Abstract
Purpose. To judge object surface properties, the visual system must estimate reflectance from the light signal arriving at the eyes. We ask to what extent observers are able to do this under geometrically varying light fields, and focus on the interaction between two distinct reflectance properties: diffuse and specular. These are physical correlates of the percepts of lightness and glossiness. If the two reflectance attributes are processed independently, future experiments can be simplified by studying each in isolation, while interactions require continued joint measurement. Methods. Observers adjusted the diffuse and specular components of one grayscale sphere to match the appearance of second, reference, sphere. Spheres were rendered using the Debevec (SIGGRAPH98) light fields and presented on a high-dynamic-range display. For symmetric matches, both spheres were rendered using the same light field. For asymmetric matches, a different light field was chosen for each sphere. Matches were collected for different combinations of reference sphere diffuse and specular reflectance. Surface roughness was held constant across the two spheres; the measurements were repeated for two levels of roughness. Performance was quantified by the slope of regression lines of matched versus reference reflectance. Results. Symmetric matches were close to veridical (average slopes 0.99 diffuse component; 1.00 specular). Asymmetric matches deviated systematically from veridical (average magnitude of slope deviation 0.06 diffuse; 0.31 specular), showing an effect of light field on perceived lightness and glossiness. The matched diffuse component decreased with increasing reference sphere specular component. In contrast, the matched specular component was roughly independent of reference sphere diffuse component. Matches were similar for the two levels of roughness (r = 0.94). Conclusions. The spatial structure of the illumination affects the perceived lightness and glossiness of 3D objects. Specular component matches were independent of the diffuse component, but not vice-versa. Changing surface roughness had little effect.
This research was funded by NIH RO1 EY10016, P30 EY001583 and the Emil Aaltonen Foundation.