Summary: Realism in Computer Graphics: A Survey
Dept. of Computer Science
University of Toronto
One of the most challenging problems in computer graphics is to generate images that appear realis-
tic; that is, images that can fool a human observer when displayed on a screen. The quest for this "Holy
Grail" began in earnest in the early 70's when memory prices dropped low enough to allow raster technolo-
gies to be cost-effective over the then prevailing calligraphic displays. Calligraphic displays could only
draw a limited number of lines and even the most capable of these displays allowed for only a handful of
colours. Previously, research work concentrated on removing "hidden lines" from objects drawn on these
displays. The objects displayed were obviously not realistic but contained enough information for the task
at hand, such as computer aided design. Raster technology, by subdividing the screen into pixels, allowed
whole regions of the screen to be filled with colours, colours that had a wide variety of intensities and tints.
This new technology, capable of displaying realistic images, opened up research in this direction and it is
this research that we will outline.
This paper will survey most of the major issues that one must deal with when generating realistic
images. We begin with an overview of the rendering process and a quick review of visible surface deter-
mination algorithms. We then discuss, in more detail, shading, anti-aliasing, texture mapping, shadows,
optical effects and close with a discussion of modeling primitives.