_current_blend_state_enabled = true; + } +} + +void RenderInterface_DX11::DisableBlend() { + if (m_current_blend_state_enabled) + { + m_d3d_context->OMSetBlendState(nullptr, nullptr, 0xFFFFFFFF); + m_current_blend_state_enabled = false; } } @@ -1873,6 +1795,7 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, if (is_flipped || is_stretched || !is_full_copy) { + DebugScope scope(L"Complex"); // Full draw call. Slow path because no equivalent in DX11 // Cache current state @@ -1891,9 +1814,8 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, ID3D11ShaderResourceView* null_shader_resource_views[2] = {nullptr, nullptr}; m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); - // Replace mode - float blendFactor[4] = {1.0f, 1.0f, 1.0f, 1.0f}; - m_d3d_context->OMSetBlendState(m_blend_state_replace, blendFactor, 0xFFFFFFFF); + // Disable blending + m_d3d_context->OMSetBlendState(nullptr, nullptr, 0xFFFFFFFF); // Resolve from source to the temporary first m_d3d_context->ResolveSubresource(temporary_rt.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); @@ -1910,48 +1832,65 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); - // Draw a quad into temporary with source bound as the texture, and the UVs lerping to match the new dimensions - // We want to map UV 0 - 1 to src min-max - // We want to map pos -1 - +1 to dst min max - UseProgram(ProgramId::Blit); - - ShaderCbuffer* dataPtr = nullptr; - - D3D11_MAPPED_SUBRESOURCE mappedResource{}; - // Lock the constant buffer so it can be written to. - HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); - if (FAILED(result)) - { - // Restore state before exiting - goto blit_restore_state; - return; - } - - // Get a pointer to the data in the constant buffer. - dataPtr = (ShaderCbuffer*)mappedResource.pData; - - dataPtr->transform = m_transform; - - // To blit we need to map the UVs from a 0 to 1 range to a src0 to src1 range - dataPtr->blit.src_x_min = float(srcX0) / float(source.width); // Map to 0 - dataPtr->blit.src_y_min = float(srcY0) / float(source.height); // Map to 0 - dataPtr->blit.src_x_max = float(srcX1) / float(source.width); // Map to +1 - dataPtr->blit.src_y_max = float(srcY1) / float(source.height); // Map to +1 - - dataPtr->blit.dst_x_min = (dstX0 / float(dest.width)); - dataPtr->blit.dst_y_min = ((dest.height - dstY0 - dest_height) / float(dest.height)); - dataPtr->blit.dst_x_max = ((dstX0 + dest_width) / float(dest.width)); - dataPtr->blit.dst_y_max = ((dest.height - dstY0) / float(dest.height)); - - // Upload to the GPU. - m_d3d_context->Unmap(m_shader_buffer, 0); - // Bind texture m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); - Gfx::BindTexture(m_d3d_context, temporary_rt); + UseProgram(ProgramId::Passthrough); - DrawFullscreenQuad(); + // Draw a quad into temporary with source bound as the texture, and the UVs lerping to match the new dimensions + // We want to map UV 0 - 1 to src min-max + // We want to map pos -1 - +1 to dst min max + + float uv_x_min = float(srcX0) / float(source.width); // Map to 0 + float uv_y_max = float(srcY0) / float(source.height); // Map to 0 + float uv_x_max = float(srcX1) / float(source.width); // Map to +1 + float uv_y_min = float(srcY1) / float(source.height); // Map to +1 + + float pos_x_min = (dstX0 / float(dest.width)) * 2.0f - 1.0f; + float pos_y_min = ((dest.height - dstY0 - dest_height) / float(dest.height)) * 2.0f - 1.0f; + float pos_x_max = ((dstX0 + dest_width) / float(dest.width)) * 2.0f - 1.0f; + float pos_y_max = ((dest.height - dstY0) / float(dest.height)) * 2.0f - 1.0f; + + Rml::Mesh mesh; + const int v0 = (int)mesh.vertices.size(); + const int i0 = (int)mesh.indices.size(); + + mesh.vertices.resize(mesh.vertices.size() + 4); + mesh.indices.resize(mesh.indices.size() + 6); + Rml::Vertex* vertices = mesh.vertices.data(); + int* indices = mesh.indices.data(); + + vertices[v0 + 0].position.x = pos_x_min; + vertices[v0 + 0].position.y = pos_y_min; + vertices[v0 + 0].tex_coord.x = uv_x_min; + vertices[v0 + 0].tex_coord.y = uv_y_min; + + vertices[v0 + 1].position.x = pos_x_max; + vertices[v0 + 1].position.y = pos_y_min; + vertices[v0 + 1].tex_coord.x = uv_x_max; + vertices[v0 + 1].tex_coord.y = uv_y_min; + + vertices[v0 + 2].position.x = pos_x_max; + vertices[v0 + 2].position.y = pos_y_max; + vertices[v0 + 2].tex_coord.x = uv_x_max; + vertices[v0 + 2].tex_coord.y = uv_y_max; + + vertices[v0 + 3].position.x = pos_x_min; + vertices[v0 + 3].position.y = pos_y_max; + vertices[v0 + 3].tex_coord.x = uv_x_min; + vertices[v0 + 3].tex_coord.y = uv_y_max; + + indices[i0 + 0] = v0 + 0; + indices[i0 + 1] = v0 + 3; + indices[i0 + 2] = v0 + 1; + + indices[i0 + 3] = v0 + 1; + indices[i0 + 4] = v0 + 3; + indices[i0 + 5] = v0 + 2; + + const Rml::CompiledGeometryHandle geometry = CompileGeometry(mesh.vertices, mesh.indices); + RenderGeometry(geometry, {}, RenderInterface_DX11::TexturePostprocess); + ReleaseGeometry(geometry); // Restore state blit_restore_state: @@ -1964,9 +1903,11 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, d3dviewport.Width = m_viewport_width; d3dviewport.Height = m_viewport_height; m_d3d_context->RSSetViewports(1, &d3dviewport); + m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); } else { + DebugScope scope(L"Simple"); // Resolve and move on m_d3d_context->ResolveSubresource(dest.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); } diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 7be4e8b18..5fa4f4344 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -194,10 +194,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11InputLayout* m_vertex_layout = nullptr; ID3D11BlendState* m_blend_state_enable = nullptr; - ID3D11BlendState* m_blend_state_disable = nullptr; ID3D11BlendState* m_blend_state_disable_color = nullptr; ID3D11BlendState* m_blend_state_color_filter = nullptr; - ID3D11BlendState* m_blend_state_replace = nullptr; ProgramId active_program = {}; Rml::Rectanglei m_scissor_state = Rml::Rectanglei::MakeInvalid(); @@ -213,6 +211,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { std::unordered_map m_geometry_cache; // D3D11 state for RmlUi rendering + bool m_current_blend_state_enabled = false; ID3D11BlendState* m_current_blend_state = nullptr; // Backup of prior D3D11 state @@ -279,19 +278,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::Vector2f dimensions; float value; } creation; - struct Blit { - int _padding[16]; - - float src_x_min; - float src_y_min; - float src_x_max; - float src_y_max; - - float dst_x_min; - float dst_y_min; - float dst_x_max; - float dst_y_max; - } blit; }; #pragma pack() From 974599420cf996b65eaebf3a23bceacb23cda7aa Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 2 Oct 2024 01:21:58 +0200 Subject: [PATCH 097/125] (fix): Correct bad blend state --- Backends/RmlUi_Renderer_DX11.cpp | 18 +++++++++--------- Backends/RmlUi_Renderer_DX11.h | 1 + 2 files changed, 10 insertions(+), 9 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 1618ec4d5..4a22a2875 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1111,10 +1111,8 @@ void RenderInterface_DX11::Cleanup() // Cleans up all general resources DX_CLEANUP_RESOURCE_IF_CREATED(m_sampler_state); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_enable); - DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_color_filter); DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_disable_color); - DX_CLEANUP_RESOURCE_IF_CREATED(m_blend_state_replace); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_disable); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_intersect); DX_CLEANUP_RESOURCE_IF_CREATED(m_depth_stencil_state_stencil_set); @@ -1955,13 +1953,15 @@ void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml:: UseProgram(ProgramId::Passthrough); + ID3D11BlendState* current_blend_state = m_current_blend_state; + if (blend_mode == BlendMode::Replace) - SetBlendState(m_blend_state_disable); + DisableBlend(); DrawFullscreenQuad(); if (blend_mode == BlendMode::Replace) - SetBlendState(m_blend_state_enable); + SetBlendState(m_current_blend_state); if (destination_handle != m_render_layers.GetTopLayerHandle()) Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); @@ -2058,7 +2058,7 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() Gfx::BindTexture(m_d3d_context, source); UseProgram(ProgramId::Passthrough); ID3D11BlendState* blend_state_backup = m_current_blend_state; - SetBlendState(m_blend_state_replace); + DisableBlend(); DrawFullscreenQuad(); @@ -2639,7 +2639,7 @@ void RenderInterface_DX11::RenderFilters(Rml::Span Date: Wed, 2 Oct 2024 01:23:53 +0200 Subject: [PATCH 098/125] (fix): Correct color matrix implementation Not finished. Flickering --- Backends/RmlUi_Renderer_DX11.cpp | 14 ++++++++------ Backends/RmlUi_Renderer_DX11.h | 1 + 2 files changed, 9 insertions(+), 6 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 4a22a2875..ea30c6cd0 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -315,8 +315,9 @@ SamplerState g_SamplerLinear : register(s0); cbuffer ConstantBuffer : register(b0) { + float4 _padding[5]; float4x4 m_color_matrix; - float4 _padding[22]; // Padding so that cbuffer aligns with the largest one (gradient) + float4 _padding2[17]; // Padding so that cbuffer aligns with the largest one (gradient) }; struct PS_Input @@ -2714,7 +2715,6 @@ void RenderInterface_DX11::RenderFilters(Rml::Spantransform = m_transform; dataPtr->translation = m_translation; memset(&dataPtr->color_matrix.color_matrix, 0, sizeof(dataPtr->color_matrix.color_matrix)); - memcpy_s(&dataPtr->color_matrix.color_matrix, sizeof(dataPtr->color_matrix.color_matrix), filter.color_matrix.data(), sizeof(dataPtr->color_matrix.color_matrix)); + memcpy_s(&dataPtr->color_matrix.color_matrix, sizeof(dataPtr->color_matrix.color_matrix), + transpose ? filter.color_matrix.Transpose().data() : filter.color_matrix.data(), sizeof(dataPtr->color_matrix.color_matrix)); // Upload to the GPU. m_d3d_context->Unmap(m_shader_buffer, 0); @@ -2747,15 +2748,16 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(0, 2, null_shader_resource_views); - m_render_layers.SwapPostprocessPrimarySecondary(); SetBlendState(blend_state_backup); + + ID3D11ShaderResourceView* const nullSRV = nullptr; + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); } break; case FilterType::MaskImage: diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 6b48013ab..40f1c17a0 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -259,6 +259,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { float stop_positions[16]; } gradient; struct ColorMatrix { + int _padding[20]; Rml::Matrix4f color_matrix; } color_matrix; struct Blur { From 6ca7379684aa09d3137d840cf7a2199c9572c82b Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 2 Oct 2024 01:24:22 +0200 Subject: [PATCH 099/125] (fix): Unbind resources when done --- Backends/RmlUi_Renderer_DX11.cpp | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index ea30c6cd0..0483f6432 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2055,6 +2055,10 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); const Gfx::RenderTargetData& destination = m_render_layers.GetBlendMask(); + // Unbind resources + ID3D11ShaderResourceView* const nullSRV = nullptr; + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); + Gfx::BindRenderTarget(m_d3d_context, destination); Gfx::BindTexture(m_d3d_context, source); UseProgram(ProgramId::Passthrough); @@ -2063,6 +2067,9 @@ Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() DrawFullscreenQuad(); + // Unbind resources + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); + SetBlendState(blend_state_backup); Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); @@ -2634,6 +2641,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(0, 2, null_shader_resource_views); } break; case FilterType::Blur: @@ -2711,6 +2719,9 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(0, 1, &nullSRV); } break; case FilterType::ColorMatrix: @@ -2777,11 +2788,9 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(0, 2, null_shader_resource_views); - m_render_layers.SwapPostprocessPrimarySecondary(); SetBlendState(blend_state_backup); + m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); } break; case FilterType::Invalid: @@ -2790,6 +2799,10 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(0, 1, &nullSRV); } } From 911b51ea70765fce1823755c970c9e75bcba4b51 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 2 Oct 2024 01:24:52 +0200 Subject: [PATCH 100/125] (fix): Correct binding order Eliminates errors in the debug layer --- Backends/RmlUi_Renderer_DX11.cpp | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 0483f6432..d8c6c94e0 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2441,7 +2441,6 @@ static void SetBlurWeights(Rml::Vector4f& p_weights, float sigma) void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& source_destination, const Gfx::RenderTargetData& temp, const Rml::Rectanglei window_flipped) { - return; DebugScope scope(L"RenderBlur"); RMLUI_ASSERT(&source_destination != &temp && source_destination.width == temp.width && source_destination.height == temp.height); RMLUI_ASSERT(window_flipped.Valid()); @@ -2633,9 +2632,9 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetSamplers(0, 1, &m_sampler_state); - Gfx::BindRenderTarget(m_d3d_context, destination); DrawFullscreenQuad(); @@ -2671,8 +2670,8 @@ void RenderInterface_DX11::RenderFilters(Rml::Span Date: Wed, 2 Oct 2024 01:25:09 +0200 Subject: [PATCH 101/125] (fix): Correct alignment of blur shader --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++-- Backends/RmlUi_Renderer_DX11.h | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index d8c6c94e0..34468bc9c 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -382,8 +382,8 @@ cbuffer SharedConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; - float4 m_weights; float2 m_texelOffset; + float4 m_weights; float2 m_texCoordMin; float2 m_texCoordMax; float4 _padding[19]; // Padding so that cbuffer aligns with the largest one (gradient) @@ -410,8 +410,8 @@ cbuffer ConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; - float4 m_weights; float2 m_texelOffset; + float4 m_weights; float2 m_texCoordMin; float2 m_texCoordMax; float4 _padding[19]; // Padding so that cbuffer aligns with the largest one (gradient) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 40f1c17a0..16998fdf3 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -264,8 +264,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { } color_matrix; struct Blur { int _padding[18]; - Rml::Vector4f weights; Rml::Vector2f texel_offset; + Rml::Vector4f weights; Rml::Vector2f texcoord_min; Rml::Vector2f texcoord_max; } blur; From 6512dc9e3dc46c56c2cd29df8dda75886a703542 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Wed, 2 Oct 2024 01:30:29 +0200 Subject: [PATCH 102/125] (chore): Remove a lot of warnings --- Backends/RmlUi_Renderer_DX11.cpp | 43 +++++++++++++++----------------- 1 file changed, 20 insertions(+), 23 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 34468bc9c..0759f90e2 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1169,8 +1169,8 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT D3D11_VIEWPORT d3dviewport; d3dviewport.TopLeftX = 0; d3dviewport.TopLeftY = 0; - d3dviewport.Width = m_viewport_width; - d3dviewport.Height = m_viewport_height; + d3dviewport.Width = (float) m_viewport_width; + d3dviewport.Height = (float) m_viewport_height; d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); @@ -1318,7 +1318,7 @@ Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span(sizeof(Rml::Vertex) * vertices.size()); vertexBufferDesc.BindFlags = D3D11_BIND_VERTEX_BUFFER; vertexBufferDesc.CPUAccessFlags = 0; vertexBufferDesc.MiscFlags = 0; @@ -1341,7 +1341,7 @@ Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Span(sizeof(int) * indices.size()); indexBufferDesc.BindFlags = D3D11_BIND_INDEX_BUFFER; indexBufferDesc.CPUAccessFlags = 0; indexBufferDesc.MiscFlags = 0; @@ -1434,7 +1434,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm m_d3d_context->IASetVertexBuffers(0, 1, &geometryData.vertex_buffer, &stride, &offset); m_d3d_context->IASetIndexBuffer(geometryData.index_buffer, DXGI_FORMAT_R32_UINT, 0); m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); - m_d3d_context->DrawIndexed(geometryData.index_count, 0, 0); + m_d3d_context->DrawIndexed(static_cast(geometryData.index_count), 0, 0); // Unbind resources ID3D11ShaderResourceView* const nullSRV = nullptr; @@ -1606,7 +1606,7 @@ Rml::TextureHandle RenderInterface_DX11::GenerateTexture(Rml::Span(-1); // Create the shader resource view for the texture. result = m_d3d_device->CreateShaderResourceView(gpu_texture, &srvDesc, &gpu_texture_view); @@ -1794,7 +1794,7 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, if (is_flipped || is_stretched || !is_full_copy) { - DebugScope scope(L"Complex"); + DebugScope scope2(L"Complex"); // Full draw call. Slow path because no equivalent in DX11 // Cache current state @@ -1892,21 +1892,20 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, ReleaseGeometry(geometry); // Restore state - blit_restore_state: m_d3d_context->OMSetRenderTargets(1, &original_render_target_view, original_depth_stencil_view); m_d3d_context->OMSetBlendState(original_blend_state, original_blend_factor, original_blend_mask); original_render_target_view->Release(); original_depth_stencil_view->Release(); original_blend_state->Release(); - d3dviewport.Width = m_viewport_width; - d3dviewport.Height = m_viewport_height; + d3dviewport.Width = static_cast(m_viewport_width); + d3dviewport.Height = static_cast(m_viewport_height); m_d3d_context->RSSetViewports(1, &d3dviewport); m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); } else { - DebugScope scope(L"Simple"); + DebugScope scope2(L"Simple"); // Resolve and move on m_d3d_context->ResolveSubresource(dest.