Research map / Video / Video generation
MultiAnimate: Pose-Guided Image Animation Made Extensible
Yingcheng Hu, Haowen Gong, Chuanguang Yang, Zhulin An, Yongjun Xu, Songhua Liu
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 |
|---|---|---|---|
| use mask-driven Identifier Assigner and Adapter over UniAnimate-DiT | MultiAnimate · Swing Dance | PSNR: 16.15 -> 19.40 | Improved |
| use mask-driven Identifier Assigner and Adapter over UniAnimate-DiT | MultiAnimate · Swing Dance | FVD: 891.89 -> 648.84 | Improved |
| use Extended model (up to 7 characters) over UniAnimate-DiT | MultiAnimate · Unseen dance videos | FVD: 624.45 -> 358.74 | Improved |
| use Extended model over UniAnimate-DiT | MultiAnimate · Unseen dance videos | PSNR: 17.94 -> 23.24 | Improved |
| use Stage 1 model over UniAnimate-DiT without training on Gen-dataset | MultiAnimate · Gen-dataset | FID-VID: 32.10 -> 33.57 | Hurt |
| use Stage 1 model over VACE | MultiAnimate · Gen-dataset | LPIPS: 0.231 -> 0.251 | Hurt |
| add Gen-dataset training (Stage 2 model) (ablation) | MultiAnimate · Gen-dataset samples | subject and background consistency (qualitative) | Improved |
| use mask-driven design instead of addition-driven per-person feature summation (ablation) | MultiAnimate · more characters than training | generalization to more characters (qualitative) | Improved |
| add Identifier Assigner and Adapter (mask-driven design) (ablation) | MultiAnimate · Swing Dance / unseen videos | identity consistency (qualitative) | Improved |
| compare to single-character baselines | MultiAnimate · TikTok (single character) | PSNR: 17.76 -> 23.68 | Improved |
| compare to single-character baselines | MultiAnimate · TikTok (single character) | FVD: 615.27 -> 342.48 | Improved |
| randomly assign identifiers from a label bank during training (ablation) | MultiAnimate · Swing Dance | symmetry problem (qualitative) | Improved |
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.
- Mimicmotion — MimicMotion: High-Quality Human Motion Video Generation with Confidence-aware Pose Guidance (ICML 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
- Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animation (CVPR 2024 · similarity 0.91)
- EverybodyDance: Bipartite Graph–Based Identity Correspondence for Multi-Character Animation (NeurIPS 2025 · similarity 0.90)
- Free-viewpoint Human Animation with Pose-correlated Reference Selection (CVPR 2025 · similarity 0.90)
- ReMask-Animate: Refined Character Image Animation Using Mask-Guided Adapters (AAAI 2025 · similarity 0.89)
- MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Model (CVPR 2024 · similarity 0.89)
- Towards Multiple Character Image Animation Through Enhancing Implicit Decoupling (ICLR 2025 · 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.
