ATGS: Anchored Temporal Gaussian Splatting for Long Volumetric Video Representation
2026-08-31 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors present a new method called ATGS for creating 3D videos that let you see scenes from any angle over time. They found that trying to track complex movements over a long time with simple parts causes problems, so their method groups these parts around special points that relate to both space and time, making the videos smoother and more stable. They also only focus on the parts needed at any given moment, which helps the method handle longer videos efficiently. Overall, their approach improves the quality of 3D videos with complicated motions compared to earlier techniques.
Volumetric videoGaussian splattingFree viewpoint renderingTemporal coherenceSpatial-temporal anchorsMotion trackingTemporal windowing3D reconstructionVisual artifacts
Authors
Jiahao Wu, Jie Liang, Die Hu, Jiayu Yang, Kaiqiang Xiong, Xiang Li, Xiaoyun Zheng, Chao Wang, Ronggang Wang
Abstract
Volumetric video enables immersive free viewpoint rendering of dynamic real world scenes, yet existing methods struggle with long sequences and complex motions, often leading to temporal instability and visual artifacts. To address these challenges, we propose \ourname, a Gaussian splatting based framework for volumetric video reconstruction. Our key insight is that explicitly tracking long term complex motion with individual Gaussian primitives is inherently unstable. Instead, we organize Gaussians around time conditioned anchors that localize their spatial and temporal support, thereby reducing long range motion complexity. We further introduce a temporal windowing strategy to activate only anchors relevant to the queried time, which improves scalability and temporal coherence. In addition, to ensure spatial and temporal stability, we design a compact set of multi level anchor features that encode global features, local spatial features, and local temporal features, jointly constraining Gaussian generation. Extensive experiments demonstrate that \ourname \ consistently outperforms prior methods on long sequence volumetric videos with complex motions. Project page: https://github.com/WuJH2001/ATGS.