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); } @@ -1954,8 +1953,6 @@ void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml:: UseProgram(ProgramId::Passthrough); - ID3D11BlendState* current_blend_state = m_current_blend_state; - if (blend_mode == BlendMode::Replace) DisableBlend(); @@ -2337,7 +2334,7 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, m_d3d_context->IASetVertexBuffers(0, 1, &geometry.vertex_buffer, &stride, &offset); m_d3d_context->IASetIndexBuffer(geometry.index_buffer, DXGI_FORMAT_R32_UINT, 0); m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); - m_d3d_context->DrawIndexed(geometry.index_count, 0, 0); + m_d3d_context->DrawIndexed(static_cast(geometry.index_count), 0, 0); } break; case CompiledShaderType::Creation: @@ -2376,7 +2373,7 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, m_d3d_context->IASetVertexBuffers(0, 1, &geometry.vertex_buffer, &stride, &offset); m_d3d_context->IASetIndexBuffer(geometry.index_buffer, DXGI_FORMAT_R32_UINT, 0); m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); - m_d3d_context->DrawIndexed(geometry.index_count, 0, 0); + m_d3d_context->DrawIndexed(static_cast(geometry.index_count), 0, 0); } break; case CompiledShaderType::Invalid: @@ -2460,8 +2457,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& D3D11_VIEWPORT d3dviewport; d3dviewport.TopLeftX = 0; d3dviewport.TopLeftY = 0; - d3dviewport.Width = source_destination.width / 2; - d3dviewport.Height = source_destination.height / 2; + d3dviewport.Width = static_cast(source_destination.width / 2); + d3dviewport.Height = static_cast(source_destination.height / 2); d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); @@ -2485,8 +2482,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& d3dviewport.TopLeftX = 0; d3dviewport.TopLeftY = 0; - d3dviewport.Width = source_destination.width; - d3dviewport.Height = source_destination.height; + d3dviewport.Width = static_cast(source_destination.width); + d3dviewport.Height = static_cast(source_destination.height); d3dviewport.MinDepth = 0.0f; d3dviewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &d3dviewport); @@ -2586,10 +2583,10 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& Gfx::BindTexture(m_d3d_context, temp); D3D11_VIEWPORT viewport; - viewport.TopLeftX = target_min.x; - viewport.TopLeftY = target_min.y; - viewport.Width = (float)(target_max.x - target_min.x); - viewport.Height = (float)(target_max.y - target_min.y); + viewport.TopLeftX = static_cast(target_min.x); + viewport.TopLeftY = static_cast(target_min.y); + viewport.Width = static_cast(target_max.x - target_min.x); + viewport.Height = static_cast(target_max.y - target_min.y); viewport.MinDepth = 0.0f; viewport.MaxDepth = 1.0f; m_d3d_context->RSSetViewports(1, &viewport); From 43799edc00ae393a9fd7e4a50254c76d23af7994 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 3 Oct 2024 09:19:08 +0200 Subject: [PATCH 103/125] (fix): Fix color matrix flickering --- Backends/RmlUi_Renderer_DX11.cpp | 64 ++++++++++++++++++++++---------- Backends/RmlUi_Renderer_DX11.h | 8 ++-- 2 files changed, 48 insertions(+), 24 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 0759f90e2..f05553110 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -313,11 +313,9 @@ static const char shader_frag_color_matrix[] = RMLUI_SHADER_HEADER R"( Texture2D g_InputTexture : register(t0); SamplerState g_SamplerLinear : register(s0); -cbuffer ConstantBuffer : register(b0) +cbuffer ConstantBuffer : register(b1) { - float4 _padding[5]; float4x4 m_color_matrix; - float4 _padding2[17]; // Padding so that cbuffer aligns with the largest one (gradient) }; struct PS_Input @@ -1021,6 +1019,15 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* { return; } + + cbufferDesc.ByteWidth = sizeof(ColorMatrixCbuffer); + + result = m_d3d_device->CreateBuffer(&cbufferDesc, nullptr, &m_color_matrix_cbuffer); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); + if (FAILED(result)) + { + return; + } } // Create sampler state for textures @@ -1121,6 +1128,7 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_disabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(m_color_matrix_cbuffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); DX_CLEANUP_RESOURCE_IF_CREATED(m_debug); } @@ -2724,46 +2732,63 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMGetBlendState(&original_blend_state, original_blend_factor, &original_sample_mask); DisableBlend(); + const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); + const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); + Gfx::BindRenderTarget(m_d3d_context, destination); + Gfx::BindTexture(m_d3d_context, source); + D3D11_MAPPED_SUBRESOURCE mappedResource{}; // Lock the constant buffer so it can be written to. - HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + HRESULT result = m_d3d_context->Map(m_color_matrix_cbuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); if (FAILED(result)) { + // Restore blend state + m_d3d_context->OMSetBlendState(original_blend_state, original_blend_factor, original_sample_mask); + original_blend_state->Release(); + m_current_blend_state_enabled = true; + + ID3D11ShaderResourceView* const nullSRV = nullptr; + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); + m_cbuffer_dirty = true; return; } // Get a pointer to the data in the constant buffer. - ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + ColorMatrixCbuffer* dataPtr = (ColorMatrixCbuffer*)mappedResource.pData; constexpr bool transpose = std::is_same::value; // Copy the data to the GPU - dataPtr->transform = m_transform; - dataPtr->translation = m_translation; - memset(&dataPtr->color_matrix.color_matrix, 0, sizeof(dataPtr->color_matrix.color_matrix)); - memcpy_s(&dataPtr->color_matrix.color_matrix, sizeof(dataPtr->color_matrix.color_matrix), - transpose ? filter.color_matrix.Transpose().data() : filter.color_matrix.data(), sizeof(dataPtr->color_matrix.color_matrix)); + memcpy_s(&dataPtr->color_matrix, sizeof(dataPtr->color_matrix), + transpose ? filter.color_matrix.Transpose().data() : filter.color_matrix.data(), sizeof(dataPtr->color_matrix)); // Upload to the GPU. - m_d3d_context->Unmap(m_shader_buffer, 0); + m_d3d_context->Unmap(m_color_matrix_cbuffer, 0); - m_cbuffer_dirty = true; - - const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); - const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessSecondary(); - Gfx::BindRenderTarget(m_d3d_context, destination); - Gfx::BindTexture(m_d3d_context, source); + m_d3d_context->PSSetConstantBuffers(1, 1, &m_color_matrix_cbuffer); DrawFullscreenQuad(); + ID3D11Buffer* null_cbuffer = nullptr; + m_d3d_context->PSSetConstantBuffers(1, 1, &null_cbuffer); + m_render_layers.SwapPostprocessPrimarySecondary(); - SetBlendState(blend_state_backup); + + // Restore blend state + m_d3d_context->OMSetBlendState(original_blend_state, original_blend_factor, original_sample_mask); + original_blend_state->Release(); + m_current_blend_state_enabled = true; ID3D11ShaderResourceView* const nullSRV = nullptr; m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); + m_cbuffer_dirty = true; } break; case FilterType::MaskImage: @@ -2944,7 +2969,6 @@ Rml::LayerHandle RenderInterface_DX11::RenderLayerStack::PushLayer() return GetTopLayerHandle(); } - void RenderInterface_DX11::RenderLayerStack::PopLayer() { RMLUI_ASSERT(m_layers_size > 0); diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 16998fdf3..8b89ebd78 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -182,6 +182,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { // Shaders ID3D11Buffer* m_shader_buffer = nullptr; + ID3D11Buffer* m_color_matrix_cbuffer = nullptr; bool m_cbuffer_dirty = true; ID3D11SamplerState* m_sampler_state = nullptr; @@ -258,10 +259,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::Vector4f stop_colors[16]; float stop_positions[16]; } gradient; - struct ColorMatrix { - int _padding[20]; - Rml::Matrix4f color_matrix; - } color_matrix; struct Blur { int _padding[18]; Rml::Vector2f texel_offset; @@ -281,6 +278,9 @@ class RenderInterface_DX11 : public Rml::RenderInterface { float value; } creation; }; + struct ColorMatrixCbuffer { + Rml::Matrix4f color_matrix; + } color_matrix; #pragma pack() Rml::CompiledGeometryHandle m_fullscreen_quad_geometry = 0; From 9b86546ddec1bb316cfe7463e8d248198c5d3672 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 3 Oct 2024 09:19:30 +0200 Subject: [PATCH 104/125] (fix): Remove padding from cbuffers in shaders --- Backends/RmlUi_Renderer_DX11.cpp | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index f05553110..e04a3d6bc 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -78,8 +78,6 @@ cbuffer SharedConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; - float2 _padding; - float4 _padding2[21]; // Padding so that cbuffer aligns with the largest one (gradient) }; PS_INPUT VSMain(const VS_Input IN) @@ -232,8 +230,6 @@ cbuffer SharedConstantBuffer : register(b0) float2 m_translate; float2 m_dimensions; float m_value; - float3 _padding; - float4 _padding2[20]; // Padding so that cbuffer aligns with the largest one (gradient) }; #define glsl_mod(x,y) (((x)-(y)*floor((x)/(y)))) @@ -276,8 +272,6 @@ cbuffer ConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; - float2 _padding; - float4 _padding2[21]; // Padding so that cbuffer aligns with the largest one (gradient) }; PS_Input VSMain(const VS_Input IN) @@ -384,7 +378,6 @@ cbuffer SharedConstantBuffer : register(b0) float4 m_weights; float2 m_texCoordMin; float2 m_texCoordMax; - float4 _padding[19]; // Padding so that cbuffer aligns with the largest one (gradient) }; PS_INPUT VSMain(const VS_Input IN) @@ -412,7 +405,6 @@ cbuffer ConstantBuffer : register(b0) float4 m_weights; float2 m_texCoordMin; float2 m_texCoordMax; - float4 _padding[19]; // Padding so that cbuffer aligns with the largest one (gradient) }; struct PS_Input @@ -444,8 +436,6 @@ cbuffer ConstantBuffer : register(b0) float2 m_texCoordMin; float2 m_texCoordMax; float4 m_color; - float2 _padding; - float4 _padding2[19]; // Padding so that cbuffer aligns with the largest one (gradient) }; struct PS_Input From 9651a9563a1a32183facad661d1c3837650f4eed Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 3 Oct 2024 09:20:46 +0200 Subject: [PATCH 105/125] (fix): Correct bind order for MaskImage --- Backends/RmlUi_Renderer_DX11.cpp | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e04a3d6bc..26a61b91e 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2766,6 +2766,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetConstantBuffers(1, 1, &null_cbuffer); @@ -2792,9 +2793,9 @@ void RenderInterface_DX11::RenderFilters(Rml::Span Date: Thu, 3 Oct 2024 18:53:21 +0200 Subject: [PATCH 106/125] (fix): Fix unnamed struct warning --- Backends/RmlUi_Renderer_DX11.cpp | 20 ++++++++++---------- Backends/RmlUi_Renderer_DX11.h | 2 +- 2 files changed, 11 insertions(+), 11 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 26a61b91e..0a4d8f402 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2307,8 +2307,8 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; // Copy the data to the GPU - dataPtr->transform = m_transform; - dataPtr->translation = translation; + dataPtr->common.transform = m_transform; + dataPtr->common.translation = translation; dataPtr->gradient.func = static_cast(shader.gradient_function); dataPtr->gradient.p = shader.p; dataPtr->gradient.v = shader.v; @@ -2354,8 +2354,8 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; // Copy the data to the GPU - dataPtr->transform = m_transform; - dataPtr->translation = translation; + dataPtr->common.transform = m_transform; + dataPtr->common.translation = translation; dataPtr->creation.value = (float)time; dataPtr->creation.dimensions = shader.dimensions; @@ -2507,8 +2507,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; // Copy the data to the GPU - dataPtr->transform = m_transform; - dataPtr->translation = m_translation; + dataPtr->common.transform = m_transform; + dataPtr->common.translation = m_translation; SetBlurWeights(dataPtr->blur.weights, sigma); @@ -2682,8 +2682,8 @@ void RenderInterface_DX11::RenderFilters(Rml::Spantransform = m_transform; - dataPtr->translation = m_translation; + dataPtr->common.transform = m_transform; + dataPtr->common.translation = m_translation; dataPtr->drop_shadow.color.x = color.red; dataPtr->drop_shadow.color.y = color.green; dataPtr->drop_shadow.color.z = color.blue; @@ -2833,8 +2833,8 @@ void RenderInterface_DX11::UpdateConstantBuffer() ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; // Copy the data to the GPU - dataPtr->transform = m_transform; - dataPtr->translation = m_translation; + dataPtr->common.transform = m_transform; + dataPtr->common.translation = m_translation; // Upload to the GPU. m_d3d_context->Unmap(m_shader_buffer, 0); diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index 8b89ebd78..ef0138dc5 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -249,7 +249,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { struct { Rml::Matrix4f transform; Rml::Vector2f translation; - }; + } common; struct Gradient { int _padding[18]; int func; From ada1c828f42a37c5ebe18aff3c3d9fb8c37a2a97 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Thu, 3 Oct 2024 23:25:11 +0200 Subject: [PATCH 107/125] (fix): Fix colour blending --- Backends/RmlUi_Renderer_DX11.cpp | 37 +++++++------------------------- 1 file changed, 8 