Lightness estimation in three-dimensional scenes with and without specular cues

Purpose: We report two experiments measuring how observers estimate the albedo ("lightness") of achromatic, matte surfaces at different depths in three-dimensional, rendered scenes. In each scene, illumination intensity varied with depth. We examined whether observers compensated for this variation in estimating albedo. Stimuli: Each scene consisted of two small rooms juxtaposed along the observer's line of sight. Scenes were viewed binocularly. The observer could see the interior of the near room and part of the interior of the far room through a doorway. All surfaces of the rooms were tiled with achromatic, matte Mondrians that varied randomly across trials. Light sources were placed in the far room and behind the observer and were never directly observable. This design was modeled on Gilchrist (1977, 1981). Methods: We used an asymmetric lightness matching task to measure observers' "equivalent illuminant intensity (EII)" (following Brainard et al, 1997) at each of ten different depths (signaled by binocular disparity) within the two-room scene. Nine observers made ten matches per depth level. Experiment 1: Illumination was intense in the far room and dim in the near room. We found that EIIs for 8 of 9 observers' matches did vary significantly with depth, but we found no clear relationship between observers' EII and the actual illuminant intensity at each depth in the scene. Experiment 2: A possible explanation of the results of Experiment 1 is that observers had difficulty in estimating illumination intensity changes across depth. We repeated experiment 1, with the addition of small specular spheres to the scene. The location of the spheres varied randomly from trial to trial. We then found that observers' equivalent illuminants at each depth were roughly proportional to actual illuminants with a constant of proportionality of 0.5. Observers substantially discount changes in illumination intensity with depth.