Research map / Video / Video generation
VerseCrafter: Dynamic Realistic Video World Model with 4D Geometric Control
Sixiao Zheng, Minghao Yin, Wenbo Hu, Xiaoyu Li, Ying Shan, Yanwei Fu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use 3D Gaussian trajectories instead of 3D bounding boxes (ablation) | VerseCrafter · VerseControl4D (joint) | ObjMC: 4.520 -> 2.507 | Improved |
| use 3D Gaussian trajectories instead of 3D point trajectories (ablation) | VerseCrafter · VerseControl4D (joint) | ObjMC: 6.896 -> 2.507 | Improved |
| use 3D Gaussian trajectories instead of 3D bounding boxes (ablation) | VerseCrafter · VerseControl4D (joint) | VBench Overall Score: 85.45 -> 88.10 | Improved |
| add depth channel to control maps (ablation) | VerseCrafter · VerseControl4D (joint) | ObjMC: 4.929 -> 2.507 | Improved |
| add depth channel to control maps (ablation) | VerseCrafter · VerseControl4D (joint) | RotErr: 1.177 -> 0.890 | Improved |
| decouple background and foreground control maps instead of merging (ablation) | VerseCrafter · VerseControl4D (joint) | ObjMC: 3.726 -> 2.507 | Improved |
| decouple background and foreground control maps (ablation) | VerseCrafter · VerseControl4D (joint) | Dynamic Degree: 83.72 -> 86.26 | Mixed |
| VerseCrafter vs Uni3C on joint control | VerseCrafter · VerseControl4D (joint) | RotErr: 1.361 -> 0.890 | Improved |
| VerseCrafter vs FlashWorld on camera-only control | VerseCrafter · VerseControl4D static subset | RotErr: 1.792 -> 0.650 | Improved |
| VerseCrafter vs Yume/Voyager on Motion Smoothness | VerseCrafter · VerseControl4D static subset | Motion Smoothness: 99.39 vs 97.62 | No clear effect |
Closest papers by meaning
- DynamicVerse: A Physically-Aware Multimodal Framework for 4D World Modeling (NeurIPS 2025 · similarity 0.89)
- WorldReel: 4D Video Generation with Consistent Geometry and Motion Modeling (CVPR 2026 · similarity 0.88)
- WorldStereo: Bridging Camera-Guided Video Generation and Scene Reconstruction via 3D Geometric Memories (CVPR 2026 · similarity 0.88)
- WorldWeaver: Generating Long-Horizon Video Worlds via Rich Perception (NeurIPS 2025 · similarity 0.87)
- Controlling Space and Time with Diffusion Models (ICLR 2025 · similarity 0.86)
- OptiWorld: Optimal Control for Video World Generation under Physical Constraints (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
