My research focuses on simplifying content creation, in particular to produce highly detailed patterns, structures and shapes. Applications range from Computer Graphics to additive manufacturing (3D printing), for rendering of highly detailed scenes, for the design of complex objects, or for the modelling of technical parts that have to follow precise specifications.
The main approach I follow is to develop fast, highly controllable by-example approaches. These techniques are able to generate new content -- images, 3D models -- resembling a given example while enforcing user specified constraints. Such techniques are easy to use through example specification, enabling users to focus on their main design task while the algorithm produces details and enforces constraints.
Detailed content can quickly lead to large, heavy models, with files of several gigabytes in size. This hinders fast processing and visualization. Instead, I focus on methods that can quickly generate content on-demand, at the resolution it is needed and only where it is needed (e.g. for the current viewpoint). To support this line of research I investigate novel data-structures and algorithms running on parallel graphics processors (GPUs).
Since 2012 I am leading the ERC funded project ShapeForge which is targeted at bringing complex, by-example shape generation to additive manufacturing, as well as optimizing shapes to consider fabrication and structural constraints. As we explore the generation of complex objects, we need tools able to process these shapes and drive the 3D printers during fabrication. Our software, IceSL, simplifies the digital modelling and 3D printing of complex objects using techniques from modern Computer Graphics.