2026-11-05 · America/Los_Angeles · 论文 · #16

AnchorComplete: Recovering Complete Scene Meshes from Partial Point Clouds

Most image-based 3D reconstructors primarily recover visible surfaces, producing partial point clouds with large holes and leaving unobserved geometry unresolved. To address this limitation, we introduce AnchorComplete, a framework that reconstructs complete scene meshes from partial point clouds while preserving observed geometry and supporting scenes of varying scales. Building on TRELLIS.2, our method combines partial-point conditioning, which integrates global scene tokens with local spatial evidence to anchor generation to the input, and metric-scale coarse-to-fine generation, which…

热度 37.1 / 100；排名与评分保留该期记录。