Abstract
The past decade has seen a surge in studies investigating how we perceive material properties of objects, with the majority focusing on the perception of gloss. This research has shown that how glossy an object or surface looks is determined by the appearance of specular highlights (their shape, contrast, sharpness, size, alignment), which in turn is determined by complex interactions between physical factors (surface properties, shape, illumination), and viewpoint. Although previous research acknowledges that highlight properties and appearance are multidimensional, very few studies have explored the dimensionality of gloss perception, that is, that a change in physical properties or appearance of the surface may not just affect the level of gloss but the quality/material of the surface altogether. We addressed this directly by investigating how properties of specular highlights affect perceived material category. We rendered complex glossy objects under natural illumination fields, manipulating five gloss parameters (specular strength, roughness, anisotropy, anisotropic rotation, specular tint), and two diffuse shading parameters (saturation, lightness). Observers judged what material each object was made from, with no restrictions. After processing for duplicates and similar terminology, over 200 words were used to describe the materials. Cluster analysis revealed that the stimuli fell into strikingly distinct perceptual categories such as plastic, glazed porcelain/ceramic, metals (e.g. gold, bronze, chrome), fabric (e.g. velvet, silk), wax/soap, and even chocolate, yet there was no simple mapping of rendering parameters onto categories. We implemented an image-based model of gloss appearance that predicted perceived material category and generalised to different objects and lighting conditions. Our results demonstrate that the appearance of specular highlights yield a much richer experience of glossy objects than suggested by previous studies. These findings are in line with the idea that the perception and neural processing of qualities like glossiness should be considered in the context of material recognition.
Acknowledgement: This research was supported by a Sofja Kovalevskaja Award from the Alexander von Humboldt Foundation, sponsored by the German Federal Ministry for Education and Research