Non-photorealistic rendering: a critical examination and proposed system.

In the first part of the program the emergent field of Non-Photorealistic Rendering is explored from a cultural perspective. This is to establish a clear understanding of what Non-Photorealistic Rendering (NPR) ought to be in its mature form in order to provide goals and an overall infrastructure for future development. This thesis claims that unless we understand and clarify NPR's relationship with other media (photography, photorealistic computer graphics and traditional media) we will continue to manufacture "new solutions" to computer based imaging which are confused and naive in their goals. Such solutions will be rejected by the art and design
community, generally condemned as novelties of little cultural worth ( i.e. they will not sell).
This is achieved by critically reviewing published systems that are naively described as Non-photorealistic
or "painterly" systems. Current practices and techniques are criticised in terms of their low ability to articulate meaning in images; solutions to this problem are given. A further argument claims that NPR, while being similar to traditional "natural media" techniques in certain aspects, is fundamentally different in other ways. This similarity has lead NPR to be sometimes proposed as "painting simulation" — something it can never be. Methods for avoiding this position are proposed. The similarities and differences to painting and drawing are presented and NPR's relationship to its other counterpart, Photorealistic Rendering (PR), is then delineated. It is shown that NPR is paradigmatically different to other forms of representation — i.e. it is not an "effect", but rather something basically different.
The benefits of NPR in its mature form are discussed in the context of Architectural Representation and Design in general. This is done in conjunction with consultations with designers and architects. From this consultation a "wish-list" of capabilities is compiled by way of a requirements capture for a proposed system.
A series of computer-based experiments resulting in the systems "Expressive Marks" and 'Magic Painter" are carried out; these practical experiments add further understanding to the problems of NPR. The exploration concludes with a prototype system "Piranesi" which is submitted as a good overall solution to the problem of NPR. In support of this written thesis are : -
• The Expressive Marks system
• Magic Painter system
• The Piranesi system (which includes the EPixel and Sketcher systems)
• A large portfolio of images generated throughout the exploration.

A dissertation submitted for the degree of Doctor of Philosophy.