Abstract
The visual appearance of fabrics allows us to infer their physical properties, like a sweater feeling warm and soft. Moreover, we usually associate specific material attributes with certain materials. We performed two online experiments to find which perceptual attributes distinguish velvet from satin depicted in 17th century paintings, and how perception changes if textiles’ context is removed. In experiment 1, six attributes (roughness, shininess, weight, softness, hairiness, warmth) were rated for 20 images of painted velvet and satin, shown either with context in full figure or without context as crops of the textile. Overall, all attributes were consistently associated with velvet, except shininess which was associated with satin. PCA showed that satin and velvet in full figure condition clustered according to their distinctive material attributes. The distinction was less clear in the cropped condition. In experiment 2 we tested the influence of the choice of the cropped area on this effect by dividing the textiles into equally sized crops, resulting in around 15 crops per painting. Participants rated these crops on either softness or shininess, attributes found to be diagnostic and independent in experiment 1. A one-way ANOVA showed that crops differed significantly in shininess, but not softness perception. For each crop, we analyzed the highlights’ features. Overall, shininess correlated positively to highlights’ intensity and percentage of coverage, whereas softness was not correlated to the highlights’ features. Other cues, like asperity scattering and folds’ shape, might be related to visual perception of softness. We showed that velvet and satin possess perceptual signatures which can be used to distinguish them. We also demonstrated that if parts of a whole are isolated, their perception differ due to the local availability of cues. Further research is needed to understand if the global judgement derives from the most representative part or from an average.