Contents 1 System Description
Our main design choice was to use a desktop system, with a mouse as the input device. This was motivated by the fact that desktop systems and mice have (typically) lower latencies than other alternatives. Also, today there are good methods to map 2D input to 3D manipulation. In particular, we use a contract assumption, snapping, sliding, collision detection, interaction only with visible objects, and viewer constraints, as previous work [4] has shown all these to be beneficial. We display the 3D scene in perspective, but do not use stereo display, as it is to us (at best) unclear, if stereo has strong benefits for 3D interaction.
