[论文] Vista4D: Video Reshooting with 4D Point Clouds

## 论文概要 **研究领域**: CV **作者**: Kuan Heng Lin, Zhizheng Liu, Pablo Salamanca **发布时间**: 2026-04-23 **arXiv**: [2604.21938](https://arxiv.org/abs/2604.21938) ## 中文摘要 我们提出了Vista4D，一个鲁棒且灵活的视频重拍框架，将输入视频和目标相机基于4D点云进行 grounding。具体而言，给定一个输入视频，我们的方法从不同相机轨迹和视角重新合成具有相同动态的场景。现有的视频重拍方法通常难以处理真实世界动态视频的深度估计伪影，同时无法保留内容外观，也无法为具有挑战性的新轨迹维持精确的相机控制。我们构建了基于4D ground的点云表示，通过静态像素分割和4D重建来显式保留已见内容并提供丰富的相机信号，我们使用重建的多视角动态数据进行训练，以增强对真实世界推理期间点云伪影的鲁棒性。我们的结果表明，与各种视频和相机路径下的最先进基线相比，4D一致性、相机控制和视觉质量均有提升。此外，我们的方法可推广到动态场景扩展和4D场景重组等真实世界应用。 ## 原文摘要 We present Vista4D, a robust and flexible video reshooting framework that grounds the input video and target cameras in a 4D point cloud. Specifically, given an input video, our method re-synthesizes the scene with the same dynamics from a different camera trajectory and viewpoint. Existing video reshooting methods often struggle with depth estimation artifacts of real-world dynamic videos, while also failing to preserve content appearance and failing to maintain precise camera control for challenging new trajectories. We build a 4D-grounded point cloud representation with static pixel segmentation and 4D reconstruction to explicitly preserve seen content and provide rich camera signals, and we train with reconstructed multiview dynamic data for robustness against point cloud artifacts dur... --- *自动采集于 2026-04-25* #论文 #arXiv #CV #小凯