用物体间相对运动关系测试视频模型的物理推理能力
Principia: Relational Physics Tests for Video Models
- 通过配对物体的运动一致性检验物理规律，无需相机标定
- 六款顶尖视频生成模型在该测试中最高得分仅0.42
- 适合评估生成内容是否符合真实物理规律的研究者
视频模型的物理推理能力评估困难，因绝对运动测量依赖帧率、物体尺度和相机标定，而这些信息在生成视频中常不明确或缺失。我们提出新思路：若场景中两个物体遵循相同物理定律，其运动应满足可预测的关系，且该关系与标定无关。为此，我们构建了Principia基准，通过配对物体间的相对一致性评估牛顿力学，涵盖重力、弹性碰撞、摩擦、转动惯量、抛体运动、动量守恒、单摆及质量-弹簧振荡共八种现象，覆盖平动、转动、碰撞与振荡动态，基于受控协议采集的真实世界场景数据。我们还提出一种无标定的一致性评分，在图像空间直接量化物理违反程度。在六款先进视频生成模型的数千次生成结果上，所有模型得分均不超过0.42，尽管它们在VBench上得分均约0.8。视觉语言模型被用于检测这种关系性物理错误，最佳模型准确率仅67%，多数接近随机水平。
原文摘要 · Abstract (English)
Evaluating physical reasoning in video models is difficult because absolute motion measurements depend on frame rate, object scale, and camera calibration, all of which are often ambiguous or unavailable in generated video. We propose a different approach. When two objects in the same scene obey the same physical law, their motions must satisfy predictable relationships, and these relationships hold independent of calibration. We introduce Principia, a benchmark that evaluates Newtonian physics through relational consistency between paired objects. Principia spans eight phenomena - gravity, restitution, friction, rotational inertia, projectile motion, momentum, pendulum, and mass-spring oscillation - across translational, rotational, collisional, and oscillatory dynamics, using real-world scenes recorded under controlled protocols. We also introduce a calibration-independent consistency score that quantifies physical violation directly in image space. Across thousands of generations from six state-of-the-art video generators, no model exceeds 0.42 on Principia despite all scoring around 0.8 on VBench. Vision-language models are evaluated on their ability to detect relational physics violations, with the best model achieving only 67% accuracy and most performing near chance level.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。