Background

Globalisation and growing international networks make efficient computer supported work of interdisciplinary teams increasingly important, in particular for local and also for distributed product development, training, education and communication. Many of these processes in today’s industry require interdisciplinary groups of people to work on collaborative tasks involving complex three-dimensional models. Virtual environments hold the promise to provide a shared space in which such collaborative interactions of multiple users with virtual 3D models could occur as easily and smoothly as they would in the real world. Large projection-based stereoscopic displays are the most successful incarnation of such immersive virtual environments. They allow groups of people to simultaneously view a virtual three-dimensional scenario and support direct communication by voice and gestures. These systems use a fixed view point or track the head position of a single user and compute a stereoscopic image pair for that point of view. Unfortunately, when non-tracked users see these images, the virtual environment may appear severely distorted and it changes shape when users are moving. These distortions make it difficult to have a correct spatial understanding of the virtual word. As a consequence users cannot properly interact with the virtual objects and each user has a different spatial mental model of the virtual word, which limits the applicability of virtual environment technology considerably.