Towards World Simulator: Crafting Physical Commonsense-Based Benchmark for Video Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Meng, Fanqing, Liao, Jiaqi, Tan, Xinyu, Shao, Wenqi, Lu, Quanfeng, Zhang, Kaipeng, Cheng, Yu, Li, Dianqi, Qiao, Yu, Luo, Ping
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912062550048768
author Meng, Fanqing
Liao, Jiaqi
Tan, Xinyu
Shao, Wenqi
Lu, Quanfeng
Zhang, Kaipeng
Cheng, Yu
Li, Dianqi
Qiao, Yu
Luo, Ping
author_facet Meng, Fanqing
Liao, Jiaqi
Tan, Xinyu
Shao, Wenqi
Lu, Quanfeng
Zhang, Kaipeng
Cheng, Yu
Li, Dianqi
Qiao, Yu
Luo, Ping
contents Text-to-video (T2V) models like Sora have made significant strides in visualizing complex prompts, which is increasingly viewed as a promising path towards constructing the universal world simulator. Cognitive psychologists believe that the foundation for achieving this goal is the ability to understand intuitive physics. However, the capacity of these models to accurately represent intuitive physics remains largely unexplored. To bridge this gap, we introduce PhyGenBench, a comprehensive \textbf{Phy}sics \textbf{Gen}eration \textbf{Ben}chmark designed to evaluate physical commonsense correctness in T2V generation. PhyGenBench comprises 160 carefully crafted prompts across 27 distinct physical laws, spanning four fundamental domains, which could comprehensively assesses models' understanding of physical commonsense. Alongside PhyGenBench, we propose a novel evaluation framework called PhyGenEval. This framework employs a hierarchical evaluation structure utilizing appropriate advanced vision-language models and large language models to assess physical commonsense. Through PhyGenBench and PhyGenEval, we can conduct large-scale automated assessments of T2V models' understanding of physical commonsense, which align closely with human feedback. Our evaluation results and in-depth analysis demonstrate that current models struggle to generate videos that comply with physical commonsense. Moreover, simply scaling up models or employing prompt engineering techniques is insufficient to fully address the challenges presented by PhyGenBench (e.g., dynamic scenarios). We hope this study will inspire the community to prioritize the learning of physical commonsense in these models beyond entertainment applications. We will release the data and codes at https://github.com/OpenGVLab/PhyGenBench
format Preprint
id arxiv_https___arxiv_org_abs_2410_05363
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Towards World Simulator: Crafting Physical Commonsense-Based Benchmark for Video Generation
Meng, Fanqing
Liao, Jiaqi
Tan, Xinyu
Shao, Wenqi
Lu, Quanfeng
Zhang, Kaipeng
Cheng, Yu
Li, Dianqi
Qiao, Yu
Luo, Ping
Computer Vision and Pattern Recognition
Text-to-video (T2V) models like Sora have made significant strides in visualizing complex prompts, which is increasingly viewed as a promising path towards constructing the universal world simulator. Cognitive psychologists believe that the foundation for achieving this goal is the ability to understand intuitive physics. However, the capacity of these models to accurately represent intuitive physics remains largely unexplored. To bridge this gap, we introduce PhyGenBench, a comprehensive \textbf{Phy}sics \textbf{Gen}eration \textbf{Ben}chmark designed to evaluate physical commonsense correctness in T2V generation. PhyGenBench comprises 160 carefully crafted prompts across 27 distinct physical laws, spanning four fundamental domains, which could comprehensively assesses models' understanding of physical commonsense. Alongside PhyGenBench, we propose a novel evaluation framework called PhyGenEval. This framework employs a hierarchical evaluation structure utilizing appropriate advanced vision-language models and large language models to assess physical commonsense. Through PhyGenBench and PhyGenEval, we can conduct large-scale automated assessments of T2V models' understanding of physical commonsense, which align closely with human feedback. Our evaluation results and in-depth analysis demonstrate that current models struggle to generate videos that comply with physical commonsense. Moreover, simply scaling up models or employing prompt engineering techniques is insufficient to fully address the challenges presented by PhyGenBench (e.g., dynamic scenarios). We hope this study will inspire the community to prioritize the learning of physical commonsense in these models beyond entertainment applications. We will release the data and codes at https://github.com/OpenGVLab/PhyGenBench
title Towards World Simulator: Crafting Physical Commonsense-Based Benchmark for Video Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.05363