Research map / Video / Video generation
InfiniDreamer: Arbitrarily Long Human Motion Generation via Segment Score Distillation
Wenjie Zhuo, Fan Ma, Hehe Fan
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 gradient masks (ablation) | InfiniDreamer · HumanML3D | R-precision: 0.643 -> 0.679 | Improved |
| add gradient masks (ablation) | InfiniDreamer · HumanML3D | FID: 0.64 -> 0.47 | Improved |
| add gradient masks (ablation) | InfiniDreamer · BABEL | R-precision: 0.522 -> 0.543 | Improved |
| add gradient masks (ablation) | InfiniDreamer · BABEL | transition FID: 2.37 -> 2.07 | Improved |
| add geometric losses (ablation) | InfiniDreamer · BABEL | FID: 1.09 -> 0.97 | Improved |
| add geometric losses (ablation) | InfiniDreamer · BABEL | transition FID: 2.15 -> 2.07 | Improved |
| use text selection strategy instead of single prompt 'transition' (ablation) | InfiniDreamer · HumanML3D | R-precision: 0.657 -> 0.679 | Improved |
| use text selection strategy instead of unconditional optimization (ablation) | InfiniDreamer · HumanML3D | transition FID: 2.15 -> 2.04 | Improved |
| increase window size W from 120 to 150 (P=30) (ablation) | InfiniDreamer · HumanML3D | R-precision: 0.679 -> 0.667 | Hurt |
| reduce window size W from 120 to 30 (P=30) (ablation) | InfiniDreamer · HumanML3D | transition FID: 2.04 -> 4.33 | Hurt |
| increase stride P to 160 (W=120) (ablation) | InfiniDreamer · HumanML3D | transition FID: 2.04 -> 7.47 | Hurt |
| increase stride P to 120 (W=120) (ablation) | InfiniDreamer · HumanML3D | MultiModal-Dist: 3.15 -> 3.25 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports InfiniDreamer as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Motion-R1: Enhancing Motion Generation with Decomposed Chain-of-Thought and RL Binding (ICLR 2026 · better on 2 of 2 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
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.89)
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.88)
- Scaling Large Motion Models with Million-Level Human Motions (ICML 2025 · similarity 0.88)
- MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation (AAAI 2024 · similarity 0.88)
- Seamless Human Motion Composition with Blended Positional Encodings (CVPR 2024 · similarity 0.88)
- MMM: Generative Masked Motion 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.
