← 研踪 PaperTrailarXiv:2610.31429 · v2
PhysTrace: Interpretable Temporal Evaluation of Physical Consistency in Video Generative Models
首次提交 2026-10-22 23:27
版本更新 2026-11-05 23:42
雷达观测 2026-11-06 12:09
摘要
World action models increasingly rely on pretrained diffusion/flow-based video generators assumed to capture physical dynamics for prediction, planning, and action generation. Existing evaluations assess plausibility, probe representations, or measure predictive surprise, yet physical consistency is history-dependent: preceding events constrain what can physically occur later. We introduce PhysTrace, a training-free framework that uses MLLMs to formulate history-conditioned physical expectations and tracks their agreement with observed video windows through temporal video--language alignment, rather than relying on direct plausibility judgments or task-specific probes. On IntPhys2, PhysTrace achieves 67.2% possible--impossible discrimination accuracy, outperforming the evaluated direct MLLM baselines. We further introduce reference-guided masked denoising for controlled reconstruction from mixed possible--impossible video latents. To support these reconstructions and evaluate temporal localization as an additional capability, we introduce IntPhys2-Timing, adding violation intervals to IntPhys2, with PhysTrace achieving a Hit Rate of up to 77.4%. On these reconstructions, human evaluations reveal model-dependent differences in physical consistency and fewer physically impossible outcomes at higher initial noise levels, trends broadly captured by PhysTrace.
作者备注：19 pages, 8 figures, 7 tables, including supplementary material
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-05 23:42 · 最新观测版本
PhysTrace: Interpretable Temporal Evaluation of Physical Consistency in Video Generative Models
World action models increasingly rely on pretrained diffusion/flow-based video generators assumed to capture physical dynamics for prediction, planning, and action generation. Existing evaluations assess plausibility, probe representations, or measure predictive surprise, yet physical consistency is history-dependent: preceding events constrain what can physically occur later. We introduce PhysTrace, a training-free framework that uses MLLMs to formulate history-conditioned physical expectations and tracks their agreement with observed video windows through temporal video--language alignment, rather than relying on direct plausibility judgments or task-specific probes. On IntPhys2, PhysTrace achieves 67.2% possible--impossible discrimination accuracy, outperforming the evaluated direct MLLM baselines. We further introduce reference-guided masked denoising for controlled reconstruction from mixed possible--impossible video latents. To support these reconstructions and evaluate temporal localization as an additional capability, we introduce IntPhys2-Timing, adding violation intervals to IntPhys2, with PhysTrace achieving a Hit Rate of up to 77.4%. On these reconstructions, human evaluations reveal model-dependent differences in physical consistency and fewer physically impossible outcomes at higher initial noise levels, trends broadly captured by PhysTrace.
查看此版本解读 →查看此版本原文 ↗