insertions(+), 29 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 0a4d8f402..7037b5ce1 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -91,11 +91,6 @@ PS_INPUT VSMain(const VS_Input IN) result.color = IN.color; result.uv = IN.uv; -#if defined(RMLUI_PREMULTIPLIED_ALPHA) - // Pre-multiply vertex colors with their alpha. - result.color.rgb = result.color.rgb * result.color.a; -#endif - return result; }; )"; @@ -587,10 +582,9 @@ static bool CreateShader(ID3DBlob*& out_shader_dxil, ShaderType shader_type, con #endif // Compile shader - const D3D_SHADER_MACRO macros[] = {"RMLUI_PREMULTIPLIED_ALPHA", nullptr, nullptr, nullptr}; ID3DBlob* p_error_buff = nullptr; - HRESULT result = D3DCompile(code_string, code_length, nullptr, macros, nullptr, shader_type == ShaderType::Vertex ? "VSMain" : "PSMain", + HRESULT result = D3DCompile(code_string, code_length, nullptr, nullptr, nullptr, shader_type == ShaderType::Vertex ? "VSMain" : "PSMain", shader_type == ShaderType::Vertex ? "vs_5_0" : "ps_5_0", default_shader_flags, 0, &out_shader_dxil, &p_error_buff); RMLUI_DX_ASSERTMSG(result, "failed to D3DCompile"); #ifdef RMLUI_DX_DEBUG @@ -899,7 +893,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* #endif ZeroMemory(&blend_desc, sizeof(blend_desc)); - blend_desc.RenderTarget[0].BlendEnable = false; blend_desc.RenderTarget[0].RenderTargetWriteMask = 0; result = m_d3d_device->CreateBlendState(&blend_desc, &m_blend_state_disable_color); RMLUI_DX_ASSERTMSG(result, "failed to CreateBlendState"); @@ -912,10 +905,10 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* #endif blend_desc.RenderTarget[0].BlendEnable = true; - blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_SRC_COLOR; + blend_desc.RenderTarget[0].SrcBlend = D3D11_BLEND_BLEND_FACTOR; blend_desc.RenderTarget[0].DestBlend = D3D11_BLEND_ZERO; blend_desc.RenderTarget[0].BlendOp = D3D11_BLEND_OP_ADD; - blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_SRC_ALPHA; + blend_desc.RenderTarget[0].SrcBlendAlpha = D3D11_BLEND_BLEND_FACTOR; blend_desc.RenderTarget[0].DestBlendAlpha = D3D11_BLEND_ZERO; blend_desc.RenderTarget[0].BlendOpAlpha = D3D11_BLEND_OP_ADD; blend_desc.RenderTarget[0].RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL; @@ -1799,10 +1792,6 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, ID3D11RenderTargetView* original_render_target_view = nullptr; ID3D11DepthStencilView* original_depth_stencil_view = nullptr; m_d3d_context->OMGetRenderTargets(1, &original_render_target_view, &original_depth_stencil_view); - ID3D11BlendState* original_blend_state = nullptr; - float original_blend_factor[4]{}; - uint32_t original_blend_mask = 0; - m_d3d_context->OMGetBlendState(&original_blend_state, original_blend_factor, &original_blend_mask); // Bind temporary as the render target const Gfx::RenderTargetData& temporary_rt = m_render_layers.GetTemporary(); @@ -1812,7 +1801,7 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); // Disable blending - m_d3d_context->OMSetBlendState(nullptr, nullptr, 0xFFFFFFFF); + DisableBlend(); // Resolve from source to the temporary first m_d3d_context->ResolveSubresource(temporary_rt.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); @@ -1891,10 +1880,7 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, // Restore state m_d3d_context->OMSetRenderTargets(1, &original_render_target_view, original_depth_stencil_view); - m_d3d_context->OMSetBlendState(original_blend_state, original_blend_factor, original_blend_mask); - original_render_target_view->Release(); - original_depth_stencil_view->Release(); - original_blend_state->Release(); + SetBlendState(m_current_blend_state); d3dviewport.Width = static_cast(m_viewport_width); d3dviewport.Height = static_cast(m_viewport_height); @@ -2723,10 +2709,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMGetBlendState(&original_blend_state, original_blend_factor, &original_sample_mask); + ID3D11BlendState* blend_state_backup = m_current_blend_state; DisableBlend(); const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); @@ -2740,9 +2723,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetBlendState(original_blend_state, original_blend_factor, original_sample_mask); - original_blend_state->Release(); - m_current_blend_state_enabled = true; + SetBlendState(blend_state_backup); ID3D11ShaderResourceView* const nullSRV = nullptr; m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); @@ -2773,9 +2754,7 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetBlendState(original_blend_state, original_blend_factor, original_sample_mask); - original_blend_state->Release(); - m_current_blend_state_enabled = true; + SetBlendState(blend_state_backup); ID3D11ShaderResourceView* const nullSRV = nullptr; m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); From 48324d08186c4931f637cbf13ab9c065f494bd96 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 4 Oct 2024 00:33:37 +0200 Subject: [PATCH 108/125] (feat): Get blur to render something? Currently broken. Needs more work --- Backends/RmlUi_Renderer_DX11.cpp | 97 ++++++++++++-------------------- 1 file changed, 36 insertions(+), 61 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 7037b5ce1..35c4af59b 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -379,8 +379,12 @@ PS_INPUT VSMain(const VS_Input IN) { PS_INPUT result = (PS_INPUT)0; + // Flip UVs + // float2 flipped_uv = float2(IN.uv.x, 1.0f - IN.uv.y); + float2 flipped_uv = IN.uv; + for (int i = 0; i < BLUR_SIZE; i++) { - result.uv[i] = IN.uv - float(i - BLUR_NUM_WEIGHTS + 1) * m_texelOffset; + result.uv[i] = flipped_uv - float(i - BLUR_NUM_WEIGHTS + 1) * m_texelOffset; } result.position = float4(IN.position.xy, 1.0, 1.0); @@ -414,7 +418,7 @@ float4 PSMain(const PS_Input IN) : SV_TARGET for(int i = 0; i < BLUR_SIZE; i++) { float2 in_region = step(m_texCoordMin, IN.uv[i]) * step(IN.uv[i], m_texCoordMax); - color += g_InputTexture.Sample(g_SamplerLinear, IN.uv[i]) * in_region.x * in_region.y * m_weights[abs(i - BLUR_NUM_WEIGHTS + 1)]; + color += g_InputTexture.Sample(g_SamplerLinear, float2(IN.uv[i].x, 1.0f - IN.uv[i].y)) * in_region.x * in_region.y * m_weights[abs(i - BLUR_NUM_WEIGHTS + 1)]; } return color; }; @@ -1764,20 +1768,20 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, int dest_height = dstY1 - dstY0; // All coords must be greater than 0 - RMLUI_ASSERTMSG(srcX0 >= 0, "Invalid source coordinate (srcX0)!"); - RMLUI_ASSERTMSG(srcX1 >= 0, "Invalid source coordinate (srcX1)!"); - RMLUI_ASSERTMSG(srcY0 >= 0, "Invalid source coordinate (srcY0)!"); - RMLUI_ASSERTMSG(srcY1 >= 0, "Invalid source coordinate (srcY1)!"); - RMLUI_ASSERTMSG(dstX0 >= 0, "Invalid destination coordinate (dstX0)!"); - RMLUI_ASSERTMSG(dstX1 >= 0, "Invalid destination coordinate (dstX1)!"); - RMLUI_ASSERTMSG(dstY0 >= 0, "Invalid destination coordinate (dstY0)!"); - RMLUI_ASSERTMSG(dstY1 >= 0, "Invalid destination coordinate (dstY1)!"); + // RMLUI_ASSERTMSG(srcX0 >= 0, "Invalid source coordinate (srcX0)!"); + // RMLUI_ASSERTMSG(srcX1 >= 0, "Invalid source coordinate (srcX1)!"); + // RMLUI_ASSERTMSG(srcY0 >= 0, "Invalid source coordinate (srcY0)!"); + // RMLUI_ASSERTMSG(srcY1 >= 0, "Invalid source coordinate (srcY1)!"); + // RMLUI_ASSERTMSG(dstX0 >= 0, "Invalid destination coordinate (dstX0)!"); + // RMLUI_ASSERTMSG(dstX1 >= 0, "Invalid destination coordinate (dstX1)!"); + // RMLUI_ASSERTMSG(dstY0 >= 0, "Invalid destination coordinate (dstY0)!"); + // RMLUI_ASSERTMSG(dstY1 >= 0, "Invalid destination coordinate (dstY1)!"); // Width and height must be greater than 0 - RMLUI_ASSERTMSG(src_width != 0, "Invalid source rectangle (width)!"); - RMLUI_ASSERTMSG(src_height != 0, "Invalid source rectangle (height)!"); - RMLUI_ASSERTMSG(dest_width != 0, "Invalid destination rectangle (width)!"); - RMLUI_ASSERTMSG(dest_height != 0, "Invalid destination rectangle (height)!"); + // RMLUI_ASSERTMSG(src_width != 0, "Invalid source rectangle (width)!"); + // RMLUI_ASSERTMSG(src_height != 0, "Invalid source rectangle (height)!"); + // RMLUI_ASSERTMSG(dest_width != 0, "Invalid destination rectangle (width)!"); + // RMLUI_ASSERTMSG(dest_height != 0, "Invalid destination rectangle (height)!"); bool is_flipped = src_width < 0 || src_height < 0 || dest_width < 0 || dest_height < 0; bool is_stretched = src_width != dest_width || src_height != dest_height; @@ -2438,9 +2442,9 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& SetScissor(scissor, true); // Downscale by iterative half-scaling with bilinear filtering, to reduce aliasing. - D3D11_VIEWPORT d3dviewport; + D3D11_VIEWPORT d3dviewport{}; d3dviewport.TopLeftX = 0; - d3dviewport.TopLeftY = 0; + d3dviewport.TopLeftY = static_cast(source_destination.height / 2); d3dviewport.Width = static_cast(source_destination.width / 2); d3dviewport.Height = static_cast(source_destination.height / 2); d3dviewport.MinDepth = 0.0f; @@ -2457,8 +2461,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& scissor.p0 = (scissor.p0 + Rml::Vector2i(1)) / 2; scissor.p1 = Rml::Math::Max(scissor.p1 / 2, scissor.p0); const bool from_source = (i % 2 == 0); - Gfx::BindTexture(m_d3d_context, from_source ? source_destination : temp); Gfx::BindRenderTarget(m_d3d_context, (from_source ? temp : source_destination)); + Gfx::BindTexture(m_d3d_context, from_source ? source_destination : temp); SetScissor(scissor, true); DrawFullscreenQuad({}, uv_scaling); @@ -2476,11 +2480,13 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& const bool transfer_to_temp_buffer = (pass_level % 2 == 0); if (transfer_to_temp_buffer) { - Gfx::BindTexture(m_d3d_context, source_destination); Gfx::BindRenderTarget(m_d3d_context, temp); + Gfx::BindTexture(m_d3d_context, source_destination); DrawFullscreenQuad(); } + // Set up uniforms. + UseProgram(ProgramId::Blur); D3D11_MAPPED_SUBRESOURCE mappedResource{}; // Lock the constant buffer so it can be written to. HRESULT result = m_d3d_context->Map(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); @@ -2497,31 +2503,26 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& dataPtr->common.translation = m_translation; SetBlurWeights(dataPtr->blur.weights, sigma); - - // Set up uniforms. - UseProgram(ProgramId::Blur); - SetBlurWeights(dataPtr->blur.weights, sigma); - SetTexCoordLimits(dataPtr->blur.texcoord_min, dataPtr->blur.texcoord_max, scissor, - {source_destination.width, source_destination.height}); + SetTexCoordLimits(dataPtr->blur.texcoord_min, dataPtr->blur.texcoord_max, scissor, {source_destination.width, source_destination.height}); auto SetTexelOffset = [dataPtr](Rml::Vector2f blur_direction, int texture_dimension) { const Rml::Vector2f texel_offset = blur_direction * (1.0f / float(texture_dimension)); dataPtr->blur.texel_offset = texel_offset; }; + SetTexelOffset({0.f, 1.f}, temp.height); + // Upload to the GPU. m_d3d_context->Unmap(m_shader_buffer, 0); // Blur render pass - vertical. - Gfx::BindTexture(m_d3d_context, temp); Gfx::BindRenderTarget(m_d3d_context, source_destination); - - SetTexelOffset({0.f, 1.f}, temp.height); + Gfx::BindTexture(m_d3d_context, temp); DrawFullscreenQuad(); // Blur render pass - horizontal. - Gfx::BindTexture(m_d3d_context, source_destination); Gfx::BindRenderTarget(m_d3d_context, temp); + Gfx::BindTexture(m_d3d_context, source_destination); // Add a 1px transparent border around the blur region by first clearing with a padded scissor. This helps prevent // artifacts when upscaling the blur result in the later step. On Intel and AMD, we have observed that during @@ -2529,8 +2530,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // hand, it looks like Nvidia clamps the pixels to the source edge, which is what we really want. Regardless, we // work around the issue with this extra step. SetScissor(scissor.Extend(1), true); - float clearColor[4] = {0.0f, 0.0f, 0.0f, 0.0f}; - m_d3d_context->ClearRenderTargetView(temp.render_target_view, clearColor); + FLOAT clear_color[4] = {0.0f, 0.0f, 0.0f, 0.0f}; + m_d3d_context->ClearRenderTargetView(temp.render_target_view, clear_color); SetScissor(scissor, true); SetTexelOffset({1.f, 0.f}, source_destination.width); @@ -2543,17 +2544,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& const Rml::Vector2i src_max = scissor.p1; const Rml::Vector2i dst_min = window_flipped.p0; const Rml::Vector2i dst_max = window_flipped.p1; - - D3D11_BOX srcBox; - srcBox.left = src_min.x; - srcBox.top = src_min.y; - srcBox.front = 0; - srcBox.right = src_max.x; - srcBox.bottom = src_max.y; - srcBox.back = 1; - - m_d3d_context->CopySubresourceRegion(source_destination.render_target_texture, 0, dst_min.x, dst_min.y, 0, temp.render_target_texture, 0, - &srcBox); + BlitRenderTarget(temp, source_destination, src_min.x, src_max.y, src_max.x, src_min.y, dst_min.x, dst_min.y, dst_max.x, dst_max.y); // The above upscale blit might be jittery at low resolutions (large pass levels). This is especially noticeable when moving an element with // backdrop blur around or when trying to click/hover an element within a blurred region since it may be rendered at an offset. For more stable @@ -2564,28 +2555,14 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& const Rml::Vector2i target_max = src_max * (1 << pass_level); if (target_min != dst_min || target_max != dst_max) { - Gfx::BindTexture(m_d3d_context, temp); - - D3D11_VIEWPORT viewport; - viewport.TopLeftX = static_cast(target_min.x); - viewport.TopLeftY = static_cast(target_min.y); - viewport.Width = static_cast(target_max.x - target_min.x); - viewport.Height = static_cast(target_max.y - target_min.y); - viewport.MinDepth = 0.0f; - viewport.MaxDepth = 1.0f; - m_d3d_context->RSSetViewports(1, &viewport); - UseProgram(ProgramId::Passthrough); - DrawFullscreenQuad(); - // Restore viewport - m_d3d_context->RSSetViewports(1, &d3dviewport); - - - // glBlitFramebuffer(src_min.x, src_min.y, src_max.x, src_max.y, target_min.x, target_min.y, target_max.x, target_max.y, GL_COLOR_BUFFER_BIT, - // GL_LINEAR); + BlitRenderTarget(temp, source_destination, src_min.x, src_min.y, src_max.x, src_max.y, target_min.x, target_min.y, target_max.x, target_max.y); } // Restore render state. SetScissor(original_scissor); + + ID3D11ShaderResourceView* const nullSRV = nullptr; + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); } void RenderInterface_DX11::RenderFilters(Rml::Span filter_handles) @@ -2626,7 +2603,6 @@ void RenderInterface_DX11::RenderFilters(Rml::Span Date: Fri, 4 Oct 2024 13:06:48 +0200 Subject: [PATCH 109/125] (fix): Fix memory leak --- Backends/RmlUi_Renderer_DX11.cpp | 2 ++ 1 file changed, 2 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 35c4af59b..87da7c405 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1890,6 +1890,8 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, d3dviewport.Height = static_cast(m_viewport_height); m_d3d_context->RSSetViewports(1, &d3dviewport); m_d3d_context->PSSetShaderResources(0, 2, null_shader_resource_views); + DX_CLEANUP_RESOURCE_IF_CREATED(original_render_target_view); + DX_CLEANUP_RESOURCE_IF_CREATED(original_depth_stencil_view); } else { From b839bffe251d1773130be08836490c1fd6a5ba9e Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Fri, 4 Oct 2024 13:07:24 +0200 Subject: [PATCH 110/125] (fix): Fix blur flipping image on large kernels --- Backends/RmlUi_Renderer_DX11.cpp | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 87da7c405..e18f340cd 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2557,7 +2557,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& const Rml::Vector2i target_max = src_max * (1 << pass_level); if (target_min != dst_min || target_max != dst_max) { - BlitRenderTarget(temp, source_destination, src_min.x, src_min.y, src_max.x, src_max.y, target_min.x, target_min.y, target_max.x, target_max.y); + BlitRenderTarget(temp, source_destination, src_min.x, src_max.y, src_max.x, src_min.y, target_min.x, target_min.y, target_max.x, + target_max.y); } // Restore render state. From 19ddcb3d36c08a197b8a68471b55328c4cdd18e8 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 5 Oct 2024 14:39:49 +0200 Subject: [PATCH 111/125] (fix): Correct blur implementation --- Backends/RmlUi_Backend_Win32_DX11.cpp | 2 +- Backends/RmlUi_Renderer_DX11.cpp | 136 ++++++++++++++++---------- Backends/RmlUi_Renderer_DX11.h | 8 +- 3 files changed, 94 insertions(+), 52 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index 92353d95f..f1f6a1703 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -178,7 +178,7 @@ bool Backend::Initialize(const char* window_name, int width, int height, bool al stbi_convert_iphone_png_to_rgb(true); #endif - data->render_interface.Init(data->d3d_resources.pd3dDevice, data->d3d_resources.pd3dDeviceContext); + data->render_interface.Init(data->d3d_resources.pd3dDevice); data->window_handle = window_handle; data->system_interface.SetWindow(window_handle); diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e18f340cd..09eb9498e 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -379,12 +379,8 @@ PS_INPUT VSMain(const VS_Input IN) { PS_INPUT result = (PS_INPUT)0; - // Flip UVs - // float2 flipped_uv = float2(IN.uv.x, 1.0f - IN.uv.y); - float2 flipped_uv = IN.uv; - for (int i = 0; i < BLUR_SIZE; i++) { - result.uv[i] = flipped_uv - float(i - BLUR_NUM_WEIGHTS + 1) * m_texelOffset; + result.uv[i] = IN.uv - float(i - BLUR_NUM_WEIGHTS + 1) * m_texelOffset; } result.position = float4(IN.position.xy, 1.0, 1.0); @@ -396,7 +392,7 @@ static const char shader_frag_blur[] = RMLUI_SHADER_BLUR_HEADER R"( Texture2D g_InputTexture : register(t0); SamplerState g_SamplerLinear : register(s0); -cbuffer ConstantBuffer : register(b0) +cbuffer SharedConstantBuffer : register(b0) { float4x4 m_transform; float2 m_translate; @@ -853,14 +849,14 @@ static uintptr_t HashPointer(uintptr_t inPtr) RenderInterface_DX11::RenderInterface_DX11() {} RenderInterface_DX11::~RenderInterface_DX11() {} -void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context) +void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device) { RMLUI_ASSERTMSG(p_d3d_device, "p_d3d_device cannot be nullptr!"); - RMLUI_ASSERTMSG(p_d3d_device_context, "p_d3d_device_context cannot be nullptr!"); // Assign D3D resources m_d3d_device = p_d3d_device; - m_d3d_context = p_d3d_device_context; + m_d3d_device->QueryInterface(__uuidof(ID3D11Device1), (void**)&m_d3d_device_1); + m_d3d_device_1->GetImmediateContext1(&m_d3d_context); m_d3d_context->QueryInterface(__uuidof(ID3DUserDefinedAnnotation), (void**)&m_debug); DebugScope::AttachDebugLayer(m_debug); m_render_layers.SetD3DResources(m_d3d_device); @@ -1007,7 +1003,12 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* return; } + cbufferDesc.Usage = D3D11_USAGE_DYNAMIC; cbufferDesc.ByteWidth = sizeof(ColorMatrixCbuffer); + cbufferDesc.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + cbufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + cbufferDesc.MiscFlags = 0; + cbufferDesc.StructureByteStride = 0; result = m_d3d_device->CreateBuffer(&cbufferDesc, nullptr, &m_color_matrix_cbuffer); RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); @@ -1015,6 +1016,20 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* { return; } + + cbufferDesc.Usage = D3D11_USAGE_DYNAMIC; + cbufferDesc.ByteWidth = sizeof(ShaderCbuffer::Blur); + cbufferDesc.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + cbufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + cbufferDesc.MiscFlags = 0; + cbufferDesc.StructureByteStride = 0; + + result = m_d3d_device->CreateBuffer(&cbufferDesc, nullptr, &m_blur_cbuffer); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); + if (FAILED(result)) + { + return; + } } // Create sampler state for textures @@ -1116,6 +1131,7 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_rasterizer_state_scissor_enabled); DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_color_matrix_cbuffer); + DX_CLEANUP_RESOURCE_IF_CREATED(m_blur_cbuffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); DX_CLEANUP_RESOURCE_IF_CREATED(m_debug); } @@ -1396,7 +1412,7 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm m_cbuffer_dirty = true; } - if (texture == TexturePostprocess) + if (texture == TexturePostprocess || texture == TexturePostprocessNoBinding) { // Do nothing. } @@ -1417,15 +1433,18 @@ void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rm UseProgram(ProgramId::Color); } - UpdateConstantBuffer(); + if (texture != TexturePostprocessNoBinding) + { + UpdateConstantBuffer(); - m_d3d_context->IASetInputLayout(m_vertex_layout); - m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); - m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); + m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); + m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); + } // Bind vertex and index buffers, issue draw call uint32_t stride = sizeof(Rml::Vertex); uint32_t offset = 0; + m_d3d_context->IASetInputLayout(m_vertex_layout); m_d3d_context->IASetVertexBuffers(0, 1, &geometryData.vertex_buffer, &stride, &offset); m_d3d_context->IASetIndexBuffer(geometryData.index_buffer, DXGI_FORMAT_R32_UINT, 0); m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); @@ -1767,22 +1786,6 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, int dest_width = dstX1 - dstX0; int dest_height = dstY1 - dstY0; - // All coords must be greater than 0 - // RMLUI_ASSERTMSG(srcX0 >= 0, "Invalid source coordinate (srcX0)!"); - // RMLUI_ASSERTMSG(srcX1 >= 0, "Invalid source coordinate (srcX1)!"); - // RMLUI_ASSERTMSG(srcY0 >= 0, "Invalid source coordinate (srcY0)!"); - // RMLUI_ASSERTMSG(srcY1 >= 0, "Invalid source coordinate (srcY1)!"); - // RMLUI_ASSERTMSG(dstX0 >= 0, "Invalid destination coordinate (dstX0)!"); - // RMLUI_ASSERTMSG(dstX1 >= 0, "Invalid destination coordinate (dstX1)!"); - // RMLUI_ASSERTMSG(dstY0 >= 0, "Invalid destination coordinate (dstY0)!"); - // RMLUI_ASSERTMSG(dstY1 >= 0, "Invalid destination coordinate (dstY1)!"); - - // Width and height must be greater than 0 - // RMLUI_ASSERTMSG(src_width != 0, "Invalid source rectangle (width)!"); - // RMLUI_ASSERTMSG(src_height != 0, "Invalid source rectangle (height)!"); - // RMLUI_ASSERTMSG(dest_width != 0, "Invalid destination rectangle (width)!"); - // RMLUI_ASSERTMSG(dest_height != 0, "Invalid destination rectangle (height)!"); - bool is_flipped = src_width < 0 || src_height < 0 || dest_width < 0 || dest_height < 0; bool is_stretched = src_width != dest_width || src_height != dest_height; bool is_full_copy = src_width == dest_width && src_height == dest_height && srcX0 == 0 && srcY0 == 0 && dstX0 == 0 && dstY0 == 0; @@ -2432,6 +2435,12 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& RMLUI_ASSERT(&source_destination != &temp && source_destination.width == temp.width && source_destination.height == temp.height); RMLUI_ASSERT(window_flipped.Valid()); + if (window_flipped.Top() == window_flipped.Bottom() || window_flipped.Left() == window_flipped.Right()) + { + // In some cases the bounds are 0 width, so early exit since there's nothing to blur + return; + } + int pass_level = 0; SigmaToParameters(sigma, pass_level, sigma); @@ -2500,22 +2509,17 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // Get a pointer to the data in the constant buffer. ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; - // Copy the data to the GPU - dataPtr->common.transform = m_transform; - dataPtr->common.translation = m_translation; - SetBlurWeights(dataPtr->blur.weights, sigma); SetTexCoordLimits(dataPtr->blur.texcoord_min, dataPtr->blur.texcoord_max, scissor, {source_destination.width, source_destination.height}); - auto SetTexelOffset = [dataPtr](Rml::Vector2f blur_direction, int texture_dimension) { - const Rml::Vector2f texel_offset = blur_direction * (1.0f / float(texture_dimension)); - dataPtr->blur.texel_offset = texel_offset; - }; - - SetTexelOffset({0.f, 1.f}, temp.height); + Rml::Vector2f texel_offset = Rml::Vector2f(0.f, 1.f) * (1.0f / float(temp.height)); + dataPtr->blur.texel_offset = texel_offset; // Upload to the GPU. m_d3d_context->Unmap(m_shader_buffer, 0); + // Unbind second pass + ID3D11Buffer* null_buffer = nullptr; + m_d3d_context->VSSetConstantBuffers(1, 1, &null_buffer); // Blur render pass - vertical. Gfx::BindRenderTarget(m_d3d_context, source_destination); @@ -2532,21 +2536,55 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // hand, it looks like Nvidia clamps the pixels to the source edge, which is what we really want. Regardless, we // work around the issue with this extra step. SetScissor(scissor.Extend(1), true); + // D3D11 does not support clearing based on scissor rects. So we use a D3D11Context1 to use ClearView, and define the D3D11_RECt here + Rml::Rectanglei scissor_ext = scissor.Extend(1); + scissor_ext = VerticallyFlipped(scissor_ext, m_viewport_height); + + // Some render APIs don't like offscreen positions (WebGL in particular), so clamp them to the viewport. + const int x_min = Rml::Math::Clamp(scissor_ext.Left(), 0, m_viewport_width); + const int y_min = Rml::Math::Clamp(scissor_ext.Top(), 0, m_viewport_height); + const int x_max = Rml::Math::Clamp(scissor_ext.Right(), 0, m_viewport_width); + const int y_max = Rml::Math::Clamp(scissor_ext.Bottom(), 0, m_viewport_height); + + D3D11_RECT rect_scissor = {}; + rect_scissor.left = x_min; + rect_scissor.top = y_min; + rect_scissor.right = x_max; + rect_scissor.bottom = y_max; + FLOAT clear_color[4] = {0.0f, 0.0f, 0.0f, 0.0f}; - m_d3d_context->ClearRenderTargetView(temp.render_target_view, clear_color); + m_d3d_context->ClearView(temp.render_target_view, clear_color, &rect_scissor, 1); SetScissor(scissor, true); - SetTexelOffset({1.f, 0.f}, source_destination.width); - DrawFullscreenQuad(); + // Update texel offset for upscale pass + result = m_d3d_context->Map(m_blur_cbuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + if (FAILED(result)) + { + ID3D11ShaderResourceView* const nullSRV = nullptr; + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); + return; + } + + ShaderCbuffer::Blur* dataPtrBlur = (ShaderCbuffer::Blur*)mappedResource.pData; + + texel_offset = Rml::Vector2f(1.f, 0.f) * (1.0f / float(source_destination.width)); + dataPtrBlur->texel_offset = texel_offset; + + // Upload to the GPU. + m_d3d_context->Unmap(m_blur_cbuffer, 0); + m_d3d_context->VSSetConstantBuffers(0, 1, &m_blur_cbuffer); + + RenderGeometry(m_fullscreen_quad_geometry, {}, RenderInterface_DX11::TexturePostprocessNoBinding); // Blit the blurred image to the scissor region with upscaling. - SetScissor(window_flipped, true); + Rml::Rectanglei window_flipped_twice = VerticallyFlipped(window_flipped, m_viewport_height); + SetScissor(window_flipped_twice, false); const Rml::Vector2i src_min = scissor.p0; const Rml::Vector2i src_max = scissor.p1; - const Rml::Vector2i dst_min = window_flipped.p0; - const Rml::Vector2i dst_max = window_flipped.p1; - BlitRenderTarget(temp, source_destination, src_min.x, src_max.y, src_max.x, src_min.y, dst_min.x, dst_min.y, dst_max.x, dst_max.y); + const Rml::Vector2i dst_min = window_flipped_twice.p0; + const Rml::Vector2i dst_max = window_flipped_twice.p1; + BlitRenderTarget(temp, source_destination, src_min.x, src_min.y, src_max.x, src_max.y, dst_min.x, dst_max.y, dst_max.x, dst_min.y); // The above upscale blit might be jittery at low resolutions (large pass levels). This is especially noticeable when moving an element with // backdrop blur around or when trying to click/hover an element within a blurred region since it may be rendered at an offset. For more stable @@ -2557,8 +2595,8 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& const Rml::Vector2i target_max = src_max * (1 << pass_level); if (target_min != dst_min || target_max != dst_max) { - BlitRenderTarget(temp, source_destination, src_min.x, src_max.y, src_max.x, src_min.y, target_min.x, target_min.y, target_max.x, - target_max.y); + BlitRenderTarget(temp, source_destination, src_min.x, src_min.y, src_max.x, src_max.y, target_min.x, target_max.y, target_max.x, + target_min.y); } // Restore render state. diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index ef0138dc5..c0a5be9d8 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -78,7 +78,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ~RenderInterface_DX11(); // Resource initialisation and cleanup - void Init(ID3D11Device* p_d3d_device, ID3D11DeviceContext* p_d3d_device_context); + void Init(ID3D11Device* p_d3d_device); void Cleanup (); // Sets up DirectX11 states for taking rendering commands from RmlUi. @@ -127,6 +127,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { static constexpr Rml::TextureHandle TextureEnableWithoutBinding = Rml::TextureHandle(-1); // Can be passed to RenderGeometry() to leave the bound texture and used program unchanged. static constexpr Rml::TextureHandle TexturePostprocess = Rml::TextureHandle(-2); + // Can be passed to RenderGeometry() to not touch the cbuffers + static constexpr Rml::TextureHandle TexturePostprocessNoBinding = Rml::TextureHandle(-3); public: /// Called by the renderer when it wants to load a texture from disk. @@ -164,7 +166,8 @@ class RenderInterface_DX11 : public Rml::RenderInterface { private: // D3D11 core resources ID3D11Device* m_d3d_device = nullptr; - ID3D11DeviceContext* m_d3d_context = nullptr; + ID3D11Device1* m_d3d_device_1 = nullptr; + ID3D11DeviceContext1* m_d3d_context = nullptr; IDXGISwapChain* m_bound_swapchain = nullptr; ID3D11RenderTargetView* m_bound_render_target = nullptr; bool m_scissor_enabled = false; @@ -183,6 +186,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { // Shaders ID3D11Buffer* m_shader_buffer = nullptr; ID3D11Buffer* m_color_matrix_cbuffer = nullptr; + ID3D11Buffer* m_blur_cbuffer = nullptr; bool m_cbuffer_dirty = true; ID3D11SamplerState* m_sampler_state = nullptr; From 1e83f4a0e4adea00949e846c51a62bc642afd4a4 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 5 Oct 2024 14:43:03 +0200 Subject: [PATCH 112/125] (fix): Cleanup some DX11 resources which were being leaked --- Backends/RmlUi_Renderer_DX11.cpp | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 09eb9498e..83783e4f3 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -855,6 +855,7 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device) // Assign D3D resources m_d3d_device = p_d3d_device; + m_d3d_device->AddRef(); m_d3d_device->QueryInterface(__uuidof(ID3D11Device1), (void**)&m_d3d_device_1); m_d3d_device_1->GetImmediateContext1(&m_d3d_context); m_d3d_context->QueryInterface(__uuidof(ID3DUserDefinedAnnotation), (void**)&m_debug); @@ -1134,6 +1135,9 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_blur_cbuffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); DX_CLEANUP_RESOURCE_IF_CREATED(m_debug); + DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_context); + DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_device_1); + DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_device); } void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) From aa444567c2828e20bdfa26feba92eccba59e2a3f Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 5 Oct 2024 16:53:56 +0200 Subject: [PATCH 113/125] (fix): Improve blur implementation to correct UV offset Also add early exit in the case of a kernel of size 0 --- Backends/RmlUi_Renderer_DX11.cpp | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 83783e4f3..602ac3503 