Peking University's UltraShape 1.0 enables high-fidelity 3D generation via geometric refinement
tom_doerr · x · 2026-08-15
Peking University's YuanGroup released UltraShape 1.0, a scalable diffusion framework for high-fidelity 3D shape generation. The method uses a two-stage pipeline: first synthesizing a coarse global structure, then refining it via voxel-based techniques to produce detailed geometry. The project also introduces a comprehensive data processing pipeline featuring novel watertight processing and high-quality filtering to enhance public 3D datasets by removing defects and preserving fine details.
More from Multimodal
- How meloform works: WASM pitch detection plus differentiable circuits learning 1,022 sine coefficients — henloitsjoyce · 2026-10-16
- Documentary filmmaker urges peers to watch 'You Can See Everything' — kyliebytes · 2026-10-16
- Redditor's first Cumfy/Seedance video nails a cinematic one-shot generation — Wave_N_Wind · 2026-10-16
- Kling 4 returns after 8 months: 10 hyper-realistic AI ads for under $5 — CurieuxExplorer · 2026-10-16
- Shanghai AI Lab unveils Intern Lumina U2, a diffusion LLM unifying image, video and 3D tasks — Shanghai-AI-Laboratory · 2026-10-16
- NAVER AI Lab proposes Speculative Forcing to speed up autoregressive video generation — naver-ai · 2026-10-16