Research map / Video / Video generation
AnyI2V: Animating Any Conditional Image with Motion Control
Ziye Li, Hao Luo, Xincheng Shuai, Henghui Ding
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 K&V consistency across frames (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | FVD: 587.69 -> 569.89 | Improved |
| add K&V consistency across frames (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | ObjMC: 16.81 -> 16.39 | Improved |
| add PCA reduction of query features (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | ObjMC: 17.14 -> 16.39 | Improved |
| add PCA reduction of query features (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | FVD: 585.04 -> 569.89 | Improved |
| add semantic mask instead of static mask (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | FVD: 598.15 -> 569.89 | Improved |
| add semantic mask instead of static mask (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | ObjMC: 16.92 -> 16.39 | Improved |
| add semantic mask (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | ObjMC: 17.62 -> 16.39 | Improved |
| align query instead of residual hidden for latent optimization (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | ObjMC: 36.23 -> 16.39 | Improved |
| align query instead of residual hidden for latent optimization (ablation) | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | FID: 129.40 -> 104.53 | Improved |
| use AnyI2V training-free control instead of AnimateDiff + SparseCtrl baseline | AnimateDiff · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | ObjMC: 38.26 -> 16.39 | Improved |
| use AnyI2V instead of best training-free ObjCtrl-2.5D | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | FID: 111.82 -> 104.53 | Improved |
| compare with trained DragAnything | AnyI2V · Trajectory-controlled I2V (web + VIPSeg, structural condition first frame) | ObjMC: 13.60 vs 16.39 | Hurt |
Closest papers by meaning
- Decouple Content and Motion for Conditional Image-to-Video Generation (AAAI 2024 · similarity 0.89)
- Extrapolating and Decoupling Image-to-Video Generation Models: Motion Modeling is Easier Than You Think (CVPR 2025 · similarity 0.89)
- ConditionVideo: Training-Free Condition-Guided Video Generation (AAAI 2024 · similarity 0.89)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.89)
- Enhancing Motion in Text-to-Video Generation with Decomposed Encoding and Conditioning (NeurIPS 2024 · similarity 0.89)
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (CVPR 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.
