Researchers introduce ViTok, a Vision Transformer-based auto-encoder that scales visual tokenization to enhance image and video generation while reducing computational costs.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback