Abstract
The color rendering of lighting is typically evaluated by how natural the objects are perceived. The influence of the illumination on the aesthetics of the scenes is another important aspect to consider. However, predicting these aspects from existing quality indices for specific applications is difficult. We estimated, psychophysically, the best lighting for naturalness and preference for food displays in commercial environments. Fifteen images of commercial food counters of fruits, vegetables, meat, and fish were digitalized by a hyperspectral system in a supermarket. The scenes were simulated to be illuminated by illuminants synthesized from Judd’s daylight spectral basis functions for a grid of chromaticities on and around the Planckian locus with correlated color temperatures (CCT) ranging 2,222 – 20,000 K and were displayed on an LCD monitor controlled by a ViSaGe visual stimulus generator. Observers adjusted the chromaticity of the illuminant under two criteria: so that the scenes looked the most natural or the most pleasant. The mean CCT averaged over all the observers and images for the preference criterion was 5900 K and for the naturalness criterion was 7100 K. This type of pattern was consistently found for each scene category. In addition, it was found that the most preferred CCTs for fish and meat were considerably lower than those for fruits and vegetables. The average chromaticities for both criteria were below the locus that maximized the general color rendering index (CRI). The convex hull volume of the colors in the CIELAB space of the each scene illuminated by the most preferred illuminant was consistently larger than that for the most natural illuminant, indicating that the observers preferred lighting rendering more saturated colors. These results show that preference and naturalness require different illuminations and that the ideal spectrum may have to be tuned for different food classes.
Meeting abstract presented at VSS 2012