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2447,6 +2447,11 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& int pass_level = 0; SigmaToParameters(sigma, pass_level, sigma); + if (sigma == 0) + { + // If sigma is zero we don't have anything to blur, so early exit to save performance + return; + } const Rml::Rectanglei original_scissor = m_scissor_state; @@ -2595,8 +2600,11 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // and accurate rendering we next upscale the blur image by an exact power-of-two. However, this may not fill the edges completely so we need to // do the above first. Note that this strategy may sometimes result in visible seams. Alternatively, we could try to enlarge the window to the // next power-of-two size and then downsample and blur that. - const Rml::Vector2i target_min = src_min * (1 << pass_level); - const Rml::Vector2i target_max = src_max * (1 << pass_level); + + // We need to flip the scissor rect to have it match DX11 coords + window_flipped_twice = VerticallyFlipped(scissor, m_viewport_height); + const Rml::Vector2i target_min = window_flipped_twice.p0 * (1 << pass_level); + const Rml::Vector2i target_max = window_flipped_twice.p1 * (1 << pass_level); if (target_min != dst_min || target_max != dst_max) { BlitRenderTarget(temp, source_destination, src_min.x, src_min.y, src_max.x, src_max.y, target_min.x, target_max.y, target_max.x, From 1dfd0c9fe0ed03c7cec9bac931a641ec20d6bd93 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 5 Oct 2024 16:59:01 +0200 Subject: [PATCH 114/125] (feat): Add early exit cases for blurring to GL3 --- Backends/RmlUi_Renderer_DX11.cpp | 2 +- Backends/RmlUi_Renderer_GL3.cpp | 11 +++++++++++ 2 files changed, 12 insertions(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 602ac3503..b2d8e5941 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2449,7 +2449,7 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& SigmaToParameters(sigma, pass_level, sigma); if (sigma == 0) { - // If sigma is zero we don't have anything to blur, so early exit to save performance + // If sigma is zero we don't have anything to blur, so early exit since blurring will have no effect on the final image return; } diff --git a/Backends/RmlUi_Renderer_GL3.cpp b/Backends/RmlUi_Renderer_GL3.cpp index 51f331701..0bb540523 100644 --- a/Backends/RmlUi_Renderer_GL3.cpp +++ b/Backends/RmlUi_Renderer_GL3.cpp @@ -1367,8 +1367,19 @@ void RenderInterface_GL3::RenderBlur(float sigma, const Gfx::FramebufferData& so RMLUI_ASSERT(&source_destination != &temp && source_destination.width == temp.width && source_destination.height == temp.height); RMLUI_ASSERT(window_flipped.Valid()); + if (window_flipped.Top() == window_flipped.Bottom() || window_flipped.Left() == window_flipped.Right()) + { + // In some cases the bounds are 0 width, so early exit since there's nothing to blur + return; + } + int pass_level = 0; SigmaToParameters(sigma, pass_level, sigma); + if (sigma == 0) + { + // If sigma is zero we don't have anything to blur, so early exit since blurring will have no effect on the final image + return; + } const Rml::Rectanglei original_scissor = scissor_state; From be849b851e840de374effaf1d6b6fcb099b3bfb5 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 6 Oct 2024 02:19:55 +0200 Subject: [PATCH 115/125] (fix): Blurring on specific kernels is incorrect --- Backends/RmlUi_Renderer_DX11.cpp | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index b2d8e5941..93ea38108 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2602,13 +2602,15 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // next power-of-two size and then downsample and blur that. // We need to flip the scissor rect to have it match DX11 coords - window_flipped_twice = VerticallyFlipped(scissor, m_viewport_height); - const Rml::Vector2i target_min = window_flipped_twice.p0 * (1 << pass_level); - const Rml::Vector2i target_max = window_flipped_twice.p1 * (1 << pass_level); + const Rml::Vector2i target_min = scissor.p0 * (1 << pass_level); + const Rml::Vector2i target_max = scissor.p1 * (1 << pass_level); if (target_min != dst_min || target_max != dst_max) { - BlitRenderTarget(temp, source_destination, src_min.x, src_min.y, src_max.x, src_max.y, target_min.x, target_max.y, target_max.x, - target_min.y); + window_flipped_twice = VerticallyFlipped(scissor, m_viewport_height); + const Rml::Vector2i target_min_blit = window_flipped_twice.p0 * (1 << pass_level); + const Rml::Vector2i target_max_blit = window_flipped_twice.p1 * (1 << pass_level); + BlitRenderTarget(temp, source_destination, src_min.x, src_min.y, src_max.x, src_max.y, target_min_blit.x, target_max_blit.y, + target_max_blit.x, target_min_blit.y); } // Restore render state. From 6249f87311a82a0ae1aea6e473783d5b96650293 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 6 Oct 2024 02:20:19 +0200 Subject: [PATCH 116/125] (chore): Remove redundant name from type --- Backends/RmlUi_Renderer_DX11.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index c0a5be9d8..ffec04290 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -284,7 +284,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { }; struct ColorMatrixCbuffer { Rml::Matrix4f color_matrix; - } color_matrix; + }; #pragma pack() Rml::CompiledGeometryHandle m_fullscreen_quad_geometry = 0; From 2f7f963396aee596bb38b4191c18e5ffaa997fc7 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 6 Oct 2024 20:25:55 +0200 Subject: [PATCH 117/125] (feat): Implement drop shadows --- Backends/RmlUi_Renderer_DX11.cpp | 42 +++++++++++++++++++++----------- Backends/RmlUi_Renderer_DX11.h | 2 +- 2 files changed, 29 insertions(+), 15 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 93ea38108..d2731bfeb 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -424,10 +424,8 @@ static const char shader_frag_drop_shadow[] = RMLUI_SHADER_HEADER R"( Texture2D g_InputTexture : register(t0); SamplerState g_SamplerLinear : register(s0); -cbuffer ConstantBuffer : register(b0) +cbuffer DropShadowBuffer : register(b1) { - float4x4 m_transform; - float2 m_translate; float2 m_texCoordMin; float2 m_texCoordMax; float4 m_color; @@ -1031,6 +1029,20 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device) { return; } + + cbufferDesc.Usage = D3D11_USAGE_DYNAMIC; + cbufferDesc.ByteWidth = sizeof(ShaderCbuffer::DropShadow); + cbufferDesc.BindFlags = D3D11_BIND_CONSTANT_BUFFER; + cbufferDesc.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE; + cbufferDesc.MiscFlags = 0; + cbufferDesc.StructureByteStride = 0; + + result = m_d3d_device->CreateBuffer(&cbufferDesc, nullptr, &m_drop_shadow_cbuffer); + RMLUI_DX_ASSERTMSG(result, "failed to CreateBuffer"); + if (FAILED(result)) + { + return; + } } // Create sampler state for textures @@ -1133,6 +1145,7 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_shader_buffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_color_matrix_cbuffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_blur_cbuffer); + DX_CLEANUP_RESOURCE_IF_CREATED(m_drop_shadow_cbuffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); DX_CLEANUP_RESOURCE_IF_CREATED(m_debug); DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_context); @@ -2672,7 +2685,6 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanMap(m_shader_buffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); + HRESULT result = m_d3d_context->Map(m_drop_shadow_cbuffer, 0, D3D11_MAP_WRITE_DISCARD, 0, &mappedResource); if (FAILED(result)) { return; } // Get a pointer to the data in the constant buffer. - ShaderCbuffer* dataPtr = (ShaderCbuffer*)mappedResource.pData; + ShaderCbuffer::DropShadow* dataPtr = (ShaderCbuffer::DropShadow*)mappedResource.pData; Rml::Colourf color = ConvertToColorf(filter.color); // Copy the data to the GPU - dataPtr->common.transform = m_transform; - dataPtr->common.translation = m_translation; - dataPtr->drop_shadow.color.x = color.red; - dataPtr->drop_shadow.color.y = color.green; - dataPtr->drop_shadow.color.z = color.blue; - dataPtr->drop_shadow.color.w = color.alpha; + dataPtr->color.x = color.red; + dataPtr->color.y = color.green; + dataPtr->color.z = color.blue; + dataPtr->color.w = color.alpha; const Rml::Rectanglei window_flipped = VerticallyFlipped(m_scissor_state, m_viewport_height); - SetTexCoordLimits(dataPtr->drop_shadow.texcoord_min, dataPtr->drop_shadow.texcoord_max, window_flipped, + SetTexCoordLimits(dataPtr->texcoord_min, dataPtr->texcoord_max, m_scissor_state, {primary.width, primary.height}); // Upload to the GPU. - m_d3d_context->Unmap(m_shader_buffer, 0); + m_d3d_context->Unmap(m_drop_shadow_cbuffer, 0); + m_d3d_context->PSSetConstantBuffers(1, 1, &m_drop_shadow_cbuffer); m_cbuffer_dirty = true; @@ -2724,6 +2735,8 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanPSSetShaderResources(0, 1, &nullSRV); + m_d3d_context->PSSetShaderResources(1, 1, &nullSRV); } break; case FilterType::ColorMatrix: diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index ffec04290..efcd1cfab 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -187,6 +187,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11Buffer* m_shader_buffer = nullptr; ID3D11Buffer* m_color_matrix_cbuffer = nullptr; ID3D11Buffer* m_blur_cbuffer = nullptr; + ID3D11Buffer* m_drop_shadow_cbuffer = nullptr; bool m_cbuffer_dirty = true; ID3D11SamplerState* m_sampler_state = nullptr; @@ -271,7 +272,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { Rml::Vector2f texcoord_max; } blur; struct DropShadow { - int _padding[18]; Rml::Vector2f texcoord_min; Rml::Vector2f texcoord_max; Rml::Vector4f color; From 353a818429b0f4a65928ddaa7b4c48e10b034706 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 6 Oct 2024 20:33:32 +0200 Subject: [PATCH 118/125] (fix): Correct blur de-jitter --- Backends/RmlUi_Renderer_DX11.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index d2731bfeb..e4ed3fd2b 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -2615,9 +2615,9 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& // next power-of-two size and then downsample and blur that. // We need to flip the scissor rect to have it match DX11 coords - const Rml::Vector2i target_min = scissor.p0 * (1 << pass_level); - const Rml::Vector2i target_max = scissor.p1 * (1 << pass_level); - if (target_min != dst_min || target_max != dst_max) + const Rml::Vector2i target_min = src_min * (1 << pass_level); + const Rml::Vector2i target_max = src_max * (1 << pass_level); + if (target_min != window_flipped.p0 || target_max != window_flipped.p1) { window_flipped_twice = VerticallyFlipped(scissor, m_viewport_height); const Rml::Vector2i target_min_blit = window_flipped_twice.p0 * (1 << pass_level); From 1b8655a94e1a9a35e78867cfd524d6e787e492e3 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 6 Oct 2024 20:33:48 +0200 Subject: [PATCH 119/125] (chore): Remove debug code --- Backends/RmlUi_Renderer_DX11.cpp | 49 +------------------------------- Backends/RmlUi_Renderer_DX11.h | 11 ------- 2 files changed, 1 insertion(+), 59 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e4ed3fd2b..ac1e26e3e 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -856,8 +856,6 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device) m_d3d_device->AddRef(); m_d3d_device->QueryInterface(__uuidof(ID3D11Device1), (void**)&m_d3d_device_1); m_d3d_device_1->GetImmediateContext1(&m_d3d_context); - m_d3d_context->QueryInterface(__uuidof(ID3DUserDefinedAnnotation), (void**)&m_debug); - DebugScope::AttachDebugLayer(m_debug); m_render_layers.SetD3DResources(m_d3d_device); // Pre-cache quad for blitting @@ -1147,7 +1145,6 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_blur_cbuffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_drop_shadow_cbuffer); DX_CLEANUP_RESOURCE_IF_CREATED(m_vertex_layout); - DX_CLEANUP_RESOURCE_IF_CREATED(m_debug); DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_context); DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_device_1); DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_device); @@ -1796,8 +1793,6 @@ void RenderInterface_DX11::SetTransform(const Rml::Matrix4f* new_transform) void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, const Gfx::RenderTargetData& dest, int srcX0, int srcY0, int srcX1, int srcY1, int dstX0, int dstY0, int dstX1, int dstY1) { - DebugScope scope(L"BlitRenderTarget"); - int src_width = srcX1 - srcX0; int src_height = srcY1 - srcY0; int dest_width = dstX1 - dstX0; @@ -1809,7 +1804,6 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, if (is_flipped || is_stretched || !is_full_copy) { - DebugScope scope2(L"Complex"); // Full draw call. Slow path because no equivalent in DX11 // Cache current state @@ -1915,7 +1909,6 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, } else { - DebugScope scope2(L"Simple"); // Resolve and move on m_d3d_context->ResolveSubresource(dest.render_target_texture, 0, source.render_target_texture, 0, DXGI_FORMAT_R8G8B8A8_UNORM); } @@ -1923,7 +1916,6 @@ void RenderInterface_DX11::BlitRenderTarget(const Gfx::RenderTargetData& source, void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_handle) { - DebugScope scope(L"BlitLayerToPostprocessPrimary"); const Gfx::RenderTargetData& source = m_render_layers.GetLayer(layer_handle); const Gfx::RenderTargetData& destination = m_render_layers.GetPostprocessPrimary(); @@ -1933,7 +1925,6 @@ void RenderInterface_DX11::BlitLayerToPostprocessPrimary(Rml::LayerHandle layer_ Rml::LayerHandle RenderInterface_DX11::PushLayer() { - m_debug->BeginEvent(L"BeginLayer"); const Rml::LayerHandle layer_handle = m_render_layers.PushLayer(); Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); @@ -1946,7 +1937,6 @@ Rml::LayerHandle RenderInterface_DX11::PushLayer() void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml::LayerHandle destination_handle, Rml::BlendMode blend_mode, Rml::Span filters) { - DebugScope scope(L"CompositeLayers"); using Rml::BlendMode; // Blit source layer to postprocessing buffer. Do this regardless of whether we actually have any filters to be @@ -1978,13 +1968,11 @@ void RenderInterface_DX11::CompositeLayers(Rml::LayerHandle source_handle, Rml:: void RenderInterface_DX11::PopLayer() { m_render_layers.PopLayer(); - m_debug->EndEvent(); Gfx::BindRenderTarget(m_d3d_context, m_render_layers.GetTopLayer()); } Rml::TextureHandle RenderInterface_DX11::SaveLayerAsTexture() { - DebugScope scope(L"SaveLayerAsTexture"); RMLUI_ASSERT(m_scissor_state.Valid()); const Rml::Rectanglei bounds = m_scissor_state; @@ -2056,7 +2044,6 @@ struct CompiledFilter { Rml::CompiledFilterHandle RenderInterface_DX11::SaveLayerAsMaskImage() { - DebugScope scope(L"SaveLayerAsMaskImage"); BlitLayerToPostprocessPrimary(m_render_layers.GetTopLayerHandle()); const Gfx::RenderTargetData& source = m_render_layers.GetPostprocessPrimary(); @@ -2286,7 +2273,6 @@ Rml::CompiledShaderHandle RenderInterface_DX11::CompileShader(const Rml::String& void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, Rml::CompiledGeometryHandle geometry_handle, Rml::Vector2f translation, Rml::TextureHandle /*texture*/) { - DebugScope scope(L"RenderShader"); RMLUI_ASSERT(shader_handle && geometry_handle); const CompiledShader& shader = *reinterpret_cast(shader_handle); const CompiledShaderType type = shader.type; @@ -2301,7 +2287,6 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, { case CompiledShaderType::Gradient: { - DebugScope scope_gradient(L"Gradient"); RMLUI_ASSERT(shader.stop_positions.size() == shader.stop_colors.size()); const int num_stops = (int)shader.stop_positions.size(); @@ -2349,7 +2334,6 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, break; case CompiledShaderType::Creation: { - DebugScope scope_creation(L"Creation"); const double time = Rml::GetSystemInterface()->GetElapsedTime(); UseProgram(ProgramId::Creation); @@ -2448,7 +2432,6 @@ static void SetBlurWeights(Rml::Vector4f& p_weights, float sigma) void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& source_destination, const Gfx::RenderTargetData& temp, const Rml::Rectanglei window_flipped) { - DebugScope scope(L"RenderBlur"); RMLUI_ASSERT(&source_destination != &temp && source_destination.width == temp.width && source_destination.height == temp.height); RMLUI_ASSERT(window_flipped.Valid()); @@ -2635,7 +2618,6 @@ void RenderInterface_DX11::RenderBlur(float sigma, const Gfx::RenderTargetData& void RenderInterface_DX11::RenderFilters(Rml::Span filter_handles) { - DebugScope scope(L"RenderFilters"); m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); // To unbind slots when necessary @@ -2650,7 +2632,6 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetBlendState(m_blend_state_color_filter, blend_factor, 0xFFFFFFFF); @@ -2685,8 +2666,6 @@ void RenderInterface_DX11::RenderFilters(Rml::SpanOMSetDepthStencilState(m_depth_stencil_state_stencil_test, m_stencil_ref_value); } else { - DebugScope scope(L"DisableClipMask"); m_d3d_context->OMSetDepthStencilState(m_depth_stencil_state_disable, 0); } } void RenderInterface_DX11::RenderToClipMask(Rml::ClipMaskOperation operation, Rml::CompiledGeometryHandle geometry, Rml::Vector2f translation) { - DebugScope scope(L"RenderToClipMask"); RMLUI_ASSERT(m_is_stencil_enabled); using Rml::ClipMaskOperation; @@ -3053,24 +3026,4 @@ const Gfx::RenderTargetData& RenderInterface_DX11::RenderLayerStack::EnsureRende Gfx::CreateRenderTarget(m_d3d_device, rt, m_width, m_height, 0, Gfx::RenderTargetAttachment::None, 0); return rt; } - -static ID3DUserDefinedAnnotation* m_debug_scope_debug_device = nullptr; - -RenderInterface_DX11::DebugScope::~DebugScope() -{ - m_debug_scope_debug_device->EndEvent(); -} -RenderInterface_DX11::DebugScope::DebugScope(LPCWSTR name) -{ - m_debug_scope_debug_device->BeginEvent(name); -} -void RenderInterface_DX11::DebugScope::operator()(LPCWSTR name) const -{ - m_debug_scope_debug_device->BeginEvent(name); -} -void RenderInterface_DX11::DebugScope::AttachDebugLayer(ID3DUserDefinedAnnotation* debug) -{ - m_debug_scope_debug_device = debug; -} -#endif // RMLUI_PLATFORM_WIN32 - +#endif // RMLUI_PLATFORM_WIN32 \ No newline at end of file diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index efcd1cfab..f0fcdc8b2 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -342,17 +342,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { }; RenderLayerStack m_render_layers; - - ID3DUserDefinedAnnotation* m_debug = nullptr; - - class DebugScope { - private: - public: - DebugScope(LPCWSTR scopeName); - void operator()(LPCWSTR scopeName) const; - ~DebugScope(); - static void AttachDebugLayer(ID3DUserDefinedAnnotation* debug); - }; }; #endif // RMLUI_PLATFORM_WIN32 From d9a7ca3e22d8bddd8c9ff18307b2195cd685318c Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 6 Oct 2024 21:46:45 +0200 Subject: [PATCH 120/125] (chore): Remove redundant parameter for BeginFrame --- Backends/RmlUi_Backend_Win32_DX11.cpp | 2 +- Backends/RmlUi_Renderer_DX11.cpp | 5 +---- Backends/RmlUi_Renderer_DX11.h | 3 +-- 3 files changed, 3 insertions(+), 7 deletions(-) diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index f1f6a1703..a43f56c7e 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -293,7 +293,7 @@ void Backend::BeginFrame() CreateRenderTarget(); } - data->render_interface.BeginFrame(data->d3d_resources.pSwapChain, data->d3d_resources.pMainRenderTargetView); + data->render_interface.BeginFrame(data->d3d_resources.pMainRenderTargetView); } void Backend::PresentFrame() diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index ac1e26e3e..4d1c47a0c 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1150,11 +1150,10 @@ void RenderInterface_DX11::Cleanup() DX_CLEANUP_RESOURCE_IF_CREATED(m_d3d_device); } -void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view) +void RenderInterface_DX11::BeginFrame(ID3D11RenderTargetView* p_render_target_view) { RMLUI_ASSERT(m_viewport_width >= 1 && m_viewport_height >= 1); - RMLUI_ASSERTMSG(p_swapchain, "p_swapchain cannot be nullptr!"); RMLUI_ASSERTMSG(p_render_target_view, "p_render_target_view cannot be nullptr!"); RMLUI_ASSERTMSG(m_d3d_context, "d3d_context cannot be nullptr!"); RMLUI_ASSERTMSG(m_d3d_device, "d3d_device cannot be nullptr!"); @@ -1188,7 +1187,6 @@ void RenderInterface_DX11::BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderT } m_bound_render_target = p_render_target_view; - m_bound_swapchain = p_swapchain; // Initialise DX11 state for RmlUi D3D11_VIEWPORT d3dviewport; @@ -1243,7 +1241,6 @@ void RenderInterface_DX11::EndFrame() m_render_layers.EndFrame(); // Reset internal state - m_bound_swapchain = nullptr; m_bound_render_target = nullptr; m_current_blend_state = nullptr; diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index f0fcdc8b2..fbce19b38 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -82,7 +82,7 @@ class RenderInterface_DX11 : public Rml::RenderInterface { void Cleanup (); // Sets up DirectX11 states for taking rendering commands from RmlUi. - void BeginFrame(IDXGISwapChain* p_swapchain, ID3D11RenderTargetView* p_render_target_view); + void BeginFrame(ID3D11RenderTargetView* p_render_target_view); void EndFrame(); void SetViewport(const int width, const int height); @@ -168,7 +168,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { ID3D11Device* m_d3d_device = nullptr; ID3D11Device1* m_d3d_device_1 = nullptr; ID3D11DeviceContext1* m_d3d_context = nullptr; - IDXGISwapChain* m_bound_swapchain = nullptr; ID3D11RenderTargetView* m_bound_render_target = nullptr; bool m_scissor_enabled = false; ID3D11RasterizerState* m_rasterizer_state_scissor_enabled = nullptr; From 8a5e50fbb1a2671955ada836e76663daef05b6f3 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 29 Oct 2024 18:49:17 +0100 Subject: [PATCH 121/125] (feat): Add GLFW backend --- Backends/CMakeLists.txt | 11 + Backends/RmlUi_Backend_GLFW_DX11.cpp | 429 +++++++++++++++++++++++++++ CMake/OptionsLists.cmake | 1 + 3 files changed, 441 insertions(+) create mode 100644 Backends/RmlUi_Backend_GLFW_DX11.cpp diff --git a/Backends/CMakeLists.txt b/Backends/CMakeLists.txt index 1d05101d4..b05e2df21 100644 --- a/Backends/CMakeLists.txt +++ b/Backends/CMakeLists.txt @@ -153,6 +153,17 @@ if(UNIX) target_link_libraries(rmlui_backend_GLFW_GL3 INTERFACE ${CMAKE_DL_LIBS}) endif() +add_library(rmlui_backend_GLFW_DX11 INTERFACE) +target_sources(rmlui_backend_GLFW_DX11 INTERFACE + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Platform_GLFW.cpp" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Renderer_DX11.cpp" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Backend_GLFW_DX11.cpp" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Platform_GLFW.h" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Renderer_DX11.h" + "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Include_Windows.h" +) +target_link_libraries(rmlui_backend_GLFW_DX11 INTERFACE rmlui_backend_common_headers d3d11.lib dxgi.lib d3dcompiler.lib glfw) + add_library(rmlui_backend_GLFW_VK INTERFACE) target_sources(rmlui_backend_GLFW_VK INTERFACE "${CMAKE_CURRENT_LIST_DIR}/RmlUi_Platform_GLFW.cpp" diff --git a/Backends/RmlUi_Backend_GLFW_DX11.cpp b/Backends/RmlUi_Backend_GLFW_DX11.cpp new file mode 100644 index 000000000..aaeced2a4 --- /dev/null +++ b/Backends/RmlUi_Backend_GLFW_DX11.cpp @@ -0,0 +1,429 @@ +/* + * This source file is part of RmlUi, the HTML/CSS Interface Middleware + * + * For the latest information, see http://github.com/mikke89/RmlUi + * + * Copyright (c) 2008-2010 CodePoint Ltd, Shift Technology Ltd + * Copyright (c) 2019-2023 The RmlUi Team, and contributors + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + * + */ + +#include "RmlUi_Backend.h" +#include "RmlUi_Platform_GLFW.h" +#include "RmlUi_Renderer_DX11.h" +#include +#include +#include +#define GLFW_EXPOSE_NATIVE_WIN32 +#include +#include + +#ifdef RMLUI_USE_STB_IMAGE_LOADER +#define STB_IMAGE_IMPLEMENTATION +#include "stb_image.h" +#include + +// stb_image based loaders +void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize); +void FreeTexture(uint8_t* pData); +#endif + +static void SetupCallbacks(GLFWwindow* window); + +// D3D Creation / Cleanup functions +static bool CreateDeviceD3D(HWND hwnd); +static void CleanupDeviceD3D(); +static void CreateRenderTarget(); +static void CleanupRenderTarget(); + +static void LogErrorFromGLFW(int error, const char* description) +{ + Rml::Log::Message(Rml::Log::LT_ERROR, "GLFW error (0x%x): %s", error, description); +} + +/** + Global data used by this backend. + + Lifetime governed by the calls to Backend::Initialize() and Backend::Shutdown(). + */ +struct BackendData { + SystemInterface_GLFW system_interface; + RenderInterface_DX11 render_interface; + GLFWwindow* window = nullptr; + int glfw_active_modifiers = 0; + bool context_dimensions_dirty = true; + + // D3D11 resources + struct { + ID3D11Device* pd3dDevice = nullptr; + ID3D11DeviceContext* pd3dDeviceContext = nullptr; + IDXGISwapChain* pSwapChain = nullptr; + bool swapchain_occluded = false; + ID3D11RenderTargetView* pMainRenderTargetView = nullptr; + } d3d_resources; + + // Arguments set during event processing and nulled otherwise. + Rml::Context* context = nullptr; + KeyDownCallback key_down_callback = nullptr; +}; +static Rml::UniquePtr data; + +bool Backend::Initialize(const char* name, int width, int height, bool allow_resize) +{ + RMLUI_ASSERT(!data); + + glfwSetErrorCallback(LogErrorFromGLFW); + + if (!glfwInit()) + return false; + + glfwWindowHint(GLFW_CLIENT_API, GLFW_NO_API); + glfwWindowHint(GLFW_RESIZABLE, allow_resize ? GLFW_TRUE : GLFW_FALSE); + glfwWindowHint(GLFW_SCALE_TO_MONITOR, GLFW_TRUE); + + GLFWwindow* window = glfwCreateWindow(width, height, name, nullptr, nullptr); + if (!window) + return false; + + // Construct the system and render interface, this includes compiling all the shaders. If this fails, it is likely an error in the shader code. + data = Rml::MakeUnique(); + if (!data) + return false; + + data->window = window; + data->system_interface.SetWindow(window); + + HWND windowHwnd = glfwGetWin32Window(window); + if (!CreateDeviceD3D(windowHwnd)) + { + Shutdown(); + return false; + } + +#ifdef RMLUI_USE_STB_IMAGE_LOADER + // Assign stb_image loader to render interface + data->render_interface.LoadTextureFromFileRaw = &LoadTexture; + data->render_interface.FreeTextureFromFileRaw = &FreeTexture; + // Disable pre-multiplication because loading images through stb_image will inconsistently return pre-multiplied images (only iPhone PNGs are + // pre-multipled) We handle pre-multiplication manually on decode + stbi_set_unpremultiply_on_load(true); + // iPhone PNGs are encoded as BGRA, this tells stb_image to unpack as RGBA + stbi_convert_iphone_png_to_rgb(true); +#endif + + data->render_interface.Init(data->d3d_resources.pd3dDevice); + + // The window size may have been scaled by DPI settings, get the actual pixel size. + glfwGetFramebufferSize(window, &width, &height); + data->render_interface.SetViewport(width, height); + + // Receive num lock and caps lock modifiers for proper handling of numpad inputs in text fields. + glfwSetInputMode(window, GLFW_LOCK_KEY_MODS, GLFW_TRUE); + + // Setup the input and window event callback functions. + SetupCallbacks(window); + + return true; +} + +void Backend::Shutdown() +{ + RMLUI_ASSERT(data); + + // Cleanup renderer resources + data->render_interface.Cleanup(); + + // Shutdown DirectX11 + CleanupDeviceD3D(); + + glfwDestroyWindow(data->window); + data.reset(); + glfwTerminate(); +} + +Rml::SystemInterface* Backend::GetSystemInterface() +{ + RMLUI_ASSERT(data); + return &data->system_interface; +} + +Rml::RenderInterface* Backend::GetRenderInterface() +{ + RMLUI_ASSERT(data); + return &data->render_interface; +} + +bool Backend::ProcessEvents(Rml::Context* context, KeyDownCallback key_down_callback, bool power_save) +{ + RMLUI_ASSERT(data && context); + + // The initial window size may have been affected by system DPI settings, apply the actual pixel size and dp-ratio to the context. + if (data->context_dimensions_dirty) + { + data->context_dimensions_dirty = false; + + Rml::Vector2i window_size; + float dp_ratio = 1.f; + glfwGetFramebufferSize(data->window, &window_size.x, &window_size.y); + glfwGetWindowContentScale(data->window, &dp_ratio, nullptr); + + context->SetDimensions(window_size); + context->SetDensityIndependentPixelRatio(dp_ratio); + + CleanupRenderTarget(); + data->d3d_resources.pSwapChain->ResizeBuffers(0, window_size.x, window_size.y, DXGI_FORMAT_UNKNOWN, 0); + CreateRenderTarget(); + } + + data->context = context; + data->key_down_callback = key_down_callback; + + if (power_save) + glfwWaitEventsTimeout(Rml::Math::Min(context->GetNextUpdateDelay(), 10.0)); + else + glfwPollEvents(); + + data->context = nullptr; + data->key_down_callback = nullptr; + + const bool result = !glfwWindowShouldClose(data->window); + glfwSetWindowShouldClose(data->window, GLFW_FALSE); + return result; +} + +void Backend::RequestExit() +{ + RMLUI_ASSERT(data); + glfwSetWindowShouldClose(data->window, GLFW_TRUE); +} + +void Backend::BeginFrame() +{ + RMLUI_ASSERT(data); + data->render_interface.BeginFrame(data->d3d_resources.pMainRenderTargetView); +} + +void Backend::PresentFrame() +{ + RMLUI_ASSERT(data); + data->render_interface.EndFrame(); + + // Present + HRESULT hr = data->d3d_resources.pSwapChain->Present(1, 0); // Present with vsync + // HRESULT hr = g_pSwapChain->Present(0, 0); // Present without vsync + data->d3d_resources.swapchain_occluded = (hr == DXGI_STATUS_OCCLUDED); + + // Optional, used to mark frames during performance profiling. + RMLUI_FrameMark; +} + +static void SetupCallbacks(GLFWwindow* window) +{ + RMLUI_ASSERT(data); + + // Key input + glfwSetKeyCallback(window, [](GLFWwindow* /*window*/, int glfw_key, int /*scancode*/, int glfw_action, int glfw_mods) { + if (!data->context) + return; + + // Store the active modifiers for later because GLFW doesn't provide them in the callbacks to the mouse input events. + data->glfw_active_modifiers = glfw_mods; + + // Override the default key event callback to add global shortcuts for the samples. + Rml::Context* context = data->context; + KeyDownCallback key_down_callback = data->key_down_callback; + + switch (glfw_action) + { + case GLFW_PRESS: + case GLFW_REPEAT: + { + const Rml::Input::KeyIdentifier key = RmlGLFW::ConvertKey(glfw_key); + const int key_modifier = RmlGLFW::ConvertKeyModifiers(glfw_mods); + float dp_ratio = 1.f; + glfwGetWindowContentScale(data->window, &dp_ratio, nullptr); + + // See if we have any global shortcuts that take priority over the context. + if (key_down_callback && !key_down_callback(context, key, key_modifier, dp_ratio, true)) + break; + // Otherwise, hand the event over to the context by calling the input handler as normal. + if (!RmlGLFW::ProcessKeyCallback(context, glfw_key, glfw_action, glfw_mods)) + break; + // The key was not consumed by the context either, try keyboard shortcuts of lower priority. + if (key_down_callback && !key_down_callback(context, key, key_modifier, dp_ratio, false)) + break; + } + break; + case GLFW_RELEASE: RmlGLFW::ProcessKeyCallback(context, glfw_key, glfw_action, glfw_mods); break; + } + }); + + glfwSetCharCallback(window, [](GLFWwindow* /*window*/, unsigned int codepoint) { RmlGLFW::ProcessCharCallback(data->context, codepoint); }); + + glfwSetCursorEnterCallback(window, [](GLFWwindow* /*window*/, int