Research map / Video / Video generation
Training-free Motion Factorization for Compositional Video Generation
Zixuan Wang, Ziqin Zhou, Feng Chen, Duo Peng, Yixin Hu, Changsheng Li, Yinjie Lei
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 LLaMA-70B instead of LLaMA-8B for motion reasoning (ablation) | Ours on VideoCrafter-v2.0 · CVGBench-m | Dynamic Degree: 75.34% -> 82.21% | Improved |
| use LLaMA-70B instead of LLaMA-8B for motion reasoning (ablation) | Ours on VideoCrafter-v2.0 · CVGBench-m | Subject Consistency: 97.55% -> 98.40% | Improved |
| use LLaMA-70B instead of LLaMA-8B for motion reasoning (ablation) | Ours on CogVideoX-2B · CVGBench-m | Dynamic Degree: 94.77% -> 96.00% | Improved |
| add reference conditioned guidance (ablation) | Ours on VideoCrafter-v2.0 · CVGBench-m | Dynamic Degree: 38.40% -> 51.60% | Improved |
| add geometric invariance guidance (ablation) | Ours on VideoCrafter-v2.0 · CVGBench-m | Dynamic Degree: 38.40% -> 53.60% | Improved |
| add spatial deformation guidance (ablation) | Ours on VideoCrafter-v2.0 · CVGBench-m | Dynamic Degree: 38.40% -> 74.85% | Improved |
| add all three guidance branches (ablation) | Ours on VideoCrafter-v2.0 · CVGBench-m | Dynamic Degree: 38.40% -> 82.21% | Improved |
| add reference conditioned guidance (ablation) | Ours on CogVideoX-2B · CVGBench-m | Subject Consistency: 91.00% -> 95.98% | Improved |
| add geometric invariance guidance (ablation) | Ours on CogVideoX-2B · CVGBench-m | Subject Consistency: 91.00% -> 96.21% | Improved |
| add spatial deformation guidance (ablation) | Ours on CogVideoX-2B · CVGBench-m | Subject Consistency: 91.00% -> 96.13% | Improved |
| add all three guidance branches (ablation) | Ours on CogVideoX-2B · CVGBench-m | Subject Consistency: 91.00% -> 98.27% | Improved |
| add spatial deformation guidance on top of RCG+GIG (ablation) | Ours on CogVideoX-2B · CVGBench-m | Dynamic Degree: 94.56% -> 96.00% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CogVideoX-2B — CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer (ICLR 2025) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Motion Prompting: Controlling Video Generation with Motion Trajectories (CVPR 2025 · similarity 0.89)
- CAGE: Unsupervised Visual Composition and Animation for Controllable Video Generation (AAAI 2025 · similarity 0.88)
- VideoMage: Multi-Subject and Motion Customization of Text-to-Video Diffusion Models (CVPR 2025 · similarity 0.88)
- ONE-SHOT: Compositional Human-Environment Video Synthesis via Spatial-Decoupled Motion Injection and Hybrid Context Integration (NeurIPS 2026 · similarity 0.87)
- Deep Compositional Phase Diffusion for Long Motion Sequence Generation (NeurIPS 2025 · similarity 0.87)
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.87)
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.
