Computer Science Department
School of Computer Science, Carnegie Mellon University


Expanding the Interaction Lexicon for 3D Graphics

Jeffrey S. Pierce

November 2001

Ph.D. Thesis

Keywords: Human-computer interaction, interaction techniques, 3D interaction, generative method, breaking assumptions, content creation, object manipulation, navigation, 3D painting, interaction surface, interaction maps, voodoo dolls, visible landmarks, place representations

Historically user interfaces evolved in a series of rapid paradigm shifts followed by periods of incremental development. Because widely adopted user interfaces are resistant to change, we can potentially have a greater impact on improving interfaces by working on the next, rather than the current, interaction paradigm. While there are several candidate paradigms, I chose to focus on 3D interaction because of its potential to leverage people's natural and learned abilities. While current 3D interfaces are promising, we have only explored a small part of the design space. I believe that the reason is that we are limited by our assumptions when we build 3D interfaces. We all grow up in a fully immersive 3D world, and when we design virtual worlds we often transfer characteristics of the real world without considering other options. We have also been building worlds long enough to start thinking of the current methods as the correct methods, and thus rather than explore new parts of the design space we emulate existing work. I believe that to realize the potential of 3D interaction we need to expand the interaction lexicon for interactive 3D graphics. My hypothesis is that we can create demonstrably useful 3D interaction techniques by identifying and breaking our assumptions about the real world and about existing practice. I present an existence proof for this hypothesis consisting of interaction techniques that I created using this approach. These techniques allow users to manipulate objects using Voodoo Dolls, navigate large virtual worlds using Places and Landmarks, and specify the interaction semantics of 3D models by painting Interaction Surfaces. I also present experimental evaluations of these techniques that demonstrate their utility.

145 pages

Return to: SCS Technical Report Collection
School of Computer Science homepage

This page maintained by