Purchase this article with an account.
Sylvia C. Pont, Andrea J. Doorn, Jan J. Koenderink; Light field matching. Journal of Vision 2004;4(8):122. doi: https://doi.org/10.1167/4.8.122.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The appearance of objects is determined by the light field, the geometrical properties, surface reflectance and surface structure of the object. Conversely, the appearance of objects provides us with cues about the light field in a scene. The illuminance flow over an object is related to the shape, the reflectance, the surface structure and the light field. Question: Are human subjects able to match the light field of scene of a rendered Lambertian sphere in space with the illumination conditions under which photographs of real, roughly spherical objects were taken? Method: The photographs of the real objects were taken under four angles of collimated illumination (similar to direct sunlight), under hemispherical diffuse illumination (similar to an overcast sky) and under fully diffuse illumination (similar to a polar white-out) and afterwards photometrically calibrated. The natural objects had surface structures ranging from smooth to very rough, reflectances of a large variety, and translucencies of different degrees. The light field probe was rendered real-time. Subjects could adjust the azimuth, elevation, ambience (from fully collimated to fully diffuse) and albedo of the probe. The albedo setting was of course ambiguous due to the light intensity — surface reflectance ambiguity, but subjects indicated that it was necessary to adjust it such that the brightnesses of the probe and the stimulus objects were of the same order of magnitude. Result: We found that subjects were well able to match the illumination of a rendered lambertian sphere with that of a real object, except a few interesting cases: all subjects showed large deviations from veridical for the translucent objects. The azimuth of the illumination was matched very closely. The elevation and the ambience settings showed interactions. This interesting finding suggests that subjects cannot separate the effects due to the elevation of the source, and the degree of collimation very well.
This PDF is available to Subscribers Only