StateCraft: Augmenting Video World Models with Abstract-State Reasoning
Video world models aim to simulate complex environments with realistic appearance and dynamics.
ProofPaper ↗
Key points
- Maintaining consistency over long interactions requires tracking past actions and understanding how those actions affect the environment.
- We introduce a paradigm that decouples a persistent, updatable abstract state from the visual state.
- We propose StateCraft, a novel architecture that pairs a language model with a video diffusion model to evolve the dynamic world.
- To support training and evaluation, we develop a data collection pipeline that produces informative, densely annotated trajectories at scale without human or agent supervision, together with a benchmark measuring generation quality, memory, and controllability.
Sources (1)
- [1]StateCraft: Augmenting Video World Models with Abstract-State ReasoningarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 14, 08:39 AM
Video world models aim to simulate complex environments with realistic appearance and dynamics.
Maintaining consistency over long interactions requires tracking past actions and understanding how those actions affect the environment.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 14, 2026SonoWorld: Towards Interactive Pose-Controllable Ultrasound World Models
- Oct 14, 2026Glance Forcing: Making Bidirectional Diffusion Autoregressive at a Glance
- Oct 13, 2026VigenFlow: FLUX.1-schnell, FLUX.2-klein-4B and Z-Image-Turbo running locally on the AMD Ryzen AI NPU, no discrete GPU needed (Windows + Ubuntu)
- Oct 13, 2026GAD: Geometry Attention Distillation for Consistent Video Generation
- Oct 13, 2026LongForcing: Scaling Temporal Manifold Matching for Long-Horizon World Rollout
- Oct 13, 2026LinMorph: Learning to Morph Weights for Efficient Video DiT Linearization