Junjie Shentu

New methods improve image generation accuracy with multi-concept prompts.

AttenCraft improves text-to-image generation by separating concepts for better visuals.

This new method simplifies how computers learn from text, images, sounds, and videos.