Research map / Video / Video generation
Free-Form Motion Control: Controlling the 6D Poses of Camera and Objects in Video Generation
Xincheng Shuai, Henghui Ding, Zhenyuan Qin, Hao Luo, Xingjun Ma, Dacheng Tao
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 camera loss L_cam (background-masked diffusion loss) (ablation) | FMC · SynFMC ablation | CamTransErr: 20.35 -> 18.12 | Improved |
| add camera loss L_cam (ablation) | FMC · SynFMC ablation | CamRotErr: 1.19 -> 1.03 | Improved |
| add object loss L_obj (foreground-masked diffusion loss) (ablation) | FMC · SynFMC ablation | ObjTransErr: 46.62 -> 42.25 | Improved |
| add object loss L_obj (ablation) | FMC · SynFMC ablation | ObjRotErr: 1.15 -> 0.96 | Improved |
| use camera poses when training the object module (ablation) | MotionCtrl trained on SynFMC · SynFMC ablation | ObjTransErr: 78.82 -> 55.33 | Improved |
| use camera poses when training the object module (ablation) | MotionCtrl trained on SynFMC · SynFMC ablation | ObjRotErr: 1.65 -> 1.26 | Improved |
| replace MotionCtrl object module with 6D-pose OMC (FMC) | MotionCtrl · SynFMC evaluation | ObjTransErr: 80.66 -> 42.25 | Improved |
| replace MotionCtrl with FMC | MotionCtrl · SynFMC evaluation | ObjRotErr: 1.77 -> 0.96 | Improved |
| use FMC instead of MotionCtrl for image quality | MotionCtrl · SynFMC evaluation | FID: 125.52 -> 133.42 | Hurt |
| use FMC instead of CameraCtrl for camera rotation | CameraCtrl · SynFMC evaluation | CamRotErr: 0.94 -> 1.03 | Hurt |
| use FMC instead of MotionCtrl for text similarity | MotionCtrl · SynFMC evaluation | CLIPSIM: 26.83 -> 31.01 | Improved |
| use FMC instead of MotionCtrl for object motion | MotionCtrl · User study | Object Motion Score: 0.53 -> 0.98 | Improved |
Closest papers by meaning
- PoseTraj: Pose-Aware Trajectory Control in Video Diffusion (CVPR 2025 · similarity 0.87)
- AC3D: Analyzing and Improving 3D Camera Control in Video Diffusion Transformers (CVPR 2025 · similarity 0.86)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.86)
- 6D Object Pose Tracking in Internet Videos for Robotic Manipulation (ICLR 2025 · similarity 0.86)
- MotionBooth: Motion-Aware Customized Text-to-Video Generation (NeurIPS 2024 · similarity 0.86)
- 3DTrajMaster: Mastering 3D Trajectory for Multi-Entity Motion in Video Generation (ICLR 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.
