Bootstrapping Physics-Grounded Video Generation through VLM-Guided Iterative Self-Refinement
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909924325326848 |
|---|---|
| author | Liu, Yang Zhao, Xilin Wen, Peisong Dai, Siran Huang, Qingming |
| author_facet | Liu, Yang Zhao, Xilin Wen, Peisong Dai, Siran Huang, Qingming |
| contents | Recent progress in video generation has led to impressive visual quality, yet current models still struggle to produce results that align with real-world physical principles. To this end, we propose an iterative self-refinement framework that leverages large language models and vision-language models to provide physics-aware guidance for video generation. Specifically, we introduce a multimodal chain-of-thought (MM-CoT) process that refines prompts based on feedback from physical inconsistencies, progressively enhancing generation quality. This method is training-free and plug-and-play, making it readily applicable to a wide range of video generation models. Experiments on the PhyIQ benchmark show that our method improves the Physics-IQ score from 56.31 to 62.38. We hope this work serves as a preliminary exploration of physics-consistent video generation and may offer insights for future research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_20280 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Bootstrapping Physics-Grounded Video Generation through VLM-Guided Iterative Self-Refinement Liu, Yang Zhao, Xilin Wen, Peisong Dai, Siran Huang, Qingming Computer Vision and Pattern Recognition Recent progress in video generation has led to impressive visual quality, yet current models still struggle to produce results that align with real-world physical principles. To this end, we propose an iterative self-refinement framework that leverages large language models and vision-language models to provide physics-aware guidance for video generation. Specifically, we introduce a multimodal chain-of-thought (MM-CoT) process that refines prompts based on feedback from physical inconsistencies, progressively enhancing generation quality. This method is training-free and plug-and-play, making it readily applicable to a wide range of video generation models. Experiments on the PhyIQ benchmark show that our method improves the Physics-IQ score from 56.31 to 62.38. We hope this work serves as a preliminary exploration of physics-consistent video generation and may offer insights for future research. |
| title | Bootstrapping Physics-Grounded Video Generation through VLM-Guided Iterative Self-Refinement |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2511.20280 |