FNRS 
Augmented Reality
Swiss National Foundation June 2000 June 2002
Objectives
A major research goal during the last decade has been to create photo-realistic
interactive synthetic worlds. Despite much effort, however, synthetic worlds
remain artificial environments and it takes a tremendous amount of effort,
both terms of design and computation, to produce truly convincing scenes.
In this work, we propose to investigate the use of augmented reality techniques
to tackle these issues.
Summary of the project
We will start from video sequences or movies of real scenes containing
both objects and people; fit graphics models to some of these; finally,
use those models to move or deform the objects as well as to animate the
people, so as to produce photo-realistic synthetic sequences. To
achieve this goal, the real objects have to be associated with virtual
objects or models that can be used to track them, redraw them at a different
place in the images and/or deform them. We will refer to this combination
of real and virtual objects as ``mixed-objects.''
Rather than recreating a whole synthetic environment, which is both
costly and time consuming, only objects that actually will be animated
or moved need to be modeled. Furthermore, because the mixed-object models
will be derived from images in a highly automated fashion, the complete
work load of the graphics designers will be reduced even more and the possibilities
will only be limited by their imagination.