Research map / Architectures and efficiency / Attention and transformers
Motion Transformer for Unsupervised Image Animation
Jiale Tao, Biao Wang, Tiezheng Ge, Yuning Jiang, Wen Li, Lixin Duan
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 position encoding to image tokens (ablation) | Motion Transformer · TaiChiHD | AKD: 5.482 -> 4.670 | Improved |
| add position encoding to image tokens (ablation) | Motion Transformer · TaiChiHD | L1: 0.047 -> 0.045 | Improved |
| add position encoding to image tokens (ablation) | Motion Transformer · TaiChiHD | AED: 0.158 -> 0.148 | Improved |
| use HRNet-W32 instead of Stem CNN (ablation) | Motion Transformer · TaiChiHD | AKD: 6.056 -> 4.670 | Improved |
| use HRNet-W32 instead of Stem CNN (ablation) | Motion Transformer · TaiChiHD | L1: 0.048 -> 0.045 | Improved |
| use heavier HRNet-W48 instead of HRNet-W32 (ablation) | Motion Transformer · TaiChiHD | AKD: 4.670 -> 4.829 | No clear effect |
| use 12 transformer layers instead of 4 (ablation) | Motion Transformer · TaiChiHD | AKD: 5.320 -> 4.670 | Improved |
| use 12 transformer layers instead of 8 (ablation) | Motion Transformer · TaiChiHD | MKR: 0.025 -> 0.021 | Improved |
| use 12 transformer layers instead of 4 (ablation) | Motion Transformer · TaiChiHD | L1: 0.046 -> 0.045 | Improved |
Closest papers by meaning
- Animating Arbitrary Objects via Deep Motion Transfer (CVPR 2019 · similarity 0.89)
- MotionDreamer: One-to-Many Motion Synthesis with Localized Generative Masked Transformer (ICLR 2025 · similarity 0.87)
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (CVPR 2025 · similarity 0.87)
- Large Displacement Motion Transfer with Unsupervised Anytime Interpolation (ICML 2025 · similarity 0.86)
- ProMotion: Prototypes As Motion Learners (CVPR 2024 · similarity 0.86)
- MotionStone: Decoupled Motion Intensity Modulation with Diffusion Transformer for Image-to-Video Generation (CVPR 2025 · similarity 0.85)
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.
