EECS 598-5/498-2:
Non-photorealistic Computer Graphics

Fall, 2003

Images above from WYSIWYG NPR: Drawing Strokes Directly on 3D Models, SIGGRAPH 2002.

Professor: Lee Markosian.
Hours: Tue/Thu 1:10 - 2:30
Location: EWRE 136 (Environment and Water Resources Engineering Building, room 136.)

Newsgroup: umich.eecs.class.498-002
Web page: http://www.eecs.umich.edu/courses/eecs598-5/
The same web page can be reached via http://www.eecs.umich.edu/courses/eecs498-2/

Quick links

Summary

The traditional "holy grail" of computer graphics has been photorealism -- the notion that the ideal computer graphics imagery would be visually indistinguishable from a photograph. But realistic modeling of humans, animals, and natural scenery is extremely difficult, and there is good reason to believe that it will always be so -- realism demands lots of details that all have to be specified correctly (in a "photorealistic" sense).

In contrast, hand-drawn images can depict natural scenes with a huge range of styles, often evoking complexity while retaining an underlying simplicity. Through abstraction and careful elision of detail, hand-drawn imagery can focus the viewer's attention on important information while downplaying extraneous or unimportant features. An artistic scene can also suggest additional semantic information, such as a specific mood (eg, ominous or whimsical), or a quality of "unfinishedness" that may be desirable, for example, in an architect's rough sketch of a new building. Can the same principles apply to computer graphics? Of course, although how to fully realize this goal remains an open research question. Research in this area has been dubbed "non-photorealistic rendering" (NPR).

The goal of this seminar is to cover research to date in NPR, and to give students the opportunity to explore beyond the cutting edge. To this end, seminar will have two components -- papers and projects. These are described in more detail in the:

Syllabus

More resources

Craig Reynold's page on the subject. An excellent overview of NPR resources on the web.
A list of recent NPR papers cached at University of Utah.
An online museum organized by Nicolas Pioch.

Administrative Information

This course is open to all students (grad and undergrad) who have taken computer graphics (EECS 487 or equivalent), or by permission.

This course has been approved to satisfy the general 500-level requirements for the M.Sc. and Ph.D.

Lee Markosian.