Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[New Feature] Control Mode #1011

Closed
lllyasviel opened this issue Apr 24, 2023 · 1 comment
Closed

[New Feature] Control Mode #1011

lllyasviel opened this issue Apr 24, 2023 · 1 comment

Comments

@lllyasviel
Copy link
Collaborator

The Guess Mode will be revised into Control Mode since 1.1.09.

Control Mode (previously called Guess Mode)

We have fixed many bugs in previous 1.0’s Guess Mode and now it is called Control Mode

image

Now you can control which aspect is more important (your prompt or your ControlNet)

Input (depth+canny+hed) Control Mode: "Balanced" Control Mode: "My prompt is more important" Control Mode: "ControlNet is more important"
image image image image

Now you can use the guess mode in a better and easier way (easier to understand what happened).

If you want to reproduce your previous results in "Guess Mode", you can always try "ControlNet is more important" and you should be able to get same results. If not, then try "My prompt is more important".

Note that you can 100% reproduce all previous results and nothing is removed.

@lllyasviel
Copy link
Collaborator Author

Some doc from readme:

"Balanced" = put ControlNet on both sides of cfg scale, same as turn off "Guess Mode" in ControlNet 1.0

"My prompt is more important" = put ControlNet on both sides of cfg scale and use progressively reduced SD U-Net injections (layer_weight*=0.825**I, where 0<=I <13, and the 13 means ControlNet injected SD 13 times). In this way, you can make sure that your prompts are perfectly displayed in your generated images.

"ControlNet is more important" = put ControlNet only on the Conditional Side (the cond in A1111's batch-cond-uncond). This means the ControlNet will be X times stronger if your cfg-scale is X. For example, if your cfg-scale is 7, then ControlNet is 7 times stronger. Note that here the X times stronger is different from "Control Weights" since your weights are not modified. This "stronger" effect usually has less artifacts and give ControlNet more space to guess what is missing from your prompts (and in 1.0, it is called "Guess Mode").

Repository owner deleted a comment from aartikov May 5, 2023
Repository owner deleted a comment from Mikubill May 5, 2023
Repository owner deleted a comment from topdeckg May 5, 2023
Repository owner deleted a comment from guzuligo May 5, 2023
Repository owner deleted a comment from Daxiongmao87 May 5, 2023
Repository owner deleted a comment from ljleb May 5, 2023
Repository owner locked and limited conversation to collaborators May 5, 2023
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants