Research map / Video / Video generation
Unified Number-Free Text-to-Motion Generation Via Flow Matching
Guanhe Huang, Oya Celiktutan
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 HumanML3D heterogeneous single-person prior (ablation) | UMF · HumanML3D | FID: 10.806 -> 0.486 | Improved |
| add HumanML3D heterogeneous single-person prior (ablation) | UMF · InterHuman | FID: 4.933 -> 4.772 | Improved |
| add latent adapter (ablation) | UMF · InterHuman | FID: 5.473 -> 4.772 | Improved |
| use multi-token instead of single-token latent (ablation) | UMF · InterHuman | FID: 5.231 -> 4.772 | Improved |
| use pyramid (temporal) flow instead of full-resolution flow (ablation) | P-Flow · InterHuman | FLOPs (G): 320.2 -> 140.3 | Improved |
| use temporal pyramid instead of spatial pyramid (ablation) | P-Flow · InterHuman | FID: 7.238 -> 4.772 | Improved |
| allocate asymmetric P-Flow steps (45/5) instead of symmetric (25/25) (ablation) | P-Flow · InterHuman | FLOPs (G): 206.0 -> 140.3 | Improved |
| use separate P-Flow and S-Flow transformers instead of shared (ablation) | UMF · InterHuman | FID: 6.206 -> 4.772 | Improved |
| add context reconstruction path (semi-noise) over noise-free path (ablation) | S-Flow · InterHuman | FID: 5.617 -> 4.772 | Improved |
| add context adapter instead of none or ControlNet (ablation) | S-Flow · InterHuman | FID: 7.038 -> 4.772 | Improved |
| add context reconstruction loss (ablation) | S-Flow · InterHuman | FID: 5.765 -> 4.772 | Improved |
| use UMF instead of specialist TIMotion | UMF vs TIMotion · InterHuman | R-Precision Top3: 0.724 -> 0.694 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 5 on most shared results.
- FreeMotion — FreeMotion: A Unified Framework for Number-free Text-to-Motion Synthesis (ECCV 2024) · better on 6 of 6 shared results
- TIMotion — TIMotion: Temporal and Interactive Framework for Efficient Human-Human Motion Generation (CVPR 2025) · better on 1 of 3 shared results
- ReGenNet — ReGenNet: Towards Human Action-Reaction Synthesis (CVPR 2024) · better on 2 of 3 shared results
- InterMask — InterMask: 3D Human Interaction Generation via Collaborative Masked Modeling (ICLR 2025) · better on 2 of 2 shared results
- InterGen — InterGen: Diffusion-Based Multi-human Motion Generation Under Complex Interactions (IJCV 2024) · 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
- Pyramidal Flow Matching for Efficient Video Generative Modeling (ICLR 2025 · similarity 0.87)
- Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregression (CVPR 2025 · similarity 0.86)
- Contextual Flow Matching: Adaptive Step Selection in Flow Models for Efficient Visual Generation (NeurIPS 2026 · similarity 0.86)
- Seamless Human Motion Composition with Blended Positional Encodings (CVPR 2024 · similarity 0.86)
- MMM: Generative Masked Motion Model (CVPR 2024 · similarity 0.86)
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · 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.
