Distortion in 3D shape estimation with changes in illumination

In many domains it is very important that observers form an accurate percept of 3-dimensional structure from 2-dimensional images of scenes or objects. This is particularly relevant for designers who need to make decisions concerning the refinement of novel objects that haven't been physically built yet. This study presents the results of two experiments whose goal was to test the effect of lighting direction on the shape perception of smooth surfaces using shading and lighting techniques commonly used in modeling and design software. The first experiment consisted of a 2 alternate forced choice task which compared the effect of the amount of shape difference between smooth surfaces lit by a single point light with whether the position of the light sources were the same or different for each surface. Results show that, as the difference between the shapes decreased, participants were more and more biased towards choosing the match shape lit by the same source as the test shape. In the second experiment, participants had to report the orientation at equivalent probe locations on pairs of smooth surfaces presented simultaneously, using gauge figures. The surfaces could either be the same or slightly different and the light source of each shape could either be at the same relative location or offset by 90° horizontally. Participants reported large differences in surface orientation when the lighting condition was different, even when the shapes were the same, confirming the first results. Our findings show that lighting conditions can have a strong effect on 3-dimensional perception, and suggest that great care should be taken when projection systems are used for 3D visualisation where an accurate representation is required, either by carefully choosing lighting conditions or by using more realistic rendering techniques.