Customer Reviews

Geometric Tools for Computer Graphics (The Morgan Kaufmann Series in Computer Graphics)

5 star:		(4)
4 star:		(2)
3 star:		(3)
2 star:		(1)
1 star:		(0)

 

 

First of all, I commend the authors on a timely and valuable book. However, a word of caution: this book is full of errors. Every couple of pages I am noting in the margin: did they mean A instead of B? Having encountered so many errors, I am reading every formula with scepticism. The errors are serious enough that I have trouble recommending the book without reservations, but I know of no suitable alternative. I can only hope that the errors will be weeded out of future editions.

First, the good:
-Once you learn everything in this book, you'll should be ~well~ prepared to start implementing a powerful 3D engine.
-The three-chapter introduction to linear algebra is, quite frankly, one of the most intuitive I've ever read. Mind you, I wouldn't suggest that anyone studying linear algebra go out of their way to buy this book (since those three chapters are a small fraction of the book). However, anyone getting their first taste of linear algebra from this book should consider themselves fortunate to have such a lucid, concrete introduction to the subject. (Granted, you'll need some 'mathematical maturity' to understand it, but it should be easier to grasp than your average linear algebra text.)

And then, the bad:
-Errors galore. Fortunately, you can get a list of corrections from the book's web site, though if you print it out you may be a bit put off by the fact that it's some 25 pages. (To be fair, however, it has all corrections listed chronologically in order they were identified, then listed again by page number, so there's really only about 12 pages of corrections.) If you happen to have the second edition of this book, then you'll only have about 5 pages of corrections.
-Some of those corrections pretty much just scrap an explanation from the book and start over....which is fine, aside from having to read things like "Cross(Dot(u,v) * w))" which isn't particularly intuitive, and the fact that some of these new explanations seem to need corrections of their own (like those that appear to confuse w-parallel with w-perp, and so on).
-It seems a bit arrogant of the authors to make the occasional appeal to things that the "astute reader" may have noticed. Such appeals seem like a subtle insult to the "less than astute reader," which, in any public forum, will only serve to alienate.


All in all, if you're willing to put up with errors and have your "astuteness" challenged, you can learn tremendous things from this book.

This book provides a solid grounding in vectors and matrices, then covers a wide range of 2D and 3D geometric algorithms, such as object/object distance and intersection, boolean operations, BSP trees, convex hulls, and more. It is a comprehensive guide, giving relevant theory, methods, and working code fragments. It's an incredible value for the amount of material it covers. I think it is a must-have for computer graphics professionals (and others in related fields). In the interest of full disclosure, I wrote the Foreword to this book, which I did because I was pleased with how good it is.

Visit Dave Eberly's "Magic Software" site for more about the book's contents.

This is the first book I found that does an incredible job of covering vector geometry from a coordinate free approach. The first 1/4 of the book covers the basics with an excellent mathematical approach. The rest of the book show excellent examples of just about any type of intersection and collision of geometry; OBB, sphere, cones, polygons etc.... The is great for doing things like Frustum culling and the like. The only weird thing was it is missing Eberly's discussion on sphere/cone intersection; but no matter you can get it at the website.

I highly recommend this book for those that want to understand the core of 3D graphics from a coordinate free approach. I am very happy I purchased the book. It has inspired me to purchase a clifford algebra book to better understand coord-free algebra.

If you are simply looking for code and are not interested in the mathematical reasoning then you problably should look elsewhere. This book is for those who want to get a better understanding of core 3D graphics from a very friendly approach.

I also noticed that those who rated this book with few stars where simply looking for something quick. Face it 3D graphics and math go hand in hand. Otherwise you are just kidding yourself.

I just got this book about a week ago and I haven't been able to put it down since. A great resource, a real treasure full of well presented gems. A large number of topics are well presented with mathematical depth that enables you to understand the code. The code is very well written and concise. Truly a great book and a pleasure to read. I hope to see more books like this in the future. I am interested in medical imaging applications and I know I will be using this book for a long time to come. In addition, the authors have a site ... that presents the code and additional very valuable materials.

I have owned this text for some time and I find it very informative as it covers a great deal of subjects. A review of the table of contents will give an indication of the material covered. The book is not only valuable to graphics programmers but engineers in other disciplines looking for a good approach for solving various geometric problems.

A previous reviewer complained about the notation being "non-standard". The book was written for someone with a mathematical background to include a sound background in vector algebra, geometry and matrix operations. The notation used is in explaining the mathematics behind the solution to a problem is standard >mathematical< notation. If you don't have a background in these areas of mathematics, then you may have a problem understanding it. But fortunately, there are many inexpensive books on those subjects available for purchase that can get you quickly up to speed.

