TRiGS: Temporal Rigid-Body Motion for Scalable 4D Gaussian Splatting
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911559467401216 |
|---|---|
| author | Yeom, Suwoong Nam, Joonsik Choi, Seunggyu Lee, Lucas Yunkyu Kim, Sangmin Park, Jaesik Kim, Joonsoo Yun, Kugjin Kong, Kyeongbo Kang, Sukju |
| author_facet | Yeom, Suwoong Nam, Joonsik Choi, Seunggyu Lee, Lucas Yunkyu Kim, Sangmin Park, Jaesik Kim, Joonsoo Yun, Kugjin Kong, Kyeongbo Kang, Sukju |
| contents | Recent 4D Gaussian Splatting (4DGS) methods achieve impressive dynamic scene reconstruction but often rely on piecewise linear velocity approximations and short temporal windows. This disjointed modeling leads to severe temporal fragmentation, forcing primitives to be repeatedly eliminated and regenerated to track complex nonlinear dynamics. This makeshift approximation eliminates the long-term temporal identity of objects and causes an inevitable proliferation of Gaussians, hindering scalability to extended video sequences. To address this, we propose TRiGS, a novel 4D representation that utilizes unified, continuous geometric transformations. By integrating $SE(3)$ transformations, hierarchical Bezier residuals, and learnable local anchors, TRiGS models geometrically consistent rigid motions for individual primitives. This continuous formulation preserves temporal identity and effectively mitigates unbounded memory growth. Extensive experiments demonstrate that TRiGS achieves high fidelity rendering on standard benchmarks while uniquely scaling to extended video sequences (e.g., 600 to 1200 frames) without severe memory bottlenecks, significantly outperforming prior works in temporal stability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_00538 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TRiGS: Temporal Rigid-Body Motion for Scalable 4D Gaussian Splatting Yeom, Suwoong Nam, Joonsik Choi, Seunggyu Lee, Lucas Yunkyu Kim, Sangmin Park, Jaesik Kim, Joonsoo Yun, Kugjin Kong, Kyeongbo Kang, Sukju Computer Vision and Pattern Recognition Recent 4D Gaussian Splatting (4DGS) methods achieve impressive dynamic scene reconstruction but often rely on piecewise linear velocity approximations and short temporal windows. This disjointed modeling leads to severe temporal fragmentation, forcing primitives to be repeatedly eliminated and regenerated to track complex nonlinear dynamics. This makeshift approximation eliminates the long-term temporal identity of objects and causes an inevitable proliferation of Gaussians, hindering scalability to extended video sequences. To address this, we propose TRiGS, a novel 4D representation that utilizes unified, continuous geometric transformations. By integrating $SE(3)$ transformations, hierarchical Bezier residuals, and learnable local anchors, TRiGS models geometrically consistent rigid motions for individual primitives. This continuous formulation preserves temporal identity and effectively mitigates unbounded memory growth. Extensive experiments demonstrate that TRiGS achieves high fidelity rendering on standard benchmarks while uniquely scaling to extended video sequences (e.g., 600 to 1200 frames) without severe memory bottlenecks, significantly outperforming prior works in temporal stability. |
| title | TRiGS: Temporal Rigid-Body Motion for Scalable 4D Gaussian Splatting |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.00538 |