TY  - RPRT
TI  - Anchored, Not Graded: Vision-Language Models Fail at Slant-from-Texture Perception
AU  - Qian Zhang
AU  - Michal Golovanevsky
AU  - Fulvio Domini
AU  - James Tompkin
PY  - 2026
UR  - https://arxiv.org/abs/2606.06714
ID  - 2606.06714
ER  -