Research map / Video / Video generation
You Think, You ACT: The New Task of Arbitrary Text to Motion Generation
Runqi Wang, Caoyuan Ma, Guopeng Li, Hanrui Xu, Yuke Li, Zheng 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 dataset filtering (ablation) | TAAT · HumanML3D++ Scene Text (trained) | Hit Accuracy: 78.6 -> 79.9 | Improved |
| add dataset filtering (ablation) | TAAT · HumanML3D++ Scene Text (trained) | Mean Hit Distance: 1.195 -> 1.075 | Improved |
| add verb usage restriction to prompt (ablation) | Scene Text generation prompt · Scene Text generation prompt (human eval) | Acc.: 58.2 -> 98.4 | Improved |
| add quantity requirement to prompt (ablation) | Scene Text generation prompt · Scene Text generation prompt (human eval) | Acc.: 76.4 -> 98.4 | Improved |
| add few-shot examples to prompt (ablation) | Scene Text generation prompt · Scene Text generation prompt (human eval) | Acc.: 37.6 -> 98.4 | Improved |
| add causality description to prompt (ablation) | Scene Text generation prompt · Scene Text generation prompt (human eval) | Acc.: 81.8 -> 98.4 | Improved |
| train on HumanML3D++ scene texts instead of zero-shot | TAAT · HumanML3D++ Scene Text | Hit Accuracy: 75.1 -> 79.9 | Improved |
| train on HumanML3D++ scene texts instead of zero-shot | MDM · HumanML3D++ Scene Text | Hit Accuracy: 64.4 -> 73.4 | Improved |
| train on HumanML3D++ scene texts instead of zero-shot | MLD · HumanML3D++ Scene Text | FID: 0.897 -> 9.408 | Hurt |
| use Gemini instead of GPT-4o for scene text generation | LLM scene text generation · Human eval of 100 scene texts | Excellent rating: 25.9% -> 23.5% | No clear effect |
| use index length 7 instead of shorter or longer (ablation) | TAAT ACT model · HumanML3D++ | generation quality | Mixed |
Closest papers by meaning
- Text Motion Translator: A Bi-Directional Model for Enhanced 3D Human Motion Generation from Open-Vocabulary Descriptions (ECCV 2024 · similarity 0.91)
- InterDreamer: Zero-Shot Text to 3D Dynamic Human-Object Interaction (NeurIPS 2024 · similarity 0.90)
- Generating Human Motion in 3D Scenes from Text Descriptions (CVPR 2024 · similarity 0.89)
- The Language of Motion: Unifying Verbal and Non-verbal Language of 3D Human Motion (CVPR 2025 · similarity 0.89)
- SnapMoGen: Human Motion Generation from Expressive Texts (NeurIPS 2025 · similarity 0.89)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.89)
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.
