Research map / Video / Video generation
Prompt-A-Video: Prompt Your Video Diffusion Model via Preference-Aligned LLM
Yatai Ji, Jiacheng Zhang, Jie Wu, Shilong Zhang, Shoufa Chen, Chongjian Ge, Peize Sun, Weifeng Chen, Wenqi Shao, Xuefeng Xiao, Weilin Huang, Ping Luo
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 DPO preference alignment after SFT (ablation) | Prompt-A-Video (CogVideoX) · WebVid in-domain | VideoScore VQ: 2.770 -> 2.900 | Improved |
| add DPO preference alignment after SFT (ablation) | Prompt-A-Video (Open-Sora 1.2) · WebVid in-domain | VideoScore VQ: 3.029 -> 3.090 | Improved |
| add a second DPO iteration (ablation) | Prompt-A-Video (Open-Sora 1.2) · WebVid in-domain | VideoScore VQ: 3.090 -> 3.254 | Improved |
| add a third DPO iteration (ablation) | Prompt-A-Video (CogVideoX) · WebVid in-domain | VideoScore DD: 3.183 -> 3.156 | Hurt |
| add a third DPO iteration (ablation) | Prompt-A-Video (Open-Sora 1.2) · WebVid in-domain | VideoScore DD: 3.411 -> 3.214 | Hurt |
| add VideoScore video-level reward to image-level rewards (ablation) | Prompt-A-Video DPO · VBench 100-prompt subset | VideoScore VQ: 2.660 -> 2.679 | Improved |
| add MPS reward (ablation) | Prompt-A-Video DPO · VBench 100-prompt subset | VBench image quality: 0.653 -> 0.662 | Improved |
| add Aesthetic Predictor reward (ablation) | Prompt-A-Video DPO · VBench 100-prompt subset | VBench aesthetic quality: 0.614 -> 0.627 | Improved |
| add DPO on top of SFT for text-to-image (ablation) | Prompt-A-Video · HPSv2 average | HPS score: 26.81 -> 27.09 | Improved |
| use text-to-image prompt booster Promptist for video prompts | Promptist · WebVid in-domain (CogVideoX) | VideoScore average: 2.989 -> 2.873 | Hurt |
| replace GLM-4 refinement with Prompt-A-Video | Prompt-A-Video · VBench (CogVideoX) | dynamic degree: 56.94 -> 65.28 | Improved |
| replace GPT-4o refinement with Prompt-A-Video | Prompt-A-Video · WebVid in-domain (Open-Sora 1.2) | VideoScore average: 3.079 -> 3.318 | Improved |
Later papers that compared with this method
2 later accepted papers report Prompt-A-Video as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Diverse Video Generation with Determinantal Point Process-Guided Policy Optimization (CVPR 2026 · better on 5 of 5 shared results)
- VPO: Aligning Text-to-Video Generation Models with Prompt Optimization (ICCV 2025 · better on 2 of 2 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
- VidProM: A Million-scale Real Prompt-Gallery Dataset for Text-to-Video Diffusion Models (NeurIPS 2024 · similarity 0.92)
- AP-Adapter: Improving Generalization of Automatic Prompts on Unseen Text-to-Image Diffusion Models (NeurIPS 2024 · similarity 0.91)
- PromptCoT: Align Prompt Distribution via Adapted Chain-of-Thought (CVPR 2024 · similarity 0.91)
- The Devil is in the Prompts: Retrieval-Augmented Prompt Optimization for Text-to-Video Generation (CVPR 2025 · similarity 0.90)
- Multimodal Prompt Optimization: Why Not Leverage Multiple Modalities for MLLMs (ICLR 2026 · similarity 0.89)
- Prompt-Free Diffusion: Taking "Text" out of Text-to-Image Diffusion Models (CVPR 2024 · 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.
