Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models
Video world models have shown immense potential in simulating the physical world, yet existing memory mechanisms primarily treat environments as static canvases. When dynamic subjects hide out of sight and later re-emerge, current methods often struggle, leading to frozen, distorted, or vanishing subjects. To address this, we introduce Hybrid Memory, a novel paradigm requiring models to simultaneously act as precise archivists for static backgrounds and vigilant trackers for dynamic subjects, ensuring motion continuity during out-of-view intervals. To facilitate research in this direction, we construct HM-World, the first large-scale video dataset dedicated to hybrid memory. It features 5...
| Authors | Kaijin Chen, Dingkang Liang, Xin Zhou et al. |
|---|---|
| Primary category | cs.CV |
| Published | 26 March 2026 |
| Last updated at source | 28 March 2026 |
Connected in the knowledge graph
- KlingTeam/HM-WorldDataset · cited by
Models and datasets referencing this paper
- KlingTeam/HM-WorldDataset
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.