Research map / Video / Video generation
DrawMotion: Generating 3D Human Motions by Freehand Drawing
Tao Wang, Lei Jin, Zhihua Wu, Qiang He, Jiaming Chu, Yu Cheng, Junliang Xing, Jian Zhao, Shuicheng Yan, Li Wang
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 Mahalanobis-distance clipping (threshold 1) to IFG (ablation) | IFG · KIT-ML | Traj.err.: 0.126 -> 0.069 | Improved |
| add Mahalanobis clipping to IFG (FID) (ablation) | IFG · KIT-ML | FID: 0.185 -> 0.141 | Improved |
| use deeper MCM layer for guidance (layer 1 to 3) (ablation) | IFG · KIT-ML | Traj.err.: 0.112 -> 0.099 | Improved |
| use deeper MCM layer for guidance (layer 1 to 3) (ablation) | IFG · KIT-ML | FID: 0.131 -> 0.146 | Hurt |
| increase IFG repeat from 10 to 100 (lr 50) (ablation) | IFG · KIT-ML | Traj.err.: 0.062 -> 0.026 | Improved |
| increase IFG repeat from 10 to 100 (time per batch) (ablation) | IFG · KIT-ML | Time/Batch (s): 7 -> 44 | Hurt |
| use efficient attention for text decoder instead of dot-product (ablation) | DrawMotion · KIT-ML | FID: 0.153 -> 0.135 | Improved |
| use efficient attention for draw decoder instead of dot-product (ablation) | DrawMotion · KIT-ML | Traj.Err.: 0.032 -> 0.113 | Hurt |
| add condition fusion and latent encoder to MCM versus mask baseline (ablation) | MCM · KIT-ML | FID: 0.151 -> 0.135 | Improved |
| add condition fusion alone (no latent encoder) (ablation) | MCM · KIT-ML | FID: 0.151 -> 0.187 | Hurt |
| bias reverse process toward drawing, p=80% instead of 20% (ablation) | DrawMotion · KIT-ML | StiSim: 52.2% -> 54.8% | Improved |
| train/test with 3 stickmen rather than 0 (ablation) | DrawMotion · KIT-ML | StiSim: 42.65% -> 51.99% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- StickMotion — StickMotion: Generating 3D Human Motions by Drawing a Stickman (CVPR 2025) · better on 4 of 6 shared results
- ReMoDiffuse — ReMoDiffuse: Retrieval-Augmented Motion Diffusion Model (ICCV 2023) · better on 2 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
- StickMotion: Generating 3D Human Motions by Drawing a Stickman (CVPR 2025 · similarity 0.95)
- AirSketch: Generative Motion to Sketch (NeurIPS 2024 · similarity 0.88)
- Generating Human Motion in 3D Scenes from Text Descriptions (CVPR 2024 · similarity 0.87)
- Sitcom-Crafter: A Plot-Driven Human Motion Generation System in 3D Scenes (ICLR 2025 · similarity 0.87)
- InterDreamer: Zero-Shot Text to 3D Dynamic Human-Object Interaction (NeurIPS 2024 · similarity 0.87)
- Enhanced Fine-Grained Motion Diffusion for Text-Driven Human Motion Synthesis (AAAI 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.
