Dear Seed-X Team,
I would like to express my gratitude for making your remarkable work available as open source. I have been exploring the editing model and encountered some unexpected results:
While using the model with a specific sample, I noticed some artifacts during the inference process. For instance, when I input the image (f1.jpg) with the prompt 'Change its color to blue', the output (f2.jpg) did not reflect the requested color change and introduced several artifacts.
Could this be an expected behavior, or is it possible that there might be an issue with my implementation? Any guidance or suggestions you could provide would be greatly appreciated.
Also, I find in detokenizer (stage2) it looks missing multiplied by vae.config.scaling_factor here. Is it a feature or bug or I missed anything? I appreciate for more explanation.
Dear Seed-X Team, I would like to express my gratitude for making your remarkable work available as open source. I have been exploring the editing model and encountered some unexpected results: While using the model with a specific sample, I noticed some artifacts during the inference process. For instance, when I input the image (f1.jpg) with the prompt 'Change its color to blue', the output (f2.jpg) did not reflect the requested color change and introduced several artifacts. Could this be an expected behavior, or is it possible that there might be an issue with my implementation? Any guidance or suggestions you could provide would be greatly appreciated.