Research map / Video / Video generation
"EMDM: Efficient Motion Diffusion Model for Fast, High-Quality Human Motion Generation"
Wenyang Zhou, Zhiyang Dou, Zeyu Cao, Zhouyingcheng Liao, Jingbo Wang, Wenjia Wang, Yuan Liu, Taku Komura, Wenping Wang, Lingjie Liu
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 geometric loss (R=100) to the denoising GAN training (ablation) | EMDM · HumanML3D | FID: 9.308 -> 0.112 | Improved |
| increase geometric loss weight R from 100 to 1000 (ablation) | EMDM · HumanML3D | FID: 0.112 -> 0.195 | Hurt |
| increase sampling steps from 10 to 50 (ablation) | EMDM · HumanML3D | FID: 0.112 -> 0.216 | Hurt |
| reduce sampling steps from 10 to 1 (pure GAN) (ablation) | EMDM · HumanML3D | FID: 0.112 -> 5.640 | Hurt |
| reduce sampling steps from 10 to 5 (ablation) | EMDM · HumanML3D | FID: 0.112 -> 1.306 | Hurt |
| condition the discriminator on text/action (ablation) | EMDM · KIT | FID: 0.510 -> 0.506 | No clear effect |
| use EMDM conditional denoising GAN instead of naive DDGAN + MDM | motion diffusion few-step sampling · HumanML3D | FID: 31.085 -> 0.112 | Improved |
| reduce MDM to 10 steps with DDIM instead of EMDM large-step GAN denoising | MDM · HumanML3D | FID: 0.508 -> 33.232 | Hurt |
| replace MDM sampling with EMDM | text-to-motion · HumanML3D | running time per frame (ms): 62.505 -> 0.280 | Improved |
| EMDM vs MLD on unconditional generation | EMDM · AMASS unconditional | FID: 1.4 -> 3.46 | Hurt |
Later papers that compared with this method
2 later accepted papers report EMDM as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- SViMo: Synchronized Diffusion for Video and Motion Generation in Hand-object Interaction Scenarios (NeurIPS 2025 · better on 4 of 4 shared results)
- GenM3: Generative Pretrained Multi-path Motion Model for Text Conditional Human Motion Generation (ICCV 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
- MixerMDM: Learnable Composition of Human Motion Diffusion Models (CVPR 2025 · similarity 0.89)
- Towards Detailed Text-to-Motion Synthesis via Basic-to-Advanced Hierarchical Diffusion Model (AAAI 2024 · similarity 0.89)
- Optimizing Diffusion Noise Can Serve As Universal Motion Priors (CVPR 2024 · similarity 0.89)
- Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregression (CVPR 2025 · similarity 0.88)
- MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation (AAAI 2024 · similarity 0.88)
- AAMDM: Accelerated Auto-regressive Motion Diffusion Model (CVPR 2024 · 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.
