Velox: Spatial Memory Makes World Models Efficient
Video world models turn actions and camera motion into long visual rollouts, making video a natural interface for interactive simulation. Such interaction requires responsive generation, yet coherent rollouts must retain visual memory of what has already been generated. Chunked video diffusion models usually keep this context in growing KV caches, so later chunks become progressively slower. To address this bottleneck, we introduce Velox, a training-free attention router for region-level context selection using frustum slab correspondences from streaming geometry. Velox represents historical blocks as camera frustum slabs with depth intervals and obtains current-block depth through recenc...
| Authors | Qiuhong Shen, Shizun Wang, Xinchao Wang |
|---|---|
| Primary category | cs.CV |
| Published | 15 October 2026 |
| Last updated at source | 15 October 2026 |
Recent changes
- 16 Oct 2026 Description: Video world models turn actions and camera motion into long visual rollouts, making video a natural interface for interactive simulation. Such interaction requires responsive generation, yet coherent rollouts must retain visual memory of wh → Video world models turn actions and camera motion into long visual rollouts, making video a natural interface for interactive simulation. Such interaction requires responsive generation, yet coherent rollouts must retain visual memory of wh
- 16 Oct 2026 Topics added cs.CV removed hf-daily-papers, video world models, chunked video diffusion, kv caches, attention router, frustum slab correspondences, multiwarp, flashattention, tiled execution
Changes detected by Silytics between refreshes of the source.
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 16 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.