entered) { RmlGLFW::ProcessCursorEnterCallback(data->context, entered); }); + + // Mouse input + glfwSetCursorPosCallback(window, [](GLFWwindow* window, double xpos, double ypos) { + RmlGLFW::ProcessCursorPosCallback(data->context, window, xpos, ypos, data->glfw_active_modifiers); + }); + + glfwSetMouseButtonCallback(window, [](GLFWwindow* /*window*/, int button, int action, int mods) { + data->glfw_active_modifiers = mods; + RmlGLFW::ProcessMouseButtonCallback(data->context, button, action, mods); + }); + + glfwSetScrollCallback(window, [](GLFWwindow* /*window*/, double /*xoffset*/, double yoffset) { + RmlGLFW::ProcessScrollCallback(data->context, yoffset, data->glfw_active_modifiers); + }); + + // Window events + glfwSetFramebufferSizeCallback(window, [](GLFWwindow* /*window*/, int width, int height) { + data->render_interface.SetViewport(width, height); + RmlGLFW::ProcessFramebufferSizeCallback(data->context, width, height); + }); + + glfwSetWindowContentScaleCallback(window, + [](GLFWwindow* /*window*/, float xscale, float /*yscale*/) { RmlGLFW::ProcessContentScaleCallback(data->context, xscale); }); +} + +static bool CreateDeviceD3D(HWND hwnd) +{ + // Setup swap chain + DXGI_SWAP_CHAIN_DESC sd; + ZeroMemory(&sd, sizeof(sd)); + sd.BufferCount = 2; + sd.BufferDesc.Width = 0; + sd.BufferDesc.Height = 0; + sd.BufferDesc.Format = DXGI_FORMAT_R8G8B8A8_UNORM; + sd.BufferDesc.RefreshRate.Numerator = 60; + sd.BufferDesc.RefreshRate.Denominator = 1; + sd.Flags = DXGI_SWAP_CHAIN_FLAG_ALLOW_MODE_SWITCH; + sd.BufferUsage = DXGI_USAGE_RENDER_TARGET_OUTPUT; + sd.OutputWindow = hwnd; + sd.SampleDesc.Count = 1; + sd.SampleDesc.Quality = 0; + sd.Windowed = TRUE; + sd.SwapEffect = DXGI_SWAP_EFFECT_FLIP_DISCARD; + + UINT createDeviceFlags = 0; +#ifdef RMLUI_DEBUG + // Enable debug layer + createDeviceFlags |= D3D11_CREATE_DEVICE_DEBUG; +#endif + D3D_FEATURE_LEVEL featureLevel; + const D3D_FEATURE_LEVEL featureLevelArray[2] = { + D3D_FEATURE_LEVEL_11_0, + D3D_FEATURE_LEVEL_10_0, + }; + HRESULT res = + D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_HARDWARE, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, + &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); + if (res == DXGI_ERROR_UNSUPPORTED) // Try high-performance WARP software driver if hardware is not available. + res = D3D11CreateDeviceAndSwapChain(nullptr, D3D_DRIVER_TYPE_WARP, nullptr, createDeviceFlags, featureLevelArray, 2, D3D11_SDK_VERSION, &sd, + &data->d3d_resources.pSwapChain, &data->d3d_resources.pd3dDevice, &featureLevel, &data->d3d_resources.pd3dDeviceContext); + if (res != S_OK) + return false; + + CreateRenderTarget(); + return true; +} + +static void CleanupDeviceD3D() +{ + CleanupRenderTarget(); + if (data->d3d_resources.pSwapChain) + { + data->d3d_resources.pSwapChain->Release(); + data->d3d_resources.pSwapChain = nullptr; + } + if (data->d3d_resources.pd3dDeviceContext) + { + data->d3d_resources.pd3dDeviceContext->Release(); + data->d3d_resources.pd3dDeviceContext = nullptr; + } + if (data->d3d_resources.pd3dDevice) + { + data->d3d_resources.pd3dDevice->Release(); + data->d3d_resources.pd3dDevice = nullptr; + } +} + +static void CreateRenderTarget() +{ + ID3D11Texture2D* pBackBuffer = nullptr; + data->d3d_resources.pSwapChain->GetBuffer(0, IID_PPV_ARGS(&pBackBuffer)); + data->d3d_resources.pd3dDevice->CreateRenderTargetView(pBackBuffer, nullptr, &data->d3d_resources.pMainRenderTargetView); + pBackBuffer->Release(); +} + +static void CleanupRenderTarget() +{ + if (data->d3d_resources.pMainRenderTargetView) + { + data->d3d_resources.pMainRenderTargetView->Release(); + data->d3d_resources.pMainRenderTargetView = nullptr; + } +} + +#ifdef RMLUI_USE_STB_IMAGE_LOADER +void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize) +{ + // Find where on disk the file is + Rml::FileInterface* file_interface = Rml::GetFileInterface(); + Rml::FileHandle file_handle = file_interface->Open(filename); + if (!file_handle) + { + // Tell the backend that the data is invalid (*pData = 0), and return + *pData = nullptr; + return; + } + + // Load the file through stb_image + int texture_width, texture_height, num_channels; + *pData = stbi_load_from_file((FILE*)file_handle, &texture_width, &texture_height, &num_channels, 0); + + // If the file data is correct + if (*pData != nullptr) + { + // Assign image parameters + *pWidth = texture_width; + *pHeight = texture_height; + + // Compute number of elements in texture + *pDataSize = texture_width * texture_height * num_channels; + + // Pre-multiply the data + uint8_t* dataView = *pData; + for (int i = 0; i < *pDataSize; i += 4) + { + dataView[i + 0] = (uint8_t)((dataView[i + 0] * dataView[i + 3]) / 255); + dataView[i + 1] = (uint8_t)((dataView[i + 1] * dataView[i + 3]) / 255); + dataView[i + 2] = (uint8_t)((dataView[i + 2] * dataView[i + 3]) / 255); + } + } +} + +void FreeTexture(uint8_t* pData) +{ + // Free the allocated memory once the texture has been uploaded to the GPU + stbi_image_free(pData); +} +#endif \ No newline at end of file diff --git a/CMake/OptionsLists.cmake b/CMake/OptionsLists.cmake index 90d0d8a9e..417b23800 100644 --- a/CMake/OptionsLists.cmake +++ b/CMake/OptionsLists.cmake @@ -16,6 +16,7 @@ set(RMLUI_BACKEND_OPTIONS "SFML_GL2" "GLFW_GL2" "GLFW_GL3" + "GLFW_DX11" "GLFW_VK" "BackwardCompatible_GLFW_GL2" "BackwardCompatible_GLFW_GL3" From 8ed475311f2a6c5bf9b0541a17be4113d2b2c217 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Tue, 29 Oct 2024 19:36:04 +0100 Subject: [PATCH 122/125] (chore): Add DirectX11 to README --- readme.md | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/readme.md b/readme.md index 712fe8c15..e909be550 100644 --- a/readme.md +++ b/readme.md @@ -168,13 +168,13 @@ The provided backends on the other hand are not intended to be used directly by ### Backends -| Platform \ Renderer | OpengGL 2 | OpengGL 3 | Vulkan | SDLrenderer | -|---------------------|:---------:|:---------:|:---------:|:-----------:| -| Win32 | ✔️ | | ✔️ | | -| X11 | ✔️ | | | | -| SFML | ✔️ | | | | -| GLFW | ✔️ | ✔️ | ✔️ | | -| SDL¹ | ✔️ | ✔️² | ✔️ | ✔️ | +| Platform \ Renderer | OpengGL 2 | OpengGL 3 | Vulkan | SDLrenderer | DirectX11 | +|---------------------|:---------:|:---------:|:---------:|:-----------:|:-----------:| +| Win32 | ✔️ | | ✔️ | | ✔️ | +| X11 | ✔️ | | | | | +| SFML | ✔️ | | | | | +| GLFW | ✔️ | ✔️ | ✔️ | | ✔️ | +| SDL¹ | ✔️ | ✔️² | ✔️ | ✔️ | | ¹ SDL backends extend their respective renderers to provide image support based on SDL_image.\ ² Supports Emscripten compilation target. From 14534e23d91d9f5b0a1e77e27abf0443d0b34bc9 Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 2 Nov 2024 16:41:57 +0100 Subject: [PATCH 123/125] (fix): Improve MSAA sampling pattern --- Backends/RmlUi_Renderer_DX11.cpp | 31 +++++++++++++++---------------- 1 file changed, 15 insertions(+), 16 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 4d1c47a0c..35f6fa9cc 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -649,8 +649,8 @@ static bool CreateRenderTarget(ID3D11Device* p_device, RenderTargetData& out_rt, texture_desc.MipLevels = 1; texture_desc.ArraySize = 1; texture_desc.Format = DXGI_FORMAT_R8G8B8A8_UNORM; - texture_desc.SampleDesc.Count = (samples > 0) ? samples : 1; // MSAA - texture_desc.SampleDesc.Quality = 0; + texture_desc.SampleDesc.Count = (samples > 1) ? samples : 1; // MSAA + texture_desc.SampleDesc.Quality = (samples > 1) ? D3D11_STANDARD_MULTISAMPLE_PATTERN : 0; texture_desc.Usage = D3D11_USAGE_DEFAULT; texture_desc.BindFlags = D3D11_BIND_RENDER_TARGET | D3D11_BIND_SHADER_RESOURCE; texture_desc.CPUAccessFlags = 0; @@ -708,22 +708,21 @@ static bool CreateRenderTarget(ID3D11Device* p_device, RenderTargetData& out_rt, else { // Create a new depth/stencil buffer - D3D11_TEXTURE2D_DESC depthDesc = {}; - depthDesc.Width = width; - depthDesc.Height = height; - depthDesc.MipLevels = 1; - depthDesc.ArraySize = 1; - depthDesc.Format = + D3D11_TEXTURE2D_DESC depth_desc = {}; + depth_desc.Width = width; + depth_desc.Height = height; + depth_desc.MipLevels = 1; + depth_desc.ArraySize = 1; + depth_desc.Format = (attachment == RenderTargetAttachment::DepthStencil) ? DXGI_FORMAT_D32_FLOAT_S8X24_UINT : DXGI_FORMAT_D24_UNORM_S8_UINT; - depthDesc.SampleDesc.Count = (samples > 0) ? samples : 1; - depthDesc.SampleDesc.Quality = 0; - depthDesc.Usage = D3D11_USAGE_DEFAULT; - depthDesc.BindFlags = D3D11_BIND_DEPTH_STENCIL; - depthDesc.CPUAccessFlags = 0; - depthDesc.MiscFlags = 0; + depth_desc.SampleDesc = texture_desc.SampleDesc; + depth_desc.Usage = D3D11_USAGE_DEFAULT; + depth_desc.BindFlags = D3D11_BIND_DEPTH_STENCIL; + depth_desc.CPUAccessFlags = 0; + depth_desc.MiscFlags = 0; ID3D11Texture2D* depth_stencil_texture = nullptr; - result = p_device->CreateTexture2D(&depthDesc, nullptr, &depth_stencil_texture); + result = p_device->CreateTexture2D(&depth_desc, nullptr, &depth_stencil_texture); RMLUI_DX_ASSERTMSG(result, "failed to CreateTexture2D"); if (FAILED(result)) { @@ -732,7 +731,7 @@ static bool CreateRenderTarget(ID3D11Device* p_device, RenderTargetData& out_rt, } D3D11_DEPTH_STENCIL_VIEW_DESC dsvDesc = {}; - dsvDesc.Format = depthDesc.Format; + dsvDesc.Format = depth_desc.Format; dsvDesc.ViewDimension = (samples > 0) ? D3D11_DSV_DIMENSION_TEXTURE2DMS : D3D11_DSV_DIMENSION_TEXTURE2D; dsvDesc.Texture2D.MipSlice = 0; From 6c5784934d161f21f9820cd2320e085d23df2a9b Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sat, 2 Nov 2024 16:45:02 +0100 Subject: [PATCH 124/125] (fix): Remove m_geometry_cache As per suggestion 1 from wh1t3lord --- Backends/RmlUi_Renderer_DX11.cpp | 161 ++++++++++++------------------- Backends/RmlUi_Renderer_DX11.h | 2 - 2 files changed, 63 insertions(+), 100 deletions(-) diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index 35f6fa9cc..e13fcb399 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -831,18 +831,6 @@ static void DestroyShaders(const ProgramData& data) } // namespace Gfx -static uintptr_t HashPointer(uintptr_t inPtr) -{ - uintptr_t Value = (uintptr_t)inPtr; - Value = ~Value + (Value << 15); - Value = Value ^ (Value >> 12); - Value = Value + (Value << 2); - Value = Value ^ (Value >> 4); - Value = Value * 2057; - Value = Value ^ (Value >> 16); - return Value; -} - RenderInterface_DX11::RenderInterface_DX11() {} RenderInterface_DX11::~RenderInterface_DX11() {} @@ -1115,17 +1103,16 @@ void RenderInterface_DX11::Init(ID3D11Device* p_d3d_device) void RenderInterface_DX11::Cleanup() { - if (program_data) + if (m_fullscreen_quad_geometry) { - Gfx::DestroyShaders(*program_data); - program_data.reset(); + RenderInterface_DX11::ReleaseGeometry(m_fullscreen_quad_geometry); + m_fullscreen_quad_geometry = {}; } - // Loop through geometry cache and free all resources - for (auto& it : m_geometry_cache) + if (program_data) { - DX_CLEANUP_RESOURCE_IF_CREATED(it.second.vertex_buffer); - DX_CLEANUP_RESOURCE_IF_CREATED(it.second.index_buffer); + Gfx::DestroyShaders(*program_data); + program_data.reset(); } // Cleans up all general resources @@ -1211,7 +1198,6 @@ void RenderInterface_DX11::BeginFrame(ID3D11RenderTargetView* p_render_target_vi m_d3d_context->ClearRenderTargetView(m_render_layers.GetTopLayer().render_target_view, clearColor); UseProgram(ProgramId::None); - // program_transform_dirty.set(); m_scissor_state = Rml::Rectanglei::MakeInvalid(); } @@ -1232,7 +1218,6 @@ void RenderInterface_DX11::EndFrame() // Instead, if we had a transparent destination that didn't use premultiplied alpha, we would need to perform a manual un-premultiplication step. Gfx::BindTexture(m_d3d_context, rt_postprocess); - // @TODO: Find a pattern for flipped textures UseProgram(ProgramId::Passthrough); DrawFullscreenQuad(); @@ -1381,93 +1366,77 @@ Rml::CompiledGeometryHandle RenderInterface_DX11::CompileGeometry(Rml::Spanvertex_buffer = vertex_buffer; + geometryData->index_buffer = index_buffer; + geometryData->index_count = indices.size(); - m_geometry_cache.emplace(handleId, geometryData); - return Rml::CompiledGeometryHandle(handleId); + return (Rml::CompiledGeometryHandle)geometryData; } void RenderInterface_DX11::ReleaseGeometry(Rml::CompiledGeometryHandle handle) { - if (m_geometry_cache.find(handle) != m_geometry_cache.end()) - { - DX11_GeometryData geometryData = m_geometry_cache[handle]; + DX11_GeometryData* geometry = (DX11_GeometryData*)handle; - DX_CLEANUP_RESOURCE_IF_CREATED(geometryData.vertex_buffer); - DX_CLEANUP_RESOURCE_IF_CREATED(geometryData.index_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(geometry->vertex_buffer); + DX_CLEANUP_RESOURCE_IF_CREATED(geometry->index_buffer); - m_geometry_cache.erase(handle); - } - else - { - Rml::Log::Message(Rml::Log::LT_WARNING, "Geometry Handle %d does not exist!", handle); - } + delete geometry; } void RenderInterface_DX11::RenderGeometry(Rml::CompiledGeometryHandle handle, Rml::Vector2f translation, Rml::TextureHandle texture) { - if (m_geometry_cache.find(handle) != m_geometry_cache.end()) - { - DX11_GeometryData geometryData = m_geometry_cache[handle]; + DX11_GeometryData* geometry = (DX11_GeometryData*)handle; - if (m_translation != translation) - { - m_translation = translation; - m_cbuffer_dirty = true; - } - - if (texture == TexturePostprocess || texture == TexturePostprocessNoBinding) - { - // Do nothing. - } - else if (texture) - { - // Texture available - UseProgram(ProgramId::Texture); - if (texture != TextureEnableWithoutBinding) - { - ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture; - m_d3d_context->PSSetShaderResources(0, 1, &texture_view); - } - m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); - } - else - { - // No texture, use color - UseProgram(ProgramId::Color); - } + if (m_translation != translation) + { + m_translation = translation; + m_cbuffer_dirty = true; + } - if (texture != TexturePostprocessNoBinding) + if (texture == TexturePostprocess || texture == TexturePostprocessNoBinding) + { + // Do nothing. + } + else if (texture) + { + // Texture available + UseProgram(ProgramId::Texture); + if (texture != TextureEnableWithoutBinding) { - UpdateConstantBuffer(); - - m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); - m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); + ID3D11ShaderResourceView* texture_view = (ID3D11ShaderResourceView*)texture; + m_d3d_context->PSSetShaderResources(0, 1, &texture_view); } - - // Bind vertex and index buffers, issue draw call - uint32_t stride = sizeof(Rml::Vertex); - uint32_t offset = 0; - m_d3d_context->IASetInputLayout(m_vertex_layout); - m_d3d_context->IASetVertexBuffers(0, 1, &geometryData.vertex_buffer, &stride, &offset); - m_d3d_context->IASetIndexBuffer(geometryData.index_buffer, DXGI_FORMAT_R32_UINT, 0); - m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); - m_d3d_context->DrawIndexed(static_cast(geometryData.index_count), 0, 0); - - // Unbind resources - ID3D11ShaderResourceView* const nullSRV = nullptr; - m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); + m_d3d_context->PSSetSamplers(0, 1, &m_sampler_state); } else { - Rml::Log::Message(Rml::Log::LT_WARNING, "Geometry Handle %d does not exist!", handle); + // No texture, use color + UseProgram(ProgramId::Color); + } + + if (texture != TexturePostprocessNoBinding) + { + UpdateConstantBuffer(); + + m_d3d_context->VSSetConstantBuffers(0, 1, &m_shader_buffer); + m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); } + + // Bind vertex and index buffers, issue draw call + uint32_t stride = sizeof(Rml::Vertex); + uint32_t offset = 0; + m_d3d_context->IASetInputLayout(m_vertex_layout); + m_d3d_context->IASetVertexBuffers(0, 