Abstract
Colors influence our perceptual and aesthetic experiences, but their role in the appreciation of paintings is difficult to quantify. A previous experiment showed that observers strongly prefer the color compositions chosen by the artist over color-manipulated versions of the same painting (Nascimento et al, 2017, Vision Research, 130, 76-84). Their manipulation - a rigid rotation of the color gamut around the L* axis in CIELAB space – kept lightness and saturation invariant and therefore preserved the chromatic relationship between the colors. However, it changed perceptual appearance by changing the color spread over color categories. Therefore, we investigated how much this chromatic manipulation affects the distribution of colors over color categories and how it may determine preference scaling of naïve observers. We modeled how the distribution of colors across categories varied with the color-gamut rotation of paintings from various periods and artists. Using hyperspectral imaging data of the paintings, we derived the Euclidean distance in L*u*v* space between each pixel and the most typical color for each color category and assigned each pixel to the closest category. The degree of uniformity across color categories was quantified as the square root of the sum of squared differences between the actual pixel distribution and a uniform distribution, for each painting at each rotation angle. The majority of the paintings had the most homogenous distribution close the original color composition. Interestingly, the orientation of the most homogeneous color distribution was also the one generally preferred by participants (N=14) that performed pair-wise comparison on a subset of the images in a psychophysical study (Albers et al, 2017, ECVP). These results show that the color manipulated versions of the paintings tend to have less uniform color distribution across color categories and suggest that observers may rely on color category diversity in their preference judgments.
Meeting abstract presented at VSS 2018