Pre-trained LaDI-VTON Models and Dress Code in-shop masks
Pre-trained Models
For both Dress Code and Viton-HD datasets we release the following models:
- EMASC
- Inversion Adapter
- Denoising UNet
- Warping
Dress Code in-shop masks
To enhance the performance of our warping module, we have discovered that using in-shop images with a white background yields better results. To facilitate this process, we now offer pre-extracted masks that can be used to remove the background from the images.