figure 1. modeling session in teddy. users create 3D models using simple sketching operations.

tent. End users constructing 3D models are also supported by a number of systems, including Google’s SketchUp ( http://sketchup.google.com/) and modeling tools in SecondLife and Spore. However, these systems use scaled-down versions of traditional interfaces and still require a certain amount of skill. This article introduces research efforts at the University of Tokyo and Brown University to make computer-graphics authoring accessible to more casual users. To achieve this goal, the author and his collaborators developed easy-to-use prototype systems to create expressive computer graphics more quickly than with traditional interfaces. Examples are sketch-based 3D modeling, clothing manipulation, animation by performance, and 2D shape manipulation. We discuss the user interfaces and technical aspects of these prototype systems, as well as the lessons learned from their development, offering ideas for future research directions.

Most of our work is highly interactive and difficult to explain in written words and still images; please see demonstration videos and prototype systems at http://www-ui.is.s.u-tokyo. ac.jp/~takeo.

 

figure 2. screenshot of teddy and sample 3D models created through the teddy system.

figure 3. using teddy to teach the concept of contour lines.

sketching 3D models Creating a 3D model in a computer (not necessarily on a screen) is the first step in most 3D computer-graphics applications yet is also the most difficult. Traditional interfaces for 3D modeling programs trace their origins to traditional pencil-and-paper professional drafting. Users place vertices in 3D space by specifying x-, y-, and z- coordinates in a three-view interface, then create polygonal faces (individual polygonal sides of a polyhedron) by connecting these vertices. Alternatively, users start with a simple primitive (such as a sphere or cylinder) and modify it by editing individual vertices and edges. Many editing tools (such as free-form deformation and Boolean operations among solids) are available for designing complicated shapes from simple primitives. Although they might be appropriate for trained professionals designing precise models, they are generally too difficult for first-time users trying to quickly generate meaningful models.