Changes in Perceived Object Shape with Changes in Lighting Model and Surface Properties

Abstract

We examined the influence of lighting and bidirectional surface reflectance distribution function (BRDF) on the perceived shape of real objects viewed monocularly and binocularly. We asked observers to adjust a ‘circle-stick’ gradient probe optically superimposed on a smooth, asymmetric, pear-shaped wooden object placed directly in front of the observer at a distance of 60 cm. By using real objects, we avoid problems with unrealistic rendering algorithms and CRT presentation of stimuli. The lighting models used were simple (a single near-punctate source and a flat black background), the viewing conditions (in the binocular case) realistic, and the BDRFs selected not implausible. We did not assume that the viewer's perception of shape is veridical under any of the conditions considered, nor did we assume that surface shape estimated from gradient probe settings is an unbiased estimate of perceived shape. We sought to establish whether changes in lighting model and BRDF affect observer performance and, by implication, perceived shape. Observers viewed the object under all eight of the possible combinations of the binary factors: (a) monocular and binocular viewing of the object, (b) near-punctate illumination from above-left or above-right, and (c) matte wooden BRDF (the natural surface of the object) and gloss white BRDF (the same object painted). For each condition there was a total of ten gradient settings at each of over one hundred surface locations. The same locations were used for each of the eight conditions. We report results of analyses and discuss their implications.