Improving shape perception in virtual reality systems using toed-in cameras

Abstract

Nowadays, actual stereoscopic 3D renderers use two cameras to render images to the screen. These two cameras have parallel optical axis. It is well know that if cameras converge, then it produces distortions called vertical parallaxes. These distortions are supposed to stress visual system, but we do not know the effect on perception. In this article, we will test if these vertical parallaxes can improve shape perception. We found out that vertical parallaxes does improve shape perception, but the effect is decreased when the object is far from the user because when the cameras converge far away it get closer to a parallel configuration.