Motion Transformations with Spacetime Constraints
Michael Gleicher <http://www.cs.wisc.edu/~gleicher>
Department of Computer Sciences
University of Wisconsin - Madison
Most motion for computer animation is single purpose: it applies to
a specific character, performing a specific action, in a specific environment.
Tools for transforming motions, for example to adapt a given motion to
a new character or environment, offer an exciting opportunity to enable
more widespread use of computer animation by enabling non-animators to
apply the work of others. However, the creation of such tools is difficult:
transforming requires us not only to meet the new needs for the results,
but also to retain as much of the desireable properties of the original
motion as possible.
In this talk, I will introduce a Spacetime Constraints approach to
motion transformation. We cast the problem as a constrained optimization:
find a new motion that, subject to the constraints of meeting the new goals
and preserving the distinguishing characteristics of the original motion,
is as similar to the original motion as possible. We call such an approach
Spacetime Constraints because it requires us to consider a duration of
motion in a single, large, constrained optimization problem. By careful
choice in how we pose the problem, by judicious use of simplifications
and approximations, and by careful implementation, the approach can be
made practical. I will demonstrate the approach on two core animation tasks:
retargetting motions to new characters and direct manipulation editing
of animated motion.