This is an HTML rendering of a working paper draft that led to a publication. The publication should always be cited in preference to this draft using the following reference:
Athens University of Economics and Business
Thomas Strothotte and Stefan Schlechtweg
Non-Photorealistic Computer Graphics
Morgan Kaufmann, San Francisco, CA, USA, 2002
470 pp. ISBN 1-55860-787-0
Computer graphics has from its inception been dominated by a quest for realism: the generation of synthetic pictures that would resemble as much as possible the appearance of real images. The painstaking modeling of the physical illumination processes has resulted in fascinating synthetic images that are often indistinguishable from reality. Few human-made images are however drawn with this target in mind. Paintings, drawings, and scientific illustrations are foremost communication tools, made to be beautiful, give rise to emotions, or efficiently convey information. Cognitive sciences, art, and graphic design play an important role in the creation of the above images. Non-photorealistic rendering (NPR) uses processes that explicitly distance themselves from the goal of photographic accuracy to generate images that better communicate specific information. The applications are diverse and range from engineering and medical illustrations, to computer-generated captions, to cartoon images.
The work by Strothotte and Schlechtweg seeks to bring under a single roof research results and practical experience that have accumulated in the area of NPR from the work of computer graphics scientists in the last 20 years. It is an intellectually stimulating and thought-provoking book that provides a scientific framework for a rapidly evolving and challenging area. The material could be used as a textbook in a graduate graphics course, as well as a guide for practitioners, and researchers in the field. Keep however in mind that the different types of techniques presented in the book require a wide breadth of background knowledge in diverse areas such as signal processing, computational geometry, analysis, and discrete mathematics.
The techniques used for NPR can be broadly distinguished between those modeling image artifacts that result from the manner or style a geometric model is rendered, and model artifacts that result from the way the geometric model represents an original object. Different NPR methods seek to simulate pen-and-ink drawings, pencil sketches, painterly effects, diverse natural media, artistic techniques, deformation, and lightning. Many of the above effects are intelligently deconstructed into the processes that result in them; the description of how traditional Chinese Suibokuga paintings can be modeled using cellular automata is a tour de force in analytical thinking. Unfortunately sometimes the analysis stops short of providing the corresponding pseudocode.
The book is divided into chapters covering the pixel manipulation of images, the generation of lines, curves, and strokes, the simulation of natural media and artistic techniques, stroke-based illustrations, data structures for modeling distance, geometric and lighting models, distortions, and applications. The applications of NPR easily convey the field's importance. The ones presented in the book include the representation of motion in still images (think of speed lines in cartoons), architectural illustrations, rendering plants, illustrating medical and technical texts, and tactile rendering for blind people.
The book's last chapter contributed by Kees van Overveld and modestly titled "A Conceptual Framework for NPR" manages to provide, in an admittedly dense writing style, a single conceptual framework that, going beyond NPR, covers in a formal manner human perception and cognition. Resembling in the style of treatment Michael Arbib's classic Brains, Machines, and Mathematics (second edition, Springer Verlag 1987) the chapter discusses how human perception, communication, and vehicles of scientific discourse delicately balance the sometimes conflicting notions of truthfulness, comprehensibility, and relevance.