Intelligent planning of communicative acts for computer animation cinematography.

摘要

The creation of animations, even with the aid of computers, is a difficult and time-consuming task. In the near future it will be possible to program an intelligent animation assistant that will enable animators, both amateur and professional, to create animations quickly and easily due to the automation of most modelling and presentation tasks. This thesis presents a tool that implements an important component of this anticipated animation assistant, namely the automation of visual presentation through cinematography techniques. The system presented here is based on the idea that an animation is a method by which an animator conveys communicative acts to an audience. A communicative act is any information, feeling, or theme that the animator would like to impart to the viewer while telling a story. Our system transforms these communicative acts, using Rhetorical Structure Theory (RST) as a guide, into a set of animation instructions for a renderer. The final output is an animation that is intended to produce in the viewer feelings and impressions that are consistent with the animator's description of the mood and theme of the narrative.; The prototype system contains three major components: a cinematography knowledge base, a communicative act planner, and a renderer. The knowledge base, which is implemented using the description logic language, LOOM, encodes cinematography concepts such as 3D vectors, spatial objects, lights, cameras, colour energies, moods, and themes. This knowledge base acts as the store of qualitative knowledge needed to reason about cinematography. The communicative act planner uses a visual interpretation of RST to guide the planning process. The RST planning agent achieves its goal of transforming communicative acts into visual effects by generating a set of cinematography instructions for the renderer. The renderer transforms the cinematography effects requested by the planner into the final animation frames that constitute the animation. The renderer transforms qualitative knowledge into quantitative instructions needed by a ray-tracer. The system is intended to apply cinematography knowledge to a pre-defined animation in order to communicate information, emotion, and intentions to the viewer. Examples show how cinematography techniques are automatically applied in short animations.