1, &geometry->vertex_buffer, &stride, &offset); + m_d3d_context->IASetIndexBuffer(geometry->index_buffer, DXGI_FORMAT_R32_UINT, 0); + m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); + m_d3d_context->DrawIndexed(static_cast(geometry->index_count), 0, 0); + + // Unbind resources + ID3D11ShaderResourceView* const nullSRV = nullptr; + m_d3d_context->PSSetShaderResources(0, 1, &nullSRV); } // Set to byte packing, or the compiler will expand our struct, which means it won't read correctly from file @@ -2273,11 +2242,7 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, const CompiledShader& shader = *reinterpret_cast(shader_handle); const CompiledShaderType type = shader.type; - DX11_GeometryData geometry{}; - if (m_geometry_cache.find(geometry_handle) != m_geometry_cache.end()) - { - geometry = m_geometry_cache[geometry_handle]; - } + DX11_GeometryData* geometry = (DX11_GeometryData*)geometry_handle; switch (type) { @@ -2322,10 +2287,10 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); uint32_t stride = sizeof(Rml::Vertex); uint32_t offset = 0; - m_d3d_context->IASetVertexBuffers(0, 1, &geometry.vertex_buffer, &stride, &offset); - m_d3d_context->IASetIndexBuffer(geometry.index_buffer, DXGI_FORMAT_R32_UINT, 0); + m_d3d_context->IASetVertexBuffers(0, 1, &geometry->vertex_buffer, &stride, &offset); + m_d3d_context->IASetIndexBuffer(geometry->index_buffer, DXGI_FORMAT_R32_UINT, 0); m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); - m_d3d_context->DrawIndexed(static_cast(geometry.index_count), 0, 0); + m_d3d_context->DrawIndexed(static_cast(geometry->index_count), 0, 0); } break; case CompiledShaderType::Creation: @@ -2360,10 +2325,10 @@ void RenderInterface_DX11::RenderShader(Rml::CompiledShaderHandle shader_handle, m_d3d_context->PSSetConstantBuffers(0, 1, &m_shader_buffer); uint32_t stride = sizeof(Rml::Vertex); uint32_t offset = 0; - m_d3d_context->IASetVertexBuffers(0, 1, &geometry.vertex_buffer, &stride, &offset); - m_d3d_context->IASetIndexBuffer(geometry.index_buffer, DXGI_FORMAT_R32_UINT, 0); + m_d3d_context->IASetVertexBuffers(0, 1, &geometry->vertex_buffer, &stride, &offset); + m_d3d_context->IASetIndexBuffer(geometry->index_buffer, DXGI_FORMAT_R32_UINT, 0); m_d3d_context->IASetPrimitiveTopology(D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST); - m_d3d_context->DrawIndexed(static_cast(geometry.index_count), 0, 0); + m_d3d_context->DrawIndexed(static_cast(geometry->index_count), 0, 0); } break; case CompiledShaderType::Invalid: diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index fbce19b38..e70471435 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -214,8 +214,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { size_t index_count = 0; }; - std::unordered_map m_geometry_cache; - // D3D11 state for RmlUi rendering bool m_current_blend_state_enabled = false; ID3D11BlendState* m_current_blend_state = nullptr; From 8c1d982b59f5505ee271436fda9c33165490bf9e Mon Sep 17 00:00:00 2001 From: Hyblocker Date: Sun, 3 Nov 2024 18:14:20 +0100 Subject: [PATCH 125/125] (fix): Move stb_image sample to backend. --- Backends/RmlUi_Backend_GLFW_DX11.cpp | 117 ++++++++++++++------------ Backends/RmlUi_Backend_Win32_DX11.cpp | 116 +++++++++++++------------ Backends/RmlUi_Renderer_DX11.cpp | 22 ----- Backends/RmlUi_Renderer_DX11.h | 12 --- 4 files changed, 121 insertions(+), 146 deletions(-) diff --git a/Backends/RmlUi_Backend_GLFW_DX11.cpp b/Backends/RmlUi_Backend_GLFW_DX11.cpp index aaeced2a4..2557bfac7 100644 --- a/Backends/RmlUi_Backend_GLFW_DX11.cpp +++ b/Backends/RmlUi_Backend_GLFW_DX11.cpp @@ -37,14 +37,67 @@ #include #ifdef RMLUI_USE_STB_IMAGE_LOADER -#define STB_IMAGE_IMPLEMENTATION -#include "stb_image.h" -#include + #define STB_IMAGE_IMPLEMENTATION + #include "stb_image.h" + #include + #include +#endif + +/** + Custom render interface example for the DX11/GLFW backend. + + Overloads the DX11 render interface to load textures through stb_image. + */ +class RenderInterface_DX11_Win32 : public RenderInterface_DX11 { +public: + RenderInterface_DX11_Win32() {} + +#ifdef RMLUI_USE_STB_IMAGE_LOADER + Rml::TextureHandle LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) override + { + int texture_width = 0, texture_height = 0, num_channels = 0; + size_t image_size_bytes = 0; + uint8_t* texture_data = nullptr; + + // Find where on disk the file is + Rml::FileInterface* file_interface = Rml::GetFileInterface(); + Rml::FileHandle file_handle = file_interface->Open(source); + if (!file_handle) + { + // Tell RmlUI that the image is invalid + texture_data = nullptr; + return false; + } -// stb_image based loaders -void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize); -void FreeTexture(uint8_t* pData); + // Load the file through stb_image + texture_data = stbi_load_from_file((FILE*)file_handle, &texture_width, &texture_height, &num_channels, 0); + + // If the file data is correct + if (texture_data != nullptr) + { + // Compute number of elements in texture + image_size_bytes = texture_width * texture_height * num_channels; + + // Pre-multiply the data + for (int i = 0; i < image_size_bytes; i += 4) + { + texture_data[i + 0] = (uint8_t)((texture_data[i + 0] * texture_data[i + 3]) / 255); + texture_data[i + 1] = (uint8_t)((texture_data[i + 1] * texture_data[i + 3]) / 255); + texture_data[i + 2] = (uint8_t)((texture_data[i + 2] * texture_data[i + 3]) / 255); + } + + texture_dimensions.x = texture_width; + texture_dimensions.y = texture_height; + + Rml::TextureHandle handle = GenerateTexture({texture_data, image_size_bytes}, texture_dimensions); + + stbi_image_free(texture_data); + return handle; + } + return false; + } #endif +}; static void SetupCallbacks(GLFWwindow* window); @@ -66,7 +119,7 @@ static void LogErrorFromGLFW(int error, const char* description) */ struct BackendData { SystemInterface_GLFW system_interface; - RenderInterface_DX11 render_interface; + RenderInterface_DX11_Win32 render_interface; GLFWwindow* window = nullptr; int glfw_active_modifiers = 0; bool context_dimensions_dirty = true; @@ -119,9 +172,6 @@ bool Backend::Initialize(const char* name, int width, int height, bool allow_res } #ifdef RMLUI_USE_STB_IMAGE_LOADER - // Assign stb_image loader to render interface - data->render_interface.LoadTextureFromFileRaw = &LoadTexture; - data->render_interface.FreeTextureFromFileRaw = &FreeTexture; // Disable pre-multiplication because loading images through stb_image will inconsistently return pre-multiplied images (only iPhone PNGs are // pre-multipled) We handle pre-multiplication manually on decode stbi_set_unpremultiply_on_load(true); @@ -381,49 +431,4 @@ static void CleanupRenderTarget() data->d3d_resources.pMainRenderTargetView->Release(); data->d3d_resources.pMainRenderTargetView = nullptr; } -} - -#ifdef RMLUI_USE_STB_IMAGE_LOADER -void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize) -{ - // Find where on disk the file is - Rml::FileInterface* file_interface = Rml::GetFileInterface(); - Rml::FileHandle file_handle = file_interface->Open(filename); - if (!file_handle) - { - // Tell the backend that the data is invalid (*pData = 0), and return - *pData = nullptr; - return; - } - - // Load the file through stb_image - int texture_width, texture_height, num_channels; - *pData = stbi_load_from_file((FILE*)file_handle, &texture_width, &texture_height, &num_channels, 0); - - // If the file data is correct - if (*pData != nullptr) - { - // Assign image parameters - *pWidth = texture_width; - *pHeight = texture_height; - - // Compute number of elements in texture - *pDataSize = texture_width * texture_height * num_channels; - - // Pre-multiply the data - uint8_t* dataView = *pData; - for (int i = 0; i < *pDataSize; i += 4) - { - dataView[i + 0] = (uint8_t)((dataView[i + 0] * dataView[i + 3]) / 255); - dataView[i + 1] = (uint8_t)((dataView[i + 1] * dataView[i + 3]) / 255); - dataView[i + 2] = (uint8_t)((dataView[i + 2] * dataView[i + 3]) / 255); - } - } -} - -void FreeTexture(uint8_t* pData) -{ - // Free the allocated memory once the texture has been uploaded to the GPU - stbi_image_free(pData); -} -#endif \ No newline at end of file +} \ No newline at end of file diff --git a/Backends/RmlUi_Backend_Win32_DX11.cpp b/Backends/RmlUi_Backend_Win32_DX11.cpp index a43f56c7e..b46f68dea 100644 --- a/Backends/RmlUi_Backend_Win32_DX11.cpp +++ b/Backends/RmlUi_Backend_Win32_DX11.cpp @@ -37,14 +37,66 @@ #include #ifdef RMLUI_USE_STB_IMAGE_LOADER -#define STB_IMAGE_IMPLEMENTATION -#include "stb_image.h" -#include + #define STB_IMAGE_IMPLEMENTATION + #include "stb_image.h" + #include +#endif + +/** + Custom render interface example for the DX11/Win32 backend. + + Overloads the DX11 render interface to load textures through stb_image. + */ +class RenderInterface_DX11_Win32 : public RenderInterface_DX11 { +public: + RenderInterface_DX11_Win32() {} + +#ifdef RMLUI_USE_STB_IMAGE_LOADER + Rml::TextureHandle LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) override + { + int texture_width = 0, texture_height = 0, num_channels = 0; + size_t image_size_bytes = 0; + uint8_t* texture_data = nullptr; + + // Find where on disk the file is + Rml::FileInterface* file_interface = Rml::GetFileInterface(); + Rml::FileHandle file_handle = file_interface->Open(source); + if (!file_handle) + { + // Tell RmlUI that the image is invalid + texture_data = nullptr; + return false; + } + + // Load the file through stb_image + texture_data = stbi_load_from_file((FILE*)file_handle, &texture_width, &texture_height, &num_channels, 0); -// stb_image based loaders -void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize); -void FreeTexture(uint8_t* pData); + // If the file data is correct + if (texture_data != nullptr) + { + // Compute number of elements in texture + image_size_bytes = texture_width * texture_height * num_channels; + + // Pre-multiply the data + for (int i = 0; i < image_size_bytes; i += 4) + { + texture_data[i + 0] = (uint8_t)((texture_data[i + 0] * texture_data[i + 3]) / 255); + texture_data[i + 1] = (uint8_t)((texture_data[i + 1] * texture_data[i + 3]) / 255); + texture_data[i + 2] = (uint8_t)((texture_data[i + 2] * texture_data[i + 3]) / 255); + } + + texture_dimensions.x = texture_width; + texture_dimensions.y = texture_height; + + Rml::TextureHandle handle = GenerateTexture({texture_data, image_size_bytes}, texture_dimensions); + + stbi_image_free(texture_data); + return handle; + } + return false; + } #endif +}; /** High DPI support using Windows Per Monitor V2 DPI awareness. @@ -118,7 +170,7 @@ Lifetime governed by the calls to Backend::Initialize() and Backend::Shutdown(). */ struct BackendData { SystemInterface_Win32 system_interface; - RenderInterface_DX11 render_interface; + RenderInterface_DX11_Win32 render_interface; TextInputMethodEditor_Win32 text_input_method_editor; HINSTANCE instance_handle = nullptr; @@ -168,9 +220,6 @@ bool Backend::Initialize(const char* window_name, int width, int height, bool al } #ifdef RMLUI_USE_STB_IMAGE_LOADER - // Assign stb_image loader to render interface - data->render_interface.LoadTextureFromFileRaw = &LoadTexture; - data->render_interface.FreeTextureFromFileRaw = &FreeTexture; // Disable pre-multiplication because loading images through stb_image will inconsistently return pre-multiplied images (only iPhone PNGs are pre-multipled) // We handle pre-multiplication manually on decode stbi_set_unpremultiply_on_load(true); @@ -524,49 +573,4 @@ static void CleanupRenderTarget() data->d3d_resources.pMainRenderTargetView->Release(); data->d3d_resources.pMainRenderTargetView = nullptr; } -} - -#ifdef RMLUI_USE_STB_IMAGE_LOADER -void LoadTexture(const Rml::String& filename, int* pWidth, int* pHeight, uint8_t** pData, size_t* pDataSize) -{ - // Find where on disk the file is - Rml::FileInterface* file_interface = Rml::GetFileInterface(); - Rml::FileHandle file_handle = file_interface->Open(filename); - if (!file_handle) - { - // Tell the backend that the data is invalid (*pData = 0), and return - *pData = nullptr; - return; - } - - // Load the file through stb_image - int texture_width, texture_height, num_channels; - *pData = stbi_load_from_file((FILE*)file_handle, &texture_width, &texture_height, &num_channels, 0); - - // If the file data is correct - if (*pData != nullptr) - { - // Assign image parameters - *pWidth = texture_width; - *pHeight = texture_height; - - // Compute number of elements in texture - *pDataSize = texture_width * texture_height * num_channels; - - // Pre-multiply the data - uint8_t* dataView = *pData; - for (int i = 0; i < *pDataSize; i += 4) - { - dataView[i + 0] = (uint8_t)((dataView[i + 0] * dataView[i + 3]) / 255); - dataView[i + 1] = (uint8_t)((dataView[i + 1] * dataView[i + 3]) / 255); - dataView[i + 2] = (uint8_t)((dataView[i + 2] * dataView[i + 3]) / 255); - } - } -} - -void FreeTexture(uint8_t* pData) -{ - // Free the allocated memory once the texture has been uploaded to the GPU - stbi_image_free(pData); -} -#endif \ No newline at end of file +} \ No newline at end of file diff --git a/Backends/RmlUi_Renderer_DX11.cpp b/Backends/RmlUi_Renderer_DX11.cpp index e13fcb399..12b740f2e 100644 --- a/Backends/RmlUi_Renderer_DX11.cpp +++ b/Backends/RmlUi_Renderer_DX11.cpp @@ -1461,28 +1461,6 @@ struct TGAHeader { Rml::TextureHandle RenderInterface_DX11::LoadTexture(Rml::Vector2i& texture_dimensions, const Rml::String& source) { - // Use the user provided image loading function if it's provided, else fallback to the included TGA one - if (LoadTextureFromFileRaw != nullptr && FreeTextureFromFileRaw != nullptr) - { - int texture_width = 0, texture_height = 0; - size_t image_size_bytes = 0; - uint8_t* texture_data = nullptr; - LoadTextureFromFileRaw(source, &texture_width, &texture_height, &texture_data, &image_size_bytes); - - if (texture_data != nullptr) - { - texture_dimensions.x = texture_width; - texture_dimensions.y = texture_height; - - Rml::TextureHandle handle = GenerateTexture({texture_data, image_size_bytes}, texture_dimensions); - - FreeTextureFromFileRaw(texture_data); - return handle; - } - - // Image must be invalid if the file failed to load. Fallback to the default loader. - } - Rml::FileInterface* file_interface = Rml::GetFileInterface(); Rml::FileHandle file_handle = file_interface->Open(source); if (!file_handle) diff --git a/Backends/RmlUi_Renderer_DX11.h b/Backends/RmlUi_Renderer_DX11.h index e70471435..643f39f2e 100644 --- a/Backends/RmlUi_Renderer_DX11.h +++ b/Backends/RmlUi_Renderer_DX11.h @@ -130,18 +130,6 @@ class RenderInterface_DX11 : public Rml::RenderInterface { // Can be passed to RenderGeometry() to not touch the cbuffers static constexpr Rml::TextureHandle TexturePostprocessNoBinding = Rml::TextureHandle(-3); -public: - /// Called by the renderer when it wants to load a texture from disk. - /// @param[in] pFilename A Rml string containing the file name. - /// @param[out] pWidth The width of the texture read from disk. - /// @param[out] pHeight The height of the texture read from disk. - /// @param[out] pData A pointer to an RGBA byte array of texture data, which will then be used to generate a texture. Set to NULL to indicate that the file failed to load for whatever reason. - /// @param[out] pDataSize A pointer to a size_t storing how many bytes is in pData. - pfnLoadTextureRaw LoadTextureFromFileRaw = nullptr; - /// Called by the renderer when it wants to free a texture from disk. Always called after a successful load from LoadTextureFromFileRaw. - /// @param[in] pData A pointer to an RGBA byte array of texture data to free. - pfnFreeTextureRaw FreeTextureFromFileRaw = nullptr; - private: // Changes blend state if necessary void SetBlendState(ID3D11BlendState* blend_state);