Digital 3D models are ubiquitous. Yet, for the creation andmodification of 3D models we rely on highly specialized tools, whichunfortunately limits the user base to trained professionals. In thistalk I will describe and demonstrate two of my recent projects thatare designed to alleviate this problem. Generating plausible andvisually pleasing shapes and deformations is far from trivial: whileour capability to derive a mental model from everyday shapes around usis well developed, we fail to properly communicate this to a machine.This is why we have to model in a loop, constantly correcting theimproper interpretation of our intentions. Ideally, a sketch-basedmodeling system for 3D shapes should use the very same sketches thatdesigners would draw on a piece of paper to convey the shape. Butwhich lines are best suited for this task? The tools I willdemonstrate use a subset of such feature lines. And although theserepresent a good trade-off between complexity and usability, I willargue that this is only one of many steps towards the notion of a"human video-out".
