For some time now, computers have had the ability to display stereo imagery, often referred to as stereovisuals1. This is a situation where each eye of the viewer is shown a slightly different image to the other eye. The individual images have the same subject, but are observed from two positions offset by a value equivalent to the distance between the viewer's eyes (sometimes called the parallax of the stereo image). The brain of the viewer then combines these left and right eye images in the same way as when viewing any real scene. The result is an image that appears to be ``really there'' with depth perception, the ability to see how far away an object is without having to consider properties such as its relative size.
Traditionally, support for using stereovisual output in an application required stereo vision to be supported natively within the application itself. However, many applications may output stereovisuals simply by outputting two standard views, each offset by a parallax separation. Tools to process and display these sorts of images are not common, and usually consist of either opaque, inextensible utilities or trivial example programs. As such, general purpose tools for stereovisuals are lacking.
In addition, computers with video cameras attached are becoming increasingly prevalent. The most common type at present are webcam devices (refer to Section 1.4), which are cheap and usually deliver reasonable quality video at good frame rates. SGI now ships webcams as standard across their range of workstations.