RESEARCH.
FAMOUS™.
This is a system which primarily works on existing motion capture
data. Therefore specific hardware to obtain the original data is
required. What is highly intuitive about this system is in the way
it works with the motion-capture data. As can be seen from the screen shot,
markers on the geometric model correspond with the traditional markers
used to generate the motion-captured detail on the actor's face.
These markers can then be further assigned vertices on the 3d model by
3d painting in weights using clusters. The
clusters are colour coded depending on the weight assigned. This
allows for explicit handling of the geometry by further defining falloff
regions for the clusters. Clusters may also be overlapped and proportionally
shared or smoothly interpolated. The system also allows for collision
detection by letting the user interactively set the minimum distance constraint
between two clusters.
As can be clearly seen FAMOUS™ allows for a lot of control over the 3d
geometry that was missing from raw mo-cap data. Another problem with
motion-capture is that most hardware systems generate key-frames for every
single frame of the animation. This can lead to a confusing mess
of data which is almost impossible to tweak. FAMOUS™ allows
the user to assign keyframes to specific clusters, and in turn, create
and delete keyframes. The system also allows for tweaking of
keyframes using familiar 'ease-in', 'ease-out' function curve handles.
Finally, it also allows for algorithmic head movement or motion-captured
movement. The relevance of this last feature being that motion capture
generated data can over exaggerate when transferred to 3d geometry.
Examples of FAMOUS™ in action :-ANIM1 - GURU.mpg (791 KB)
ANIM2 - KING TAMAHO.mpg (1.55 MB)