TY  - RPRT
TI  - Do Vision Models Learn Physical Constraints or Rendering Shortcuts? A Counterfactual Benchmark for Grounded Physical Consistency
AU  - M. Moein Esfahani
AU  - Sepehr Salem
AU  - Mohammed Alser
AU  - Vince Calhoun
PY  - 2026
UR  - https://arxiv.org/abs/2610.09205
ID  - 2610.09205
ER  -