A Framework for Non-Realistic Projections
Author(s): Jonathan Levene.
Master Thesis: Massachusetts Institute of Technology, May,
1998.
[BibTeX]
Abstract:
Over the last thirty years, most research efforts in computer graphics have been
directed towards producing photorealistic images. Photorealism, however, exhibits
some very real shortcomings in practical and artistic settings. Non-photorealistic
renderers (NPRs) offer new solutions by abandoning the accurate modeling of optics
in order to achieve more expressive results. Previous NPR systems have focused on
simulating traditional media by displaying geometric results in a non-realistic fashion.
By abandoning photorealism, however, NPR systems also have the option of perform-
ing projection, in addition to lighting and visibility resolution, non-realistically. Projection techniques are particularly useful as a means of controlling how information is presented to a viewer, expressing the various ways in which the shape of objects, and the spatial relations between them, can be represented in pictures.
We describe a framework for interactively computing non-realistic projections from
3D world space to 2D screen space. The framework provides a means of (1) using
curved projection surfaces; (2) controlling the degree to which orthogonals converge
to or diverge from a vanishing point in the image; (3) controlling the behavior of
orthogonals as they converge or diverge; and (4) projecting different objects indepen-
dently and compositing their images together. We demonstrate our approach with a
variety of expressive projections applied to complex models.