Animal Modeling and Animation

Animal Modeling, Animation, and Motion Capture

We are interested in modeling and animating articulated creatures such as humans and other animals. Our modeling paradigm is anatomically-based: animals have underlying bones, muscles, and generalized tissues modeled with polygon meshes or ellipsoids. Components are highly parameterized for re-use between animals. Skin can be generated from the underlying parts, or a polygon mesh skin from elsewhere can be attached. Skin points are associated with their nearest underlying parts and move with them. When joints move, muscles automatically reshape and skin points initially move with their underlying parts. An iterative elastic relaxation results in a balanced skin mesh.

Deformable models are just a start. We also need realistic joint motion. Studio motion capture provides detailed animation, but isn't appropriate for many situations, such as children playing in the surf, or cats hunting in the wild. We are exploring extraction of 3D motion from monocular digitized video sequences. Our long-term goal is to develop a movement library that encodes motion that can be reused in different environments and for different creatures.

Research projects are described below from newest to oldest. This research is based upon work supported by NSF grants number CCR 9972464 and MRI 9724237.
A list of our publications in computer graphics and computer animation is available.

The Fauna Group

Professors: Jane Wilhelms and Allen Van Gelder .
Graduate students: Mark Slater.
Post-Doctoral Researcher: Maryann Simmons.
Alumni: Leon Atkinson-Derman, Jeffrey Lapierre, Alison Luo, Philip Schneider, Marlon Veal, Amanda Wahl, and Rose Wahlin,

Integrating methods from graphics and vision (see below) in developing virtual horse models and animation.

Exploring methods that combine computer vision techiques with interactive manipulation to extract arbitrary motion from video

For better control over a model with many joints, methods for using super-segments -- a chain of segments that move together using forward or inverse kinematics.

Methods for attaching a arbitrary polygon mesh skin to underling geometry and components designed to fit it.

An improved model used polygon-mesh bones, a more correct elastic model for skin, and non-proportional scaling for greater detail. We also experimented with fast fur modeling. (The Fur Page)

Initially, all underlying constituents (bones, muscles, etc.) were ellipsoids. Next, we developed a model including all the movable vertebrate joints. The muscles were modeled as polygon meshes.

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. (as required by NSF).