arXiv cs.AI· Liyuan Zhu, Shengyu Huang, Amrita Mazumdar, Tianye Li, Zan Gojcic, Gordon Wetzstein, Iro Armeni, Shalini De Mello, Alex Trevithick·· 12 小时前AI 评分35
World from Motion：从单目视频生成动态 3D 高斯重建
World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video
AI 导读
World from Motion 提出一种从单目视频生成可自由渲染动态 3D 高斯表示的方法，通过让视频模型以像素对齐的稠密渲染为条件，修正初始重建的伪影并补全缺失区域。该方法构建了对齐多视角视频对与动态 3DGS 数据集，测试时将生成结果蒸馏回单一一致的动态 3DGS，在 4D 重建上取得新的 SOTA，并可泛化到视角变化大、运动剧烈的野外视频。
正文 · 原文
Abstract:We present World from Motion, a method for generating freely renderable dynamic 3D Gaussian representations from monocular videos. Our approach conditions a video model on dense, pixel-aligned renderings that encode appearance, geometry, and 3D scene motion along both input and target camera trajectories to correct rendering artifacts and fill in missing regions from an initial reconstruction. To train this model, we construct a dataset of aligned multiview video pairs and dynamic 3DGS representations, with simulated artifacts characteristic of monocular reconstruction. At test time, we distill the model's generations, including newly observed regions and motions, back into a single consistent, high-quality dynamic 3DGS, improving both novel-view synthesis and the underlying 3D motion. Our method sets a new state of the art in 4D reconstruction and seamlessly generalizes to in-the-wild videos with large viewpoint changes and dynamic motions.
| Comments: | NeurIPS 2026 Camera Ready. Project page: this https URL |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR) |
| Cite as: | arXiv:2607.01202 [cs.CV] |
| (or arXiv:2607.01202v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2607.01202 arXiv-issued DOI via DataCite |
Submission history
From: Liyuan Zhu [view email]
[v1]
Wed, 1 Jul 2026 17:41:00 UTC (6,531 KB)
[v2]
Fri, 23 Oct 2026 04:36:58 UTC (13,196 KB)
来源：arXiv cs.AI · arxiv.org