Humans have always looked for some way to temporarily escape from the world. It seems like second nature for us to seek an activity that allows us to unwind, removing the stresses that daily life places upon us and replacing that tension with a more pleasant feeling of relaxation and enjoyment. This is why it comes as no surprise that our culture has adopted a love for videogames that rivals the love of stories and theatre that helped to carry previous generations though their lives. Videogames provide, in many respects, an even greater method of escape than what is offered by stories or theatre: after all, none of those offered a person the ability to actually step into the shoes of another person, and see a completely new world almost literally though that persons eyes. Our society is certainly responding to this new escape as well: names like Mario, Luigi, and Sonic the Hedgehog have become as well known as Mickey Mouse and Goofy were to previous generations. Financially, videogames have become successful too. With $6 billion in revenue generated by videogame title sales alone in 2001, the videogame industry is quickly becoming comparable to the $7.7 billion movie industry (Savitz).
However, computer videogames have not always been this big. Just a decade ago, the industry floundered about, not really having any real sense of direction or any means of distinguishing itself from the console industry. Then, in 1991, Id Software released the game that would revolutionize computer video gaming and set it apart from consoles forever: Wolfenstein 3D. Wolfenstein 3D was the first game ever to truly put the player in the world, with an interactive environment and the first graphics that were actually three dimensional and allowed the player complete freedom of movement (Id Software). The industry suddenly had a sense of direction, and for the next five years, every developer in the business was creating three dimensional games, using the (then) brute force of Intels 486 processor to push their game engines and turn a collection of numbers and a few pictures into a world that looked more real than anything anyone had ever seen before.
If the industry had simply continued on this path, chances are it would not be anywhere close to the state it is in now, a serious contender for the heavy weight entertainment crown. In 1996, Id Software again reached into its bag of tricks and pulled out a revolution. With the release of Quake, Id took the game industry by the horns and turned it in a completely different direction (Id Software). What made this game different was its ability to run on the 3D cards that were being used to render complex scenes in movies, by making use of Silicon Graphics API (Advanced Programming Interface), OpenGL. By using the 3D card instead of the computers main processor to render the graphics, Quake was able to achieve a level of graphical quality that was unseen before. This game alone is recognized as the father of consumer level 3D hardware, as the very first customer level 3D video cards were released to cater to the market demands that Quake created. Within months, a plethora of new 3D hardware solutions flooded the market, and because of Ids revolutionary concept of programming the game for an API instead of the computer itself, every developer on the market suddenly had a much better chance of creating a game on time and under budget.
What is an API?
A 3D graphics Advanced Programming Interface, or API, is a set of programs that allows developers to take the programming code for their game or application and make it work for every potential hardware configuration on the market with enough power to run it. Without an API, developers would have to literally reprogram their game thousands of times, once for every processor, video card, and chipset combination. An API is easily understood if it is thought of as a magic box with plugs on both sides. One side plugs into a program, and the other side plugs directly into the computer. When the program sends a signal down one of the plugs into the box, the box takes that signal and turns it into something that the computer can work with. Thus, the features offered by an API, or the plugs that it will accept on both sides, can directly influence what developers can do with their games. Choosing the correct API is crucial to creating a game that both fits the developers concepts and works on enough customers computers to actually sell.
The videogame market is dominated by two API standards. The most popular is Microsofts API, DirectX. More than just a 3D graphics tool, DirectX 8.1 also includes support for just about every sound card and networking setup used in the Windows environment. The only other contender is Silicon Graphics OpenGL 2.0, which is a stand alone 3D rendering API, and does not including support for any sound or networking. However, the different facets of use for each API are not the only differences. OpenGL, despite lacking the components to handle the sound or networking requirements of videogames, is a much more flexible API than DirectX, and also allows the developers to take the games they just programmed for a computer running Microsoft Windows and re-release them for all Apple computers as well. The OpenGL API is even flexible enough to work on some console systems like Sonys Playstation2 and Segas Dreamcast, although these re-releases require massively large amounts of work and usually take a great deal of time. DirectX also has its advantages, the main one being that it is an integrated package, and one licensing fee covers it all. When all the advantages are added up and the current maneuvering of both companies is considered, it becomes apparent that although OpenGL 2.0 is the superior new 3D graphics API, popular developer support and the powerhouse backing of Microsoft will make DirectX 8.1 the dominating standard in the years to come.
