2xLexicaRRDBNet

Name: 2xLexicaRRDBNet
Author: Philip Hofmann
Release Date: 01.06.2023
License: CC BY 4.0
Network: RRDBNet
Scale: 2
Purpose: Upscaling AI generated images
Iterations: 185'000
batch_size: 4
HR_size: 128
Epoch: 17 (require iter number per epoch: 10964)
Dataset: lexica-aperture-v3-small
Number of train images: 43856
OTF Training: No
Pretrained_Model_G: None

Description: 2x upscaler for the AI generated image output. Trained on 43856 images from lexica.art, so its trained specifically on that model but should work in general on ai generated images.

16 Examples with Input, Upscaled (Normal and Sharp) and GT Files, plus example data: https://drive.google.com/drive/folders/1LT20d5u1m8CryCrXOJ7pWJd0mlN7X5yA

Name: 2xLexicaRRDBNet_Sharp
Author: Philip Hofmann
Release Date: 01.06.2023
License: CC BY 4.0
Network: RRDBNet
Scale: 2
Purpose: Upscaling AI generated images - a bit sharper then above model
Iterations: 220'000
batch_size: 4
HR_size: 128
Epoch: 18 (require iter number per epoch: 10964)
Dataset: lexica-aperture-v3-small
Number of train images: 43856
OTF Training: No
Pretrained_Model_G: None

Description: Its like the above model, but trained for some more with l1_gt_usm and percep_gt_usm set to true, resulting in sharper outputs. I provide both so they can be chosen based on preferrence of the user.