Research map / Video / Video generation
MoVideo: Motion-Aware Video Generation with Diffusion Models
Jingyun Liang, Yuchen Fan, Kai Zhang, Radu Timofte, Luc Van Gool, Rakesh Ranjan
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 flow-warped latent video as condition (ablation) | MoVideo · DAVIS image-to-video | FVD: 796.82 -> 335.53 | Improved |
| add occlusion mask as condition (ablation) | MoVideo · DAVIS image-to-video | FID: 45.61 -> 40.14 | Improved |
| replace 2D decoder with flow-augmented 3D decoder (ablation) | video decoding · DAVIS decoding | FVD: 401.66 -> 335.53 | Improved |
| upgrade 2D decoder to 3D decoder (ablation) | video decoding · DAVIS decoding | PSNR: PSNR 30.11 -> 29.99, FVD 401.66 -> 379.73 | Mixed |
| add flow augmentation to decoder (PSNR) (ablation) | video decoding · DAVIS decoding | PSNR: 29.99 -> 29.88 | Hurt |
| use MoVideo instead of Gen1 for image-to-video | image-to-video · DAVIS image-to-video | FVD: 973.71 -> 335.53 | Improved |
| use MoVideo instead of Make-A-Video for zero-shot text-to-video | text-to-video · UCF-101 | FVD: 367.23 -> 313.41 | Improved |
| randomly mask the warped latent video with probability 0.5 in training | MoVideo · latent video generation | motion quality | 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.
- Make-A-Video — Make-A-Video: Text-to-Video Generation without Text-Video Data (ICLR 2023) · better on 3 of 3 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
- Video Diffusion Models are Training-free Motion Interpreter and Controller (NeurIPS 2024 · similarity 0.90)
- Make Pixels Dance: High-Dynamic Video Generation (CVPR 2024 · similarity 0.89)
- Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition (ICLR 2024 · similarity 0.89)
- Decouple Content and Motion for Conditional Image-to-Video Generation (AAAI 2024 · similarity 0.89)
- Optical-Flow Guided Prompt Optimization for Coherent Video Generation (CVPR 2025 · similarity 0.89)
- CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer (ICLR 2025 · similarity 0.89)
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.
