The Study of the Input Tool Based on Computer Vision

This study empirically explores a cube-shape tool based on computer vision, focusing on the properties of its input modalities and design space. By tracking the fiducial marker on each side of the cube in real-time, basic operations such as flipping, rotating and tilting can produce various input modalities. We present the results of an experiment that investigate the use of these input modalities to perform a set of experimental tasks. In the experiment we design a discrete angle selection task, varying the angle position and interval. Based on the experimental results, we discuss implications for the design of cube-based interaction techniques.