Date: Tue, 10 Dec 1996 21:34:06 GMT Server: NCSA/1.4.2 Content-type: text/html Last-modified: Wed, 20 Nov 1996 17:09:35 GMT Content-length: 2876
I do graphics stuff.
I started out doing user interface research, but have migrated into graphics since then.
My advisor is David Salesin. He came to our department in 1992 with a whole bunch of cool ideas for graphics research. One of those was non-photorealistic rendering, specifically pen-and-ink illustration. Our graphics group worked on two projects in this area. One was the automatic generation of pen-and-ink illustrations given three-dimensional models and some texture information. The other one, the one I worked on, addressed the interactive creation of pen-and-ink illustrations given a two-dimensional grayscale starting image.
During that project, we ran into a problem stemming from the fixed resolution of the images we were using. We wanted to be able to magnify these images while maintaining sharp features. That is, instead of blurring the image to compensate for the magnification, we wanted features to stay crisp. This requires that a little more information be stored with the image, but nothing that couldn't be captured as the image was created. Anyway, we did a project addressing these problems, and the resulting paper was accepted by SIGGRAPH. For a quick preview, here's a brief demonstration of what we're trying to do. Note the sharp line segment in the righthand image.
![]() |
![]() |
![]() |
by pixel replication | by blurring | with a discontinuity |
The rest of our graphics people are really good, and we're getting more people all the time. Our group is the Graphics and Imaging Lab, or Grail.