Video World Models · 2026-10-06
Kepler4D: Controllable Future Video Generation via 4D Scene State Evolution
Feiran Wang, Bin Duan, Junyi Wu, Gaowen Liu, Yan Yan
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
Kepler4D is a controllable future video generation framework that builds a shared 4D scene state from a monocular video (static background geometry, foreground object motion histories, coarse spatial supports, and semantic context). Chain-of-Motion selects structured speed/heading decisions (optionally a lower-bound height constraint) and deterministic rollout produces editable future 3D object-center trajectories, which are rendered as geometric controls for a pretrained video generator.
Why it matters
The authors connect explicit future object-center trajectories to video synthesis by evolving a single shared 4D scene state and rendering predicted/edited trajectories as geometric control maps. This aims to enable automatic future evolution and language-guided trajectory editing while preserving reconstructed scene structure during camera motion.
Method
- Construct a shared 4D scene state from a monocular video: static background geometry, an object proxy bank with semantic labels, observed 3D center history, and coarse spatial supports, plus semantic context.
- Use Chain-of-Motion to summarize observed motion and scene context, then infer structured speed/heading decisions and whether to apply a lower bound on predicted object-center height.
- Apply deterministic 3D trajectory rollout to produce future (or edited) object-center trajectories, render them as geometric control maps (with background and foreground controls), and guide a pretrained video generator.
Limitation
Kepler4D targets editable object-center motion under a supplied camera path. Fixed coarse supports do not explicitly model joint motion, shape deformation, or contact dynamics, so geometric guidance alone does not guarantee physically correct interactions.
Abstract (from arXiv)
Video world models aim to preserve scene structure and predict how dynamic objects evolve beyond visual observations. We present Kepler4D, a framework for future video generation through explicit 4D scene state evolution. Given a monocular video, Kepler4D constructs a shared 3D representation of background geometry, object motion histories, coarse spatial supports, and semantic context. Chain-of-Motion summarizes observed motion and uses a vision-language model to select structured speed and heading decisions and decide whether to bound object-center height from below. A deterministic rollout converts these decisions into future object trajectories for inspection and editing before synthesis. We render the evolving proxies into geometric controls for a pretrained video generator, separating coarse object motion from the synthesis of appearance and articulation. Experiments on real-world videos demonstrate that Kepler4D enables controllable object motion and plausible future rollout while preserving scene consistency.
Related papers
- Demiurge: Forging Navigable 4D Worlds via Autoregressive Video Rerendering
- 4Director: Controlling Video World Models with Rigid 3D Geometry
- Learning the 4D World through Point Trajectory Forecasting
- 4LC: 4D Latent Cache for Efficient Conditioning in Dynamic World Modeling
- Momentum: Persistent Dynamics for Video World Models
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-02 · Indexed by Pub-AI: 2026-10-06 UTC
Kepler4D: Controllable Future Video Generation via 4D Scene State Evolution
Preprint posted: 2026-10-02 · Original record ↗
- arxiv
- 2610.04152
Explicit source versions (1)
- v1 · 2026-10-02
- Preprint available
Source date: 2026-10-02 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
Video world models aim to preserve scene structure and predict how dynamic objects evolve beyond visual observations. We present Kepler4D, a framework for future video generation through explicit 4D scene state evolution. Given a monocular video, Kepler4D constructs a shared 3D representation of background geometry, object motion histories, coarse spatial supports, and semantic context. Chain-of-Motion summarizes observed motion and uses a vision-language model to select structured speed and heading decisions and decide whether to bound object-center height from below. A deterministic rollout converts these decisions into future object trajectories for inspection and editing before synthesis. We render the evolving proxies into geometric controls for a pretrained video generator, separating coarse object motion from the synthesis of appearance and articulation. Experiments on real-world videos demonstrate that Kepler4D enables controllable object motion and plausible future rollout while preserving scene consistency.