Evaluating 3D pointing techniques

"This dissertation investigates various issues related to the empirical evaluation of 3D pointing interfaces. In this context, the term ""3D pointing"" is appropriated from analogous 2D pointing literature to refer to 3D point selection tasks, i.e., specifying a target in three-dimensional space. Such pointing interfaces are required for interaction with virtual 3D environments, e.g., in computer games and virtual reality. Researchers have developed and empirically evaluated many such techniques. Yet, several technical issues and human factors complicate evaluation. Moreover, results tend not to be directly comparable between experiments, as these experiments usually use different methodologies and measures.

Based on well-established methods for comparing 2D pointing interfaces this dissertation investigates different aspects of 3D pointing. The main objective of this work is to establish methods for the direct and fair comparisons between 2D and 3D pointing interfaces. This dissertation proposes and then validates an experimental paradigm for evaluating 3D interaction techniques that rely on pointing. It also investigates some technical considerations such as latency and device noise. Results show that the mouse outperforms (between 10% and 60%) other 3D input techniques in all tested conditions. Moreover, a monoscopic cursor tends to perform better than a stereo cursor when using stereo display, by as much as 30% for deep targets. Results suggest that common 3D pointing techniques are best modelled by first projecting target parameters (i.e., distance and size) to the screen plane."