“…Our visual perception of an object is guided by its material properties; but, also involves factors such as geometry [VLD07, HFM16], light conditions [HLM06, KFB10, CK15] or motion [DFY * 11, MLMG19]. To reduce the dimensionality of the problem, previous work have focused on understanding single, high‐level appearance properties like glossiness [PFG00, WAKB09, CK15], translucency [GXZ * 13, XZG * 20, GWA * 15] and softness [SFV20, CDD21]; or draw inspiration from artists' implicit understanding of the key visual cues that guide visual perception [DCWP19, DSMG21]. Recent works suggest that material perception may be driven by complex non‐linear statistics better approximated by highly non‐linear models such as neural networks [FS19, SAF21, DLG * 20, LMS * 19].…”