Research map / Video / Video generation
MotionAgent: Fine-grained Controllable Video Generation via Motion Field Agent
Xinyao Liao, Xianfang Zeng, Liao Wang, Gang Yu, Guosheng Lin, Chi Zhang
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 Grounded-SAM detection tool for object identification (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Object Movement Q&A: 34.33 -> 45.69 | Improved |
| add object movement control (trajectory plotting) (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Object Movement Q&A: 10.51 -> 45.69 | Improved |
| add camera motion control (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Complex Camera Motion: 0.30 -> 77.76 | Improved |
| add analytical optical flow composition (vs directly adding flows) (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Complex Camera Motion: 64.92 -> 77.76 | Improved |
| add analytical optical flow composition (vs directly adding flows) (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Object Movement Q&A: 30.07 -> 45.69 | Improved |
| fine-tune optical flow adapter on unified flow maps (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Complex Camera Motion: 48.27 -> 77.76 | Improved |
| fine-tune optical flow adapter on unified flow maps (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Object Movement Q&A: 40.56 -> 45.69 | Improved |
| use grid selection instead of direct offset generation for trajectories (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Object Movement Q&A: 29.28 -> 45.69 | Improved |
| use open-source Llama3 instead of GPT-4o as agent LLM (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Total score: 65.10 -> 63.80 | No clear effect |
| use open-source Qwen2 instead of GPT-4o as agent LLM (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Total score: 65.10 -> 61.73 | No clear effect |
| use Motion-I2V instead of SVD as base I2V model (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Total score: 65.10 -> 61.00 | No clear effect |
| add rethinking step (ablation) | MotionAgent · Controllable I2V benchmark (VBench subset) | Complex Camera Motion: 77.76 -> 89.04 | 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.
- DynamiCrafter — DynamiCrafter: Animating Open-domain Images with Video Diffusion Priors (ECCV 2024) · better on 4 of 5 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
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.89)
- Make Pixels Dance: High-Dynamic Video Generation (CVPR 2024 · similarity 0.87)
- Motion Prompting: Controlling Video Generation with Motion Trajectories (CVPR 2025 · similarity 0.87)
- Optical-Flow Guided Prompt Optimization for Coherent Video Generation (CVPR 2025 · similarity 0.87)
- MotionClone: Training-Free Motion Cloning for Controllable Video Generation (ICLR 2025 · similarity 0.87)
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (CVPR 2025 · similarity 0.87)
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.