Videogame Development
The development of videogames is a long and expensive process. First, a game concept must be conceived. This concept is taken to focus groups and goes though analysis before any computer programming starts. Assuming the project is well received and a publisher is found, the development of the game starts with one difficult challenge: how to turn these ideas into reality. The most cost effective way of creating a game out of an idea is to license a game engine from another development company, in effect to take someone elses game and change things to fit what a developer wants it to do. The longer, more expensive, but generally more fruitful method of creating a game is to design a graphics engine from the ground up; however, regardless of the chosen path, the developers must always decide if they want to use DirectX or OpenGL as the base plate to build their game.
The choice of API can have effects on what developers can do with their games. OpenGL, although harder to work with, is an extremely flexible API that works on just about every system with every video card under every operating environment. OpenGL was originally designed for Hollywood studios to use on their special effects rendering farms, and thus was designed with the ability to do anything in mind. DirectX, on the other hand, is relatively limited in what it can do. Although some developers may argue that the single minded nature of DirectXs programming makes it better tuned for games (Wilson), it is a completely closed box systemthat is if developers are programming for DirectX, the developers have no idea what is under the hood. All the developers know is what will come out when a line of code is sent in. With OpenGL, programmers can go into the API and add their own features, tweak a feature if they dont like the way a specific function is handled, and generally make it do everything and anything the game requires the API to do, regardless of whether or not it is supported by the base specifications of OpenGL.
Flexibility Matters
The difference between DirectX and OpenGL boils down to a difference in philosophy. OpenGL follows the idiom that the code behind the program should be available and modifiable, the idea being that if users all add the functions they need to the software and then distribute it to others, the end result will be more feature filled and robust than anything a group of corporate programmers would be able to create. DirectX, on the other hand, follows the white box programming philosophy, where the inner workings of the program are unknown to its users; the users only know what comes out when they send something in.
The flexibility of OpenGL allows creative programmers to do things that no one else has ever even thought of doing before, and not have to wait for Microsofts programmers to catch up, but rather be able to use the new idea immediately. For example, the current method of rendering an image does a great job of creating something that looks three dimensional and in perspective; but what if the developer does not want the graphics to be three dimensional, desiring instead for them to look more like an animated character one would see in a cartoon show? The normal method of accomplishing this would be to hire an animator to draw all the characters needed in the game from a few different angles, and to load the image of the character at the angle closest to the one the player is viewing it from. This method is not very good, as it would take literally 1080 (360 degrees of rotation and 3 axis of movement) times the number of frames of animation needed for every action in order to make the character seamlessly integrated into the world. Last year, however, using OpenGL, a programmer named Raskar Ramesh figured out a way to create cartoonish or cell animated characters using three dimensional rendering and a few tricks (Ramesh 410). Using this new method, games utilizing the cell shading graphical style that once would have been held back for more than a year for artwork creation can be released much sooner and with a much smaller budget.
The method of rendering these cell shaded characters is both innovative and elegant. By having the API automatically attach geometry to the outer edge if the model being cartoonized, and using solid colored textures with real time environmental shading, the model can be changed from something that looks three dimensional to something that looks like it was drawn by hand and scanned into the game. However, the model can be viewed from any angle and takes hundreds less man hours to create than the old method would require (Ramesh 412). Because OpenGL was flexible enough to handle this method of rendering from the moment it was conceived, video games featuring real time rendered cell shading have already been released under OpenGL, while Microsoft just now is getting this rendering ability on the features list for the new DirectX 9.0, due out late this summer.
Most creative and unusual video games that are released run under OpenGL. A game released late last year, Red Faction, was the first game to allow the player to literally blow walls out of the levels, destroy buildings, and walk though the dynamically created rubble. However, with the ability for the users to alter level geometry, an old game development issue again reared its ugly head. One of the problems earlier developers had was the adverse effect repetitive textures had on the realistic appearance of game levels. Previously developers simply used great caution and a certain amount of strategy in planning out the levels in their games so that there would be no long flat stretches textured with the same file, and that level geometry would break up the monotony and make the patterns harder to recognize. This issue popped back up when users blow out terrain and noticed that the new geometry had a very unrealistic uniform appearance. This issue plagued developers until another programmer, John Hart, came up with an idea that could solve this issue once and for all. Again with a little ingenuity and the flexibility of OpenGL, Hart