The code is written in the C language. Typically, code is provided only for a specific algorithm (problem being solved). Complete applications that give examples of using the algorithm implementation aren't provided as this is beyond the scope of the book. The scope of the book is to teach you how to solve specific mathematical problems of interest. Not to teach you the many different ways that mathematical solution may be employed in all genres of programming.

The reason that I failed to give this book a 5-star rating is due to the many errors in the text. There was an impressive (to put it kindly) list of errata published on line for the initial printing. The 2nd printing of the text (and how do you know on Amazon if it is a 2nd printing?) is supposed to have most of the errors corrected. However, since the 2nd printing, errors continue to be reported. A complete list of the errata for this book is available at the web site [...].

Having experience in writing many complex technical works, I can say that it takes great dilligence and peer review to capture errors in an intense tecnicaly work such as this book. While this book seems to have excessive errors in it, this type of problem is common with most publishers. There simply isn't enough effort/expense put into having a sufficient number of qualified technical people to review the work and look for errors overlooked by the authors. And believe me, quality peer reviews are necessary for complex technical works such as this one with mathematics on virtually every page. So I do agree with a previous reviewer that you should be cautious at taking the solution/implementation of a mathematical problem from this book at "face value" without questioning if it is correct for all possible test cases. Test and verify the solution.

Given the above comments, I would still recommend this text as it covers so many different topics and problems encountered in 2D and 3D geometry. This book is valuable to many engineers other that programmers working in graphics or game development.

But I would like to see a re-print that has been 100% thoroughly peer reviewed by **mathematicians**, the algorithm implementations analyzed for correctness and an error-free copy printed.

Many of Mr. Eberly's books leave me dazed and confused. His Game Physics book, though quite useful, is so wedded to the Wild Magic framework that I felt like learning that framework became the task at hand rather than trying to learn underlying algorithms.

This book is different, because it is organized as a collection of tools. Each tool is pretty much independent of all the others, so you can see via the figures what Mr. Eberly is trying to accomplish, read the accompanying text and equations, and then read his pseudocode to understand what you need to do in whatever programming language you are trying to do it in. You can, in most cases, just lift out the algorithm/tool you need. This is the beauty of the book.

The first three or so chapters are dedicated to giving you a quick brush-up in the underlying math, primarily linear algebra. They are useful if you just need to remember something you have already learned at some point, but it is not detailed enough to teach you from scratch.

In short, this is an excellent book on the algorithms needed for the implementation of computer graphics tasks in both two and three dimensions if you already have a good big-picture understanding of computer graphics and a detailed understanding of the mathematics commonly used in such tasks.

This book is used to implement geometric algorithms. The authors present both coordinate-free (which they favour) and coordinate-based approaches to geometry, and there is plenty of intuitive motivation.

While much of the book is reference material, (look at the table of contents) it is written at the right level of detail, and chapters 2-4 provide an excellent introduction. It does not assume much more than patience from the reader, and provides an extensive list of recommended books. Some reviewers have complained of the book's difficulty. I would suggest the difficulty is not with the presentation but the content.

Breadth has made error inevitable, but there is a currently updated (as of 2007-12-16) errata listing at the book's website.

I almost gave this book 2 stars but decided that punishing the authors for my stupidity wasn't fair. This book is obviously chucked full of just about every geometric formula you will ever need in computer graphics. But you had better be a darn serious mathematician! I could simply never grasp mathematics at this level. I could never look at those formulas, written in some alien heiroglyphs, and understand them enough to translate them into code. This book is one of those "Written by Professors for Professors." I bought the book because of the reviews that wrote about the code samples. Well, they are there (pseudo-code, but that is better than nothing) but nowhere near what I was expecting. My mistake was thinking it was a programming book. It's not! Bottom line, if you are looking for computer graphics formulas and understand mathematics, this book is for you. If you are a programmer and are expecting a computer graphics library of functions, look somewhere else.

This book has the potential to be useful to computer programmers (experienced practitioners as they call them), unfortunately I don't think it might be as useful as it could be. If the reader is mathematically inclined then the book is not bad but for a programmer who wants quick solutions I am not sure it is the right book. The authors also use a notation that might be unfamiliar to most programmers making it more difficult to read. The code samples are poor (of which there are only a few) and the algorithms they describe are not practical and it would take a lot more work to implement actual working code. I wish there were a geometry book that could combine theory and practical tips and hints plus real code. I know how difficult it is to write a book and the authors are to be commended on the intention and scope of material they cover. The theory sections are quite good and would allow someone to gain a deeper understanding of the underlining approaches. However I think it would do the community a great service if the authors could publish a second edition and perhaps take a cue from the Numerical Recipes series by Press et al. as an example of a book that combines practice and theory, although their recent C++ editions is not quite as good.