Zhejiang Univ Open-Sources PhyEdit for 3D-Aware Image Editing
量子位 · wechat · 2026-08-13
The ReLER team at Zhejiang University has proposed and open-sourced PhyEdit, a DiT-based image editing model guided by explicit 3D geometric previews. It solves common failures in traditional models when handling object scale, occlusion, and multi-object manipulation.
The method decouples geometry from generation: it estimates scene depth and camera parameters, moves object point clouds in 3D space to create a rough preview, and then uses a generative model to recover textures and details. The team also released the RealManip-40K dataset and the ManipEval benchmark. Experiments show PhyEdit significantly outperforms existing commercial models in 3D accuracy while retaining standard editing capabilities.
More from Multimodal
- Musician Shares AI Remix Demo as Industry Hesitates on First-Mover Backlash — jordiponsdotme · 2026-08-13
- AI Music Invades the Charts: Guardian Uncovers Industry Concerns — nordicinst · 2026-08-13
- Stable Audio 3 Medium Ranks 2nd in Music Generation Evaluation — jordiponsdotme · 2026-08-13
- Open Source Online3DViewer: Explore 3D Models Directly in Your Browser — tom_doerr · 2026-08-13
- Minimax H3 Face Detailer Workflow Fixes Distant Face Distortion — listopalafoto · 2026-08-13
- Minimax Music 3 Open-Weight Release Imminent as Samples Surface — rerri · 2026-08-13