Research map / Video / Video generation
Perception-as-Control: Fine-grained Controllable Image Animation with 3D-aware Motion Representation
Yingjie Chen, Yifang Men, Yuan Yao, Miaomiao Cui, Liefeng Bo
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 3D unit spheres with depth-scaled radius instead of 2D point positions (ablation) | Perception-as-Control · WebVid test set | ObjMC: 30.61 -> 29.47 | Improved |
| add world envelope to 3D unit spheres (ablation) | Perception-as-Control · WebVid test set | FVD: 176.989 -> 161.076 | Improved |
| add world envelope to 3D unit spheres (ablation) | Perception-as-Control · WebVid test set | ObjMC: 29.47 -> 23.32 | Improved |
| use 3D-aware spheres plus envelope instead of 2D point offset (ablation) | Perception-as-Control · WebVid test set | FVD: 254.907 -> 161.076 | Improved |
| use 3D-aware spheres plus envelope instead of 2D point offset (ablation) | Perception-as-Control · WebVid test set | ObjMC: 53.28 -> 23.32 | Improved |
| add world envelope to 3D unit spheres (ablation) | Perception-as-Control · WebVid test set | FID: 16.458 -> 16.465 | No clear effect |
| use three-stage training instead of one-stage (ablation) | Perception-as-Control · WebVid test set | FVD: 318.941 -> 161.076 | Improved |
| use three-stage training instead of two-stage (ablation) | Perception-as-Control · WebVid test set | FVD: 254.907 -> 161.076 | Improved |
| use three-stage training instead of one-stage (ablation) | Perception-as-Control · WebVid test set | FID: 59.040 -> 16.465 | Improved |
| incorporate 3D-aware motion representation into MotionCtrl (MotionCtrl++) | MotionCtrl · RealEstate-10K test set | FVD: 122.570 -> 104.020 | Improved |
| incorporate 3D-aware motion representation into CameraCtrl (CameraCtrl*++) | CameraCtrl* (16-frame) · RealEstate-10K test set | FVD: 136.602 -> 113.165 | Improved |
| use full framework instead of CameraCtrl*++ built on SVD | CameraCtrl*++ · RealEstate-10K test set | FVD: 113.165 -> 52.421 | Improved |
Closest papers by meaning
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (CVPR 2025 · similarity 0.88)
- Follow-Your-Click: Open-domain Regional Image Animation via Motion Prompts (AAAI 2025 · similarity 0.88)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.87)
- Boximator: Generating Rich and Controllable Motions for Video Synthesis (ICML 2024 · similarity 0.87)
- TokenMotion: Decoupled Motion Control via Token Disentanglement for Human-centric Video Generation (CVPR 2025 · similarity 0.86)
- Consistent and Controllable Image Animation with Motion Diffusion Models (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.
