Add RealPLKSR as an upscale model#32
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
After more testing of BSRGAN on real photos, its quality wasn't great as a one-size-fits-all upscaler. The root cause is architectural: BSRGAN was trained with the RealESRGAN-style degradation pipeline, so it implicitly denoises and deblurs as part of upscaling. On a clean source – an edited raw – that denoise side has nothing to do and tends to soften texture the user wanted to keep.
This PR adds RealPLKSR (Real-world PLKSR, Dongheon Lee et al., University of Seoul) as an alternative. MSSIM-pretrain stage only, no GAN finetune – stays conservative and faithful, no implicit denoise. Loaded via Spandrel, exported as static ONNX at 512×512 (x2) and 256×256 (x4).