发表机构
TU Darmstadt; Zuse School ELIZA（达姆施塔特工业大学; 楚斯学校ELIZA）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
ReconSplat是基于3DGS与MV-LDM的前馈3D场景重建模型，解决未观测区域视图生成与几何一致性的权衡问题，在RealEstate10K、DL3DV-10K基准上优于现有方法，可外推未观测视点并保持精确几何。
AI 中文摘要
我们提出ReconSplat，一种用于3D场景重建的前馈模型，旨在解决未观测区域的合理视图生成与几何一致性之间长期存在的权衡问题，同时提供几何对齐的新视图与精确的深度估计。我们的方法以3D高斯溅射（3DGS）作为中间可微场景表示，将其与多视图潜在扩散模型（MV-LDM）集成，该模型被训练为同时作为外观与场景几何的细化器和修复器。我们通过用前馈3DGS表示编码的外观和几何的变分3D潜在特征引导扩散过程，将其光栅化到2D潜在空间，从而强制几何一致性。ReconSplat在真实世界基准数据集RealEstate10K和DL3DV-10K上生成逼真的新视图和准确的深度图，在具有挑战性的外推设置中优于现有方法。值得注意的是，ReconSplat允许对未观测的具有挑战性的视点进行外推，同时保持连贯且精确的场景几何。
英文摘要
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 diffusion process with variational 3D latent features for appearance and geometry, encoded by the feed-forward 3DGS representation and rasterized to 2D latent space. ReconSplat produces both photorealistic novel views and accurate depth maps on real-world benchmarks, RealEstate10K and DL3DV-10K, outperforming existing methods in challenging extrapolation setups. Notably, ReconSplat allows the extrapolation of unseen and challenging viewpoints jointly with coherent and precise scene geometry.
CommentsECCV 2026. Code and additional visual results are available on our project page: https://visinf.github.io/reconsplat
