There was a nice article in the Baltimore Sun on the state of Maryland’s games industry.
Word of the day: “turbulent”. Second word of the day: “indie”
Game Development at the University of Maryland, Baltimore County
There was a nice article in the Baltimore Sun on the state of Maryland’s games industry.
Word of the day: “turbulent”. Second word of the day: “indie”
According to a story on the Washington Post web site December 17th, a study of game sales per capita and gun-related murders per capita in 10 countries shows no real correlation. At the end, there’s a graph showing a linear fit through the data, showing a weak negative correlation, but as scattered as the data is, even drawing that line makes no sense to me. None the less, it certainly doesn’t look like correlation (much less causation) between the two.
Here’s a post that’s interesting for both programmers and artists. Sébastian LeGarde is an game engine and graphics programmer at Dontnod entertainment in Paris. He has a blog with a variety of good posts on the programming side of achieving realistic physically-based illumination. He’s just started a series on rendering rainy and wet scenes, starting with a collection of photos and observations showing the different ways rain actually appears in the real world. Whether you are an artist or programmer, when you’re trying to get something new, it’s always a good idea to go out and observe the world. You can’t get it right, if you don’t know what it is supposed to look like. Check it out!
Another linked article, this one about a set of foundation libraries released by bitsquid under the MIT license, the Bitsquid Foundation Library
Some observations…
Memory is often as critical a resource as execution time. Bad memory management can result in lots of little fragments of unusable free memory and a game that crashes when trying to load the next level. In addition, many standard memory allocators serialize the memory allocation in order to be thread safe when running on multiple CPU cores. With lots of memory allocations, this can turn into a performance bottleneck. The C++ Standard Template Libraries are handy, but mostly consider memory allocation and deallocation as free operations, and are waaay more willing to do them than most game developers would prefer. Consequently, you’ll find a profusion of different memory allocators and data structures that most people think of as being part of standard C++. This is one such library.
Compile time is a big issue for major game projects. Some use Visual Studio precompiled headers. They are good for relatively unchanging headers, but any change to any file in the precompiled header will result is a sloooow rebuild of the entire project.
Some try to reduce the dependencies between header files to only include what you absolutely need, and (in C++ at least) declare by name the classes you just need for pointers or references. This reduces the number of files that will need be recompiled for a header change. I first read about this technique in Large Scale C++ Software Design, by John Lakos. That book also advocates the pimpl approach, where rather than declaring private members in the public include file, the class will just contain a pointer to an implementation class. The implementation class is declared and defined with all of the implementation code, so does not need to be part of the class include file, and implementation changes do not cause recompilation for anyone using the class. Most games that I’ve seen do not use the pimpl method, because each class (even if created on the stack) results in an extra memory allocation, and each member function has an extra pointer dereference into the implementation class.
Bitsquid uses another technique, where the public class only contains data declarations. Rather than use member functions, you use regular functions to operate on the class. The idea is that most of the header interdependencies that increase compile time are just member-to-member dependencies which only need to know the data layout. Source implementation code may need to know about the functions too, but that’s still fewer dependencies (and faster compilation).
I have not done much coding in this style, but the key takeaway here is that game developers spend a lot of time compiling, and anything that makes compilation slower is potentially hours per day of lost work.
Most game developers I know have a network of other developers and online sources they follow to keep up with game development trends and ideas. Most students haven’t developed this set of sources yet, so I plan to start posting a series of links to things I run across and find interesting. Since my own postings are admittedly graphics focused, I thought I would start with an AI article. This one, originally from Game Developer Magazine, gives a high-level comparison of AI options as used in games: AI Architectures: A Culinary Guide
© 2024 UMBC Games, Animation and Interactive Media
Theme by Anders Noren — Up ↑