Purpose. What determines the color appearance of real objects viewed under natural conditions? The light reflected from different locations on a single object can vary enormously, even when the object is made of a uniform material. One source of variation is inhomogeneity in illumination; another is that the relative contributions of diffuse and specular reflectance change across the object. Yet humans have no trouble assigning color names to most things. We have begun to study how this works. Methods. Subjects viewed a graphics simulation of a three-dimensional scene containing two spheres, test and match. The subject's task was to adjust the match sphere until its color appearance was the same as the test sphere. The match sphere was always matte, and subjects varied its color by changing the simulated diffuse spectral reflectance function. A variety of test sphere materials were simulated by varying the strength and roughness of the specular reflectance component. This was done using Ward's parametric BRDF model. The test sphere's diffuse reflectance component (“body color”) was also varied. Scenes were rendered as stereo pairs using RADIANCE, combined with custom software that ensured spectral accuracy. Subjects viewed the stereo pairs on a calibrated computer-controlled haploscope. To date, data have been collected from two non-naïve observers. Results & Conclusion. For fixed test sphere body color, observers' matches depend on the simulated test sphere material. The data thus reject the hypothesis that perceived object color depends only on the object's diffuse reflectance. A successful theory of object color appearance will need to account for the interaction between material properties and perceived color. One plausible conjecture is that the matches depend only on the spatial average of the light reflected from the test sphere. The data reject this simple possibility. Indeed, the diffuse reflectance of the test predicts matches better than the spatial average.
Supported by: NIH Grant #EY10016