Expanding the Interaction Lexicon for 3D Graphics

摘要

This thesis focuses on one of the key challenges for interactive 3D graphics: creating the interaction lexicon. The author's hypothesis is that we can create new interaction techniques by breaking our assumptions about the real world and about existing practice. These new interaction techniques will provide new capabilities and/or make the user more efficient. This research makes several contributions to 3D interaction and virtual reality. The Voodoo Dolls technique is a new technique for manipulating objects in immersive 3D environments in which users manipulate hand-held copies of objects. A formal evaluation comparing Voodoo Dolls and Indirect HOMER (Hand-centered Object Manipulation Extending Ray-casting), a best practice manipulation technique, demonstrates that Voodoo Dolls users can position and orient objects more accurately than Indirect HOMER users, perhaps because of the additional feedback the technique provides. A new technique for navigating large virtual worlds uses place representations and visible landmarks. A formal evaluation comparing navigation with Places and Landmarks with navigation using panning and zooming demonstrates that users can travel between locations faster when navigating with Places and Landmarks. A new process for specifying a 3D model's interaction semantics allows designers to paint interaction surfaces onto the model. The system then saves the painted surfaces in an interaction map. A qualitative evaluation of this process demonstrates that designers, regardless of artistic ability, can easily specify the interaction semantics of 3D models and assign them interactive behaviors when using it. The author also presents 18 ideas that will enhance existing techniques or provide starting points for further exploration, including creating copies to aid selection, multiple light- weight views of an object, system maintained repositories, and symbolic linking.