Velox: Learning Representations of 4D Geometry and Appearance
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913094108708864 |
|---|---|
| author | Malik, Anagh Chan, Dorian Zhao, Xiaoming Lindell, David B. Tuzel, Oncel Chang, Jen-Hao Rick |
| author_facet | Malik, Anagh Chan, Dorian Zhao, Xiaoming Lindell, David B. Tuzel, Oncel Chang, Jen-Hao Rick |
| contents | We introduce a framework for learning latent representations of 4D objects which are descriptive, faithfully capturing object geometry and appearance; compressive, aiding in downstream efficiency; and accessible, requiring minimal input, i.e., an unstructured dynamic point cloud, to construct. Specifically, Velox trains an encoder to compress spatiotemporal color point clouds into a set of dynamic shape tokens. These tokens are supervised using two complementary decoders: a 4D surface decoder, which models the time-varying surface distribution capturing the geometry; and a Gaussian decoder, which maps the tokens to 3D Gaussians, helping learn appearance. To demonstrate the utility of our representation, we evaluate it across three downstream tasks -- video-to-4D generation, 3D tracking, and cloth simulation via image-to-4D generation -- and observe strong performances in all settings. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_04527 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Velox: Learning Representations of 4D Geometry and Appearance Malik, Anagh Chan, Dorian Zhao, Xiaoming Lindell, David B. Tuzel, Oncel Chang, Jen-Hao Rick Computer Vision and Pattern Recognition We introduce a framework for learning latent representations of 4D objects which are descriptive, faithfully capturing object geometry and appearance; compressive, aiding in downstream efficiency; and accessible, requiring minimal input, i.e., an unstructured dynamic point cloud, to construct. Specifically, Velox trains an encoder to compress spatiotemporal color point clouds into a set of dynamic shape tokens. These tokens are supervised using two complementary decoders: a 4D surface decoder, which models the time-varying surface distribution capturing the geometry; and a Gaussian decoder, which maps the tokens to 3D Gaussians, helping learn appearance. To demonstrate the utility of our representation, we evaluate it across three downstream tasks -- video-to-4D generation, 3D tracking, and cloth simulation via image-to-4D generation -- and observe strong performances in all settings. |
| title | Velox: Learning Representations of 4D Geometry and Appearance |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.04527 |