Abstract
Purpose: In a ‘daylight’ scene, with a punctate light source (‘sun’) and a diffuse light source (‘sky’), the spectral power distribution (SPD) of the effective light source illuminating a surface patch is a weighted linear combination of the SPDs of the punctuate and diffuse light sources. The weight assigned to the punctuate SPD is proportional to the cosine of the angle between the surface normal and the direction to the light source. We tested whether human observers take into account the orientation of a surface with respect to the punctuate light source in making achromatic settings in such ‘daylight’ scenes and whether their achromatic settings were consistent with the cosine law above.
Stimuli: The stimuli were stereo image pairs of complex scenes all containing similar specular and matte objects. The scenes were rendered with diffuse and punctate light sources that could differ in chromaticity. Each scene contained a gray matte test patch which could appear at any of seven different angles (from 50 degrees to −50) degrees with respect to the punctuate light source.
Task: Observers were asked to adjust the chromaticity of the test patch until it appeared achromatic. They could adjust the chromaticity of the patch in small increments in two directions in color space, corresponding approximately to perceived blue-yellow and red-green directions. Observers repeated this adjustment task ten times for each scene. Four observers participated in the study.
Results: We used each Observers mean achromatic settings to estimate the chromaticity of the equivalent illuminant (Brainard et al, 1997) that he is effectively discounting. We plotted equivalent illuminant as a function of orientation. As the test patch was rotated away from the direction of the punctate light, the contribution of the punctate light to chromaticity did decrease with the cosine of the rotation and relative contribution of the diffuse light increases, as predicted.