Research map / Video / Video segmentation and tracking
FlowMotion: Training-Free Flow Guidance for Video Motion Transfer
Zhen Wang, Youcan Xu, Jun Xiao, Long Chen
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 |
|---|---|---|---|
| add velocity regularization (ablation) | FlowMotion · Video motion transfer | Motion Fidelity: 0.809 -> 0.850 | Improved |
| add difference alignment loss (ablation) | FlowMotion · Video motion transfer | Motion Fidelity: 0.842 -> 0.850 | Improved |
| add velocity regularization (ablation) | FlowMotion · Video motion transfer | Text Similarity: 0.313 -> 0.347 | Improved |
| build guidance from latent prediction rather than velocity outputs (ablation) | FlowMotion · Wan2.1-1.3B memory | GPU memory (G): 93.1 -> 19.3 | Improved |
| build guidance from attention maps or features (ablation) | FlowMotion · Wan2.1-1.3B memory | GPU memory (G): OOM | Hurt |
| use FlowMotion instead of DiTFlow | motion transfer · Video motion transfer | Motion Fidelity: 0.691 -> 0.850 | Improved |
| use clean source latent instead of latent prediction as source motion representation (ablation) | FlowMotion · qualitative examples | motion transfer precision vs text alignment | Mixed |
| use FlowMotion instead of training-based methods | motion transfer · Video motion transfer | Text Similarity: 0.347 vs 0.340 (DeT) | No clear effect |
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.
- MotionClone — MotionClone: Training-Free Motion Cloning for Controllable Video Generation (ICLR 2025) · better on 1 of 1 shared result
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
- FlowMo: Variance-Based Flow Guidance for Coherent Motion in Video Generation (NeurIPS 2025 · similarity 0.91)
- PhysFlow: Physics-Intrinsic Velocity Regularization for Motion-Intensive Video Generation (NeurIPS 2026 · similarity 0.89)
- FlowVid: Taming Imperfect Optical Flows for Consistent Video-to-Video Synthesis (CVPR 2024 · similarity 0.88)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.88)
- Optical-Flow Guided Prompt Optimization for Coherent Video Generation (CVPR 2025 · similarity 0.88)
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (CVPR 2025 · similarity 0.88)
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.
