WorldRecon first runs a reconstruction model (VGGT-Ω) on unposed images to obtain a noisy global point cloud and approximate camera poses.
Generative framework reconstructs complete large-scale 3D meshes from unposed photos
WorldRecon combines noisy point-cloud scaffolding, block-wise generation, and neighbour-aware inpainting to produce coherent, detailed meshes, cutting Chamfer distance by up to 43.1 percent on real-world scenes.
Chinese Tech
Mengchen Zhang · Yang Li · Chunshi Wang · Tong Wu · Chunchao Guo
Zhejiang University · Tencent Hunyuan · Fudan University
Research Digest··2 min read
The authors introduce WorldRecon, a framework that reconstructs complete, large-scale 3D scene meshes with PBR materials from unposed multi-view RGB images.
Why this paper
From Tencent Hunyuan and 2 others
In one line
WorldRecon reconstructs complete, large-scale 3D scene meshes from unposed multi-view images.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (2 noted)
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§