Real-time generation of user- and context-sensitive three-dimensional animated explanations

Intelligent real-time animated 3D explanations and virtual cinematography are essential for responding to arbitrary user viewing queries in increasingly sophisticated virtual environments. Users should be able to pose questions whose animated explanatory responses entail complicated multi-object views subject to constraints on how those objects should appear. Additionally, the resulting animations should be dynamically adapted to the user's style preferences, knowledge of the domain, or task. The intriguing possibility of designing intelligent 3D animation systems with the ability to generate user-customized visualizations, cinema, or explanations is a very recent and largely unexplored area of research.;For example, interactive 3D learning environments could allow students to explore a virtual representation of a domain, engage in problem solving, and receive animated explanations in response to their questions. These systems could compose customized animated explanations that include establishing shots and highlights to introduce unfamiliar material. A significant challenge is determining the placement (position and aim direction) of the virtual camera and the transitions from one camera placement to another to film the action in the 3D environment. This problem is particularly challenging since the virtual camera must respond to arbitrary viewing goals in a virtual environment composed of multiple static and moving objects. Partial constraint satisfaction is employed to compute camera placements that clearly depict the salient features of an arbitrary set of objects subject to viewing constraints. If an optimal camera solution is not possible, then weak constraints may be relaxed, or the constraint problem may be decomposed into sub-problems resulting in multiple views that each convey a portion of the desired camera viewing goal.;A real-time 3D animated explanation generator has been implemented with examples for the domains of plant biology and electromagnetism. The automated cinematography sub-system has been used to implement a task-sensitive cinematography interface for a 3D learning environment and a user-customizable cinematography interface for an interactive fiction system. Each prototype system and the underlying animation and camera planning sub-systems have undergone extensive evaluation.