Rice University
presents

Aaron Hertzmann

Media Research Laboratory

Department of Computer Science

New York University

Abstract

Algorithms for Expressive Painting and Animation

Aaron Hertzmann will present methods for painting images and video. He begins with a still image algorithm inspired by painting with real media. The input is a photograph or 3D rendering to be painted. The algorithm first paints a rough sketch with a large brush, and then refines it with smaller brushes. Long, curved brush strokes are placed to follow image contours. This method produces images with a much greater subjective impression of looking hand-made than do earlier methods. By adjusting algorithm parameters a variety of styles can be generated, such as styles inspired by the Impressionists and the Expressionists.

The still image algorithm is extended to processing video by analyzing motion in the input video, and placing brush stokes over time accordingly. Different choices of parameters give different styles of animation. Hertzmann will show several examples, including video of a jazz recording session commissioned for an upcoming DVD release, and clips from an interactive exhibit.

Hertzmann is developing a new relaxation-based algorithm, in which he searches for the painting that minimizes some energy function. Different energy functions correspond to different painting styles. He will present some encouraging preliminary results that extend the range of styles available, and provide more direct control to the artist.

In many contexts, painting holds several benefits over photorealistic computer graphics, related to aesthetics, expression, and computational requirements.

Monday, November 29 at 4:00 in Duncan Hall 1064
Reception to follow.