3D modeling with silhouettes

Abstract

With the increasing power of computers and the spread of dedicated graphics hardware, 3D content has become ubiquitous in every field, from medicine to video games. However, designing 3D models remains a time-consuming and costly process, so far limited mainly to professionals who have trained extensively with complex modeling suites. We present a new sketch-based modeling approach which aims to make this process simpler and cheaper for professionals, while also making it easy enough for amateurs to use. In our approach, models are interactively designed by drawing their 2D silhouettes from different views. The core idea of our paper is to restrict the input to 2D operations, removing the need to explicitly create or position 3D elements. To create complex shapes, we allow the user to perform CSG operations on silhouettes. We introduce a new algorithm to handle CSG operations that leverages special properties of silhouette cylinders to convert the CSG problem into one that can be handled entirely with 2D operations, making implementation much simpler and more robust. We also introduce a smoothing operator that can generate a smooth shape that matches a set of input silhouettes. We evaluate our approach by using it to model a random sampling of man-made objects taken from the words in WordNet, and show that all of the tested man-made objects are suitable to being modeled quickly and easily using their silhouettes. We also present a user study that demonstrates that novice users are able to use our implemented modeler more effictively than Google SketchUp, a leading commercial 3D modeler also aimed at novice users.