Recovering 3D shape from concept and pose drawings

Modern tools to create 3D models are cumbersome and time-consuming. Sketching is a natural way to communicate ideas quickly, and human observers, given a sketch, typically imagine a unique 3D shape; thus, a tool to algorithmically interpret sketches recovering the intended 3D shape would significantly simplify 3D modeling. However, developing such tool is known to be a difficult problem in computer science due to multitude of ambiguities, inaccuracies and incompleteness in the sketches. In this thesis, we introduce three novel approaches in CAD and character modeling that successfully overcome those problems, inferring artist-intended 3D shape from sketches. First, we introduce a system to infer the artist-intended surface of a CAD object from a network of closed 3D curves. Second, we propose a new system for recovering a 3D model of a character, given a single complete drawing and a correspondingly posed 3D skeleton. Finally, we introduce a novel system to pose a 3D character using a single gesture drawing. While developing each system, we derive our key insights from perceptual and artist literature, and confirm our algorithmic choices by various evaluations and comparisons to ground truth data.