Paper Overview
Research Area: Computer Vision Authors: Mengchen Zhang, Yang Li, Chunshi Wang, Tong Wu, Chunchao Guo Published: 2026-10-13 arXiv: 2610.18846
Abstract
Reconstructing complete, high-quality meshes of large-scale scenes from unposed multi-view images requires faithful geometric grounding while plausibly completing surfaces that are unobserved or only sparsely captured. We present WorldRecon, a generative multi-view reconstruction framework for recovering complete, large-scale 3D scene meshes from unposed RGB images. WorldRecon uses a coarse, noisy global point cloud and camera poses estimated from the input images by a reconstruction model as a global scaffold for whole-scene generation and partitions it into overlapping spatial blocks.
To robustly leverage these imperfect geometric priors, we equip the native 3D generative model TRELLIS.2 with a dedicated reconstruction-based conditioning architecture that generates each block from its local point-cloud crop and block-related image observations, producing sparse structures, detailed geometry, and PBR materials. To maintain scene-level consistency, we introduce neighbor-aware latent inpainting and a causal world-generation process that propagates structure, geometry, and material context across overlapping blocks. Together, these designs extend multi-view generative reconstruction to large-scale scenes, yielding coherent and unified meshes.
Extensive experiments show that WorldRecon achieves state-of-the-art reconstruction performance, reducing Chamfer Distance by 14.9% and 43.1% and improving F-score by 17.3% and 12.0% on synthetic and real indoor scenes, respectively.
---
*Automatically collected on 2026-10-15.*