In this thesis, we are concerned with animation models for virtual characters. More specifically, we wish to promote the reuse of animation models and the components and relationships used to construct them both within a single character and between multiple characters. We also wish to provide the possibility for reuse between separate virtual character architectures. In order to accomplish this, we have provided both a low-level representation for animation models and a specification mechanism for building animation models for this representation.; We derived our model representation from the areas of Synergetics and Dynamic Pattern Theory. The resulting Virtual Behavior (ViB) model represents motion as a collection of time varying variables, or channels, that are evaluated at discrete time intervals. Each channel has a corresponding signal, a function of the model state, which is used to determine the change, or delta; in the variable since the previous evaluation time. A variable may be modified by several sources. Consequently, a channel's signal is made up of multiple independent contributions functions whose outputs are combined, or blended, to determine the output of the signal.; than that provided by the ViB model. Consequently, we have combined concepts from Alexander's work on Pattern Languages with some programming language scoping concepts in order to define pattern oriented composition. Using this of these specifications into the ViB model representation. We have also defined barriers between model components.
展开▼
