Research map / Video / Video generation
InfinityHuman: Towards Long-Term Audio-Driven Human Animation
Xiaodi Li, Pan Xie, Yi Ren, Qijun Gan, Chen Zhang, Fangyuan Kong, Xiang Yin, Zehuan Yuan, Bingyue Peng
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 the pose-guided refiner (ablation) | InfinityHuman · EMTD subset | FID / FSIM: FID 109.54 -> 91.74; FSIM 0.79 -> 0.88 | Improved |
| add pose guidance condition to the refiner (ablation) | PG-Refiner · EMTD subset | FID / FVD: FID 156.74 -> 91.74; FVD 1163.75 -> 758.98 | Improved |
| add low-resolution latent condition to the refiner (ablation) | PG-Refiner · EMTD subset | FVD / FSIM: FVD 1001.00 -> 758.98; FSIM 0.86 -> 0.88 | Improved |
| add low-resolution latent condition to the refiner (ablation) | PG-Refiner · EMTD subset | FID: 91.92 -> 91.74 | No clear effect |
| add hand-specific reward feedback learning (ablation) | InfinityHuman · EMTD subset | HKC: 0.85 -> 0.87 | Improved |
| add hand-specific reward feedback learning (ablation) | InfinityHuman · EMTD subset | FID / FVD: FID 86.32 -> 91.74 (worse); FVD 844.57 -> 758.98 (better) | Mixed |
| use InfinityHuman instead of strongest baselines | long-video audio-driven animation · EMTD | FID / FVD: FID 82.54 -> 60.71; FVD 1040.43 -> 979.88 | Improved |
| use InfinityHuman instead of baselines | lip-sync on HDTF · HDTF | Sync-C / FaceSIM: Sync-C 8.59 vs SadTalker 8.87; FaceSIM 0.89 vs 0.93 | No clear effect |
| generate longer videos (10s to 50s) | InfinityHuman · EMTD subset (long-form) | FID / FSIM: FID 36.83 -> 35.50; FSIM 0.8357 -> 0.8057 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Hallo3 — Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Video Diffusion Transformer (CVPR 2025) · better on 6 of 6 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
- HumanVid: Demystifying Training Data for Camera-controllable Human Image Animation (NeurIPS 2024 · similarity 0.88)
- EchoMimicV2: Towards Striking, Simplified, and Semi-Body Human Animation (CVPR 2025 · similarity 0.88)
- KeyFace: Expressive Audio-Driven Facial Animation for Long Sequences via KeyFrame Interpolation (CVPR 2025 · similarity 0.87)
- AudCast: Audio-Driven Human Video Generation by Cascaded Diffusion Transformers (CVPR 2025 · similarity 0.87)
- OpenHumanVid: A Large-Scale High-Quality Dataset for Enhancing Human-Centric Video Generation (CVPR 2025 · similarity 0.87)
- Teller: Real-Time Streaming Audio-Driven Portrait Animation with Autoregressive Motion Generation (CVPR 2025 · 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.
