Repository navigation
Conversation
Cast images to uint8 before dwtDct encode so chroma matches what cv2.imread feeds the decoder; Fixes huggingface#15025. Signed-off-by: GokayAI <60583610+gokay-ai@users.noreply.github.com>
github-actions
Bot
added
tests
pipelines
fixes-issue
size/M
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes #15025.
StableDiffusionXLWatermarker.apply_watermarkwas passing float32 pixels in[0, 255]toimwatermarkdwtDctencode. Stability'sdetect.pyreads the saved PNG withcv2.imread(uint8), which applies a +128 chroma offset that the float encode path never sees, so about a third of watermarked images fail detection.This casts to
uint8(round + clip) before encode and restoresfloat32for the existing torch conversion, matching the uint8 encode path that recovers all 48 watermark bits on the issue repro image.Test plan
pytest tests/pipelines/stable_diffusion_xl/test_stable_diffusion_xl_watermark.py -v(needsinvisible-watermark+ OpenCV)0/48(float encode) to48/48(this change) after PNG round-trip decodeAI assistance was used for implementation and tests; I reviewed the diff and ran the tests above.