Research map / Video / Video generation
CMDer: Controllable Mode Decomposition-Based Single Motion Synthesis with Diffusion
Junliang Chen, Sihang Chen, Xiaojuan Gu, Yoonsang Lee, Kevin Romond, Fang-Lue Zhang
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 Mode Enhancement Block (ablation) | CMDer · Mixamo benchmark | Coverage: 56.60 -> 97.94 | Improved |
| Use MVMD instead of FFT K-band decomposition (ablation) | CMDer · Mixamo benchmark | Harmonic mean: 0.46 -> 0.78 | Improved |
| Use CMDer vs SinMDM | CMDer · Mixamo benchmark | Harmonic mean: 0.76 -> 0.78 | Improved |
| Use CMDer vs GenMM on coverage | CMDer · Mixamo benchmark | Coverage: 99.26 -> 97.94 | Hurt |
| Increase depthwise kernel from 3x3 to 21x21 (ablation) | CMDer · Mixamo benchmark | Coverage / Global diversity: Cov 93.03 -> 98.10; GloDiv 1.65 -> 1.27 | Mixed |
| Increase number of modes K from 4 to 8 (ablation) | MVMD · MVMD sensitivity | Reconstruction (%): 96.75 -> 98.70 | Improved |
| Increase K from 8 to 20 (ablation) | MVMD · MVMD sensitivity | Frequency separation (Hz): 0.533 -> 0.376 | Hurt |
| CMDer vs SinMDM on perceived diversity | CMDer · User study | Diversity (1-5): 3.29 vs 3.30 | No clear effect |
| CMDer vs SinMDM on perceived quality | CMDer · User study | Quality (1-5): 3.39 -> 3.71 | Improved |
Closest papers by meaning
- MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation (AAAI 2024 · similarity 0.88)
- Single Motion Diffusion (ICLR 2024 · similarity 0.88)
- Enhanced Fine-Grained Motion Diffusion for Text-Driven Human Motion Synthesis (AAAI 2024 · similarity 0.87)
- MotionDreamer: One-to-Many Motion Synthesis with Localized Generative Masked Transformer (ICLR 2025 · similarity 0.87)
- MMM: Generative Masked Motion Model (CVPR 2024 · similarity 0.87)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · 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.
