Stereo displays using LCD shutterglasses are used for a variety of tasks, including Computer Aided Design of industrial products. Graphics for these displays use Emmert's Law to covary retinal angle and stereo disparity of an object displayed at different depths. Anecdotal reports from designers suggest that depth judgments made in these environments show systematic errors, and that the same object displayed at different depth is often seen as different in size. If the range of perceived depths were compressed relative to the stereo disparity, distortions such as those reported would result.
Our experiment asked subjects to judge the size and distance of physical and computer-generated graphical spheres in low context conditions (a single sphere positioned on a dark background). Subjects consistently underestimated the distance of spheres that were further away than the projection screen and overestimated the distance of spheres that were closer to them, resulting in a compression of distance around the plane of the screen. Size judgments in the virtual condition were highly inaccurate relative to the physical control, however the high variance in virtual condition size judgments did not permit any clear explanation of information processing in that task.
Since extraretinal inflow is a potential cue for depth of a fixated object, we also evaluated a technique for manipulating the extraretinal signal by placing the display well above eye level (eccentric gaze). Our results showed a similar but smaller effect for eccentric gaze in the physical control condition compared to the stereo display. This suggests that extraretinal inflow alone is not sufficient to explain the biases in depth and size judgments in these environments, although it may be a contributing factor. Further work is being conducted on comparisons between physical and virtual as well as comparisons of virtual to virtual stimuli.
We would like to acknowledge the support of the BC Advanced Systems Institute and General Motors of Canada