Masked image reconstruction
The model reuses FG-CLIP, it takes a reference image, then reconstructs the masked image. The prediction output is a series of discrete numbers representing the masked tokens.
Datasets
- animelover/touhou-images
- Chars/pixiv_rank_daily_2018_2023
- Makki2104/difference_images_Cloth-Nude
- picollect/12TPICS
- recoilme/tst72
- sugarquark/kiradepth-v1.1-character-index
- sugarquark/nai-mixed-400
Disclaimer
The license requires a link to the Hugging Face profile.
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support