Multi-Resolution Behaviors for Large-Scale Environments
Click on picture for large version
|
|
|
Small World
|
Waltz
|
Our goal is to build systems for the authoring and emulation of highly
interactive, large scale narrative virtual environments. In order to
support very large worlds, we want to present the user with an
approximation of both the geometry and behaviors, computing only
enough detail to emulate a plausible experience based on the limits of
the user's time-varying perception, knowledge, and expectation of the
environment. Our two approaches for achieving this goal are
procedural modeling and a multi-resolution description of behaviors.
Both as a modeling aid and as a means of data compression, we describe
the world procedurally using stochastic subdivision techniques,
generating geometric and behavioral detail only as needed by the
application. Authored content, in a structure akin to a scene graph,
provides the parameters that guide the procedural generation of the
world. Its description includes both how the world should change over
time as well as how it should adapt to changes in the user's position
and orientation within it. The runtime system must make trade-offs
between the available computation power and the visual and
experiential fidelity of the world including, especially, the author's
intent. The behavioral descriptions, accordingly, contain hints as to
their application importance such that the runtime system may choose
appropriately how to adapt their detail both over time and as the user
moves about the world.
Participants: Steven Dollins
Interactive Densely Populated Scenes
In computer graphics, we seldom see interactive scenes that contain
a large population. For example, the settings for most computer games
are indoors where dungeons and rooms provide a physical limitation to
the number of characters that can be present. Furthermore, scenes that
do incorporate a large population are usually not interactive due to
the inherent computation required to simulate them. This is more apparent
as the complexity of each character increases. In addition to individual
complexity, if we allow the characters to interact with each other (in
a pairwise manner), the computation required to simulate them may grow
at a quadratic rate with respect to the number of characters in the scene.
We have developed multi-resolution techniques in geometry, animation,
and behavior. Whereas most multi-resolution research has focused on
simplifying geometry, animation, or behavior separately, we allow our
system to leverage information from the three different areas. We apply
these techniques to a Ballroom Dancing simulation.
Participants: Dom Bhuphaibool
Animation and Multi-Resolution Behavior
graphics web master
