Zhejiang University & Ant Group open source 4DAnyone for high-quality 4D avatars from phone video
aigclink · x · 2026-08-24
Zhejiang University and Ant Group have open-sourced 4DAnyone, a model that turns a casual monocular video into high-quality, viewable dynamic 3D avatars. To address 'structure drift' caused by batch processing due to compute limits, it introduces RCP (compressing reference info) and TCR (transferring structure info between batches), ensuring consistency and enabling cross-role transfer.
More from Multimodal
- Grok 4.6 recreates The Matrix in ASCII art — Daniel_Farinax · 2026-08-24
- Flickart's helicopter video praised for continuity and cockpit realism — eyishazyer · 2026-08-24
- UniSpace: Unified Visual Representation without VAE — meituan-longcat · 2026-08-24
- Wan 3.0 demo impresses with lifelike faces and stable motion — JaynitMakwana · 2026-08-24
- Grafting Krea2 into MiniMax H3: improving texture via component transplant — Key-Philosopher-9327 · 2026-08-24
- Cinematic FLUX 3 Test: Motorcyclist in the Desert with Full Prompt — LudovicCreator · 2026-08-24