We present an experiment on depth cues for knot visualization in desktop virtual reality (fishtank VR). We used a within-subjects experiment to investigate stereoscopic visuals and 3D lit shading. Participants were required to count crossings in visualized knots – a fundamental task in classical knot diagram interpretation. Participants made significantly more errors with stereoscopic visuals than with monoscopic visuals, while no effect was detected for shading. Since this contrasts evaluations of fishtank VR for other applications, it provides knowledge about limitations of stereo imaging. In addition to the experiment, we outline our knot visualization approach, which uses a dynamic simulation and charged rope analogy to generate knot shape, and includes a technique for rendering gaps like those found in classical knot diagrams.
Nicholas G. Lipari, Christoph W. Borst