GaussFusion：用几何感知视频生成器改进野外 3D 重建
GaussFusion: Improving 3D Reconstruction in the Wild with A Geometry-Informed Video Generator
把 3DGS 渲染出的深度/法线/不透明度/协方差缓冲当作条件输入，用视频到视频生成器做时序一致的后处理精修，可即插即用地接在优化式与前馈式重建管线之后，为消除浮点、闪烁与位姿误差伪影提供了一条通用思路。
原文依据 we render a Gaussian primitive video buffer encoding depth, normals, opacity, and covariance, which the generator refines
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文提出 GaussFusion，一种通过几何感知视频生成改进野外 3D 高斯泼溅（3DGS）重建的新方法。GaussFusion 可缓解 3DGS 中常见的伪影，包括由相机位姿误差、覆盖不完整以及几何初始化含噪所导致的浮点、闪烁与模糊。与先前仅局限于单一重建管线的基于 RGB 的方法不同，本方法引入一个几何感知的视频到视频生成器，可同时精修基于优化的方法与前馈方法的 3DGS 渲染结果。给定一个已有重建，我们渲染一个编码深度、法线、不透明度与协方差的高斯基元视频缓冲，生成器对该缓冲进行精修，从而产生时序一致、无伪影的帧。我们进一步提出一套伪影合成管线，用于模拟多种退化模式，以确保鲁棒性与泛化能力。GaussFusion 在新视角合成基准上取得最优性能，其高效变体在保持相近性能的同时能以 21 FPS 实时运行，从而支持交互式 3D 应用。
英文原文摘要 1122 字符
We present GaussFusion, a novel approach for improving 3D Gaussian splatting (3DGS) reconstructions in the wild through geometry-informed video generation. GaussFusion mitigates common 3DGS artifacts, including floaters, flickering, and blur caused by camera pose errors, incomplete coverage, and noisy geometry initialization. Unlike prior RGB-based approaches limited to a single reconstruction pipeline, our method introduces a geometry-informed video-to-video generator that refines 3DGS renderings across both optimization-based and feed-forward methods. Given an existing reconstruction, we render a Gaussian primitive video buffer encoding depth, normals, opacity, and covariance, which the generator refines to produce temporally coherent, artifact-free frames. We further introduce an artifact synthesis pipeline that simulates diverse degradation patterns, ensuring robustness and generalization. GaussFusion achieves state-of-the-art performance on novel-view synthesis benchmarks, and an efficient variant runs in real time at 21 FPS while maintaining similar performance, enabling interactive 3D applications.