Jinbin Bai

Latest

• RecTok: Reconstruction Distillation along Rectified Flow
• Muddit: Liberating generation beyond text-to-image with a unified discrete diffusion model
• Conditional panoramic image generation via masked autoregressive modeling
• Decouple and track: Benchmarking and improving video diffusion transformers for motion transfer
• DreamRelation: Bridging Customization and Relation Generation