Nonphotorealistic Rendering

 

Dr. Alejo Hausner

Computer Science Department

University of New Hampshire

Durham, New Hampshire

 

Date: Monday September 18, 2006

Time: 12:20 p.m. - 1:10 p.m.

Location: 367 Votey

 

 

Abstract

 

Nonphotorealistic Rendering (NPR) aims to reproduce the techniques of artists and illustrators.  Since its beginnings in the 1970s, computer graphics research has focused on photorealism, which uses the physics of light to accurately reproduce a photographic camera.  However, human artists and illustrators have different goals, and their methods take advantage of features of the human visual system to create images that communicate effectively, even though no photographs are involved. In this talk, I will describe several NPR methods: simulated tile mosaics, pointillist painting, and artistic dither.  Tile mosaics create images by assembling many small colored shapes. Some mosaic artists are remarkably subtle, and computer simulations still cannot attain their level of skill.  Nevertheless, I will present a method based on centroidal voronoi diagrams that addresses a central task in mosaic construction: how to distribute square tiles, to cover an image with minimal gaps, while allowing the tiles to have arbitrary orientations.  Pointillist painters produced images with many small dots of pure color, which the human eye blended into the desired image.  Colors in their laboriously-constructed paintings are richer and more scintillating than those obtained by mixing the paints before applying the brush strokes.  This talk presents methods for choosing dot colors based on error diffusion.  Artistic dithering uses techniques in halftoning (printing on a device with only two or four ink colors) to create decorative patterns.  The methods presented use hierarchical graph coloring to create ordered dither matrices.  These matrices can be used for ordinary printing, or to impose a decorative pattern on an image.

 

 

Speaker bio:

Alejo Hausner earned a PhD in CS from Princeton in 2001, after studying physics, mathematics and computer science at McGill and Queen's in Canada.  His primary research area is computer graphics, focusing on nonphotorealistic rendering, geometry and halftoning.