Realistic Animation of Human Motion
High quality human motion data is expensive, whether it comes from
captured human motion, motion created by a talented animator, or
motion developed by an experienced behavior designer. The goal of
this project is to reduce the expense involved in creating new motion
and altering existing motion in a realistic manner. One objective is
to develop a real-time environment that allows an animator to interact
with physically-based models. We are also developing techniques to
allow users to incrementally create physically-based animations.
For more information...
Fast dynamic scaling using simple
machine abstractions
Dynamic scaling of control system parameters
Participants: Nancy Pollard, Jonathan
Cummings, Carolyn Uy, Fareed Behmaram-Mosavat
Even simple activities that involve interacting with objects in the
environment are difficult to animate. Activities involving
interaction may range from very subtle (e.g. picking up an object on
the desktop) to very dynamic (e.g. the collision of two players on the
football field). The human eye is quite good at picking out
physically unrealistic behavior in such scenarios. We are developing
new models of interaction between characters and their environment
that can be used to simulate these types of activities in a realistic
manner.
Participants: Anthony Fang
Animated human characters have not yet achieved a high level of visual
realism. We are exploring techniques to better model the biological
system in an animation setting. This involves developing more
accurate representations of underlying structures such as tissue and
joints.
Participants: Remco Chang
2D or 3D plots are not always the best way to visualize information
about the physics of motion. This project is investigating techniques
to assist a user in understanding motion from a physical point of
view.
Participants: Moira Burke (U. of Oregon)
For more information...
Results -- Rendering Dynamic Motion
Past Projects
We are exploring image-based techniques to allow large groups of
characters to be viewed and controlled in real-time. As rendering
algorithms become more complex, the rendering step may become the main
bottleneck to achieving real-time animation of large scenes. If we
can avoid re-rendering characters for a number of frames yet still
represent the character's appearance and motion with an acceptable
level of fidelity, then we can display and control larger scenes in
real-time.
For more information...
Results -- applying image-based rendering techniques to animation
Participants: Audrey Wong
Football coaches create and distribute playbooks for their teams at
the start of every season. These playbooks represent a static image
of each play. A computer representation of such a playbook has a
large number of advantages, including the ability to view the
evolution of a play over time, develop new plays on the fly, and
present a view of the field from any player's point of view. We are
creating a gesture-based implementation of an electronic playbook.
For more information...
Results -- COACH application
Participants: Jeff Pickering