SOURCE-LINKED INTELLIGENCE

ReconSplat: Generalizable 3D Scene Reconstruction Beyond Observed Views

arXiv · AI, language, vision and robotics · article · Aug 28, 2026 · UTC

We introduce ReconSplat, a feed-forward model for 3D scene reconstruction that aims to address the longstanding trade-off between plausible view generation for unobserved regions and geometric consistency, providing both geometrically aligned novel views and sharp depth estimates. Our approach builds on 3D Gaussian splatting (3DGS) as an intermediate differentiable scene representation and integrates it with a multi-view latent diffusion model (MV-LDM) trained to act simultaneously as a refiner and an inpainter for appearance and scene geometry. We enforce geometric consistency by guiding the

Read original source ↗ Open in workspace

recordType

paper

region

Global