Browse Books

Reviewer: Don Herbison-Evans

Whereas the human retina communicates to the brain via the optic nerve at a rate of about 1M baud, humans can only read text at about 1K baud. Computer graphics is the art of drawing pictures with a computer for better communication with a human user and trying to capitalize on this three-orders-of-magnitude discrepancy. Computer graphics is demanding of all computer resources, but as computer power has become cheaper, it is economic to use some of that power in order to provide the better user interaction that pictorial computer output provides. The data structures and algorithms employed in the process have a large impact on the cost and viability of the graphical interface. Thus, university courses on these techniques are a common part of computer science syllabi. There are approximately two dozen different books on computer graphics. Many are aimed at the home hobbyists, some at the users of personal computers; some are aimed at artists, some at engineers and mathematicians. A few are appropriate for a tertiary-level course on computer graphics within a computer science course. Four examples of this last type of book are reviewed below. All four books have abundant diagrams, photographs, coded program segments, sample questions, and a bibliography. All are excellent for both tutorial and reference purposes. The major topics covered by texts such as these include the following: :9Bintroduction (e.g., history, applications), output devices (e.g., hardcopy, raster architecture), input devices, data structures (e.g., segments, polyhedra), interactive techniques, geometric algorithms, transformations (2D and 3D), hidden line and surface removal, and shading (including color, ray tracing). An important factor is whether a particular book defines its own graphics primitives, or whether it uses those of the ISO standard: the Graphical Kernel System (GKS). This is becoming increasingly important as more hardware and software suppliers offer the GKS interface to their systems. Newman and Sproull This is the oldest book, which is a second and greatly revised edition of a book first published in 1973. Readers in the offshore colonies can obtain a paperback International Student Edition of the book at a markedly reduced price ($8.25), but it is not available on mainland North America. Newman and Sproull is the only one of the four books to include a movie (a rotating skeleton of a molecule homologous with 3,4 di-para-ethyl-phenyl-n-hexane, printed on the corners of pp. 293–409). The age of the book is betrayed by its early chapter on line drawing displays and the code in the procedure “Transmit Coords” on p. 89, which produces graphic commands for the Tektronic 4010 series of displays. Nevertheless, five out of the 28 chapters are on raster graphics, and point plotting is introduced before line plotting. It covers well all the topics described above. However, its publication date (1979) puts limits on the utility of the references, and no mention is made of GKS. Foley and Van Dam The largest book, it is competitively priced compared with the hardcover Newman and Sproull, covering much the same material, but in greater depth. This book mentions the GKS standard (six times), but it does not include even such key items as “bundles” or “polylines” in the index. Nevertheless, it has excellent explanations of the use of hierarchical graphic data structures, and also the design of user computer-graphic conversations. A feature of the book is the continuous interweaving of the various topics listed above, giving students an integrated development of computer graphics. The lack of orthogonality between chapters makes it a disconcerting book to teach from, but the student maintains a useful, coherent understanding of the whole field at every stage. One small point is the use of capitals for procedure names in the code segments. This may be comforting to old FORTRAN programmers (like the reviewer), but it makes the code harder for the eye to run over. Rogers This is a newer book, devoted mainly to geometric software. It has only five chapters: (1) Introduction, (2) raster scan graphics (including line and circle drawing, various filling algorithms, and antialiasing), (3) clipping (2D and 3D), (4) hidden line and surface removal (in particular, Roberts, Warnock, and Weiler-Atherton), and (5) rendering (shading, transparency, shadows, and texture). Unlike the other books, it assumes familiarity with vectors, matrices, and perspective. However, it has nearly a page on fractals, whereas Foley and Van Dam has only a paragraph, and Newman and Sproull has no mention of them. While the other two books have their code segments in Pascal, Rogers uses a pseudocode (which is like Pascal). Hearn and Baker The most recent book uses GKS throughout. Indeed, it even has a second index specifically on GKS keywords. It covers all the topics listed above in a straightforward manner. It also includes some more modern material, such as 3D displays, octrees, constructive solid geometry, and eight pages on fractals. The book uses color a great deal not only in the photographs, but also in the diagrams, giving them extra clarity. A curious exception is the set of pages on 295 to 308, in the chapter on Shading and Color Models, where color in the diagrams and photographs would have been very beneficial. Conclusions There are a number of limitations to all the books. They all have a very similar style of writing. While the style is not overly academic, it is not particularly exciting. There is room for more distinction in each between a reference book and a tutorial text. There are also a number of topics which are of importance but are barely treated, if at all, in these four books: e.g., electroluminescent displays, memory contention, animation, quaternions. The books are good, and are excellent texts, but coupling the omissions with the rapid pace of computer graphics research, there is clearly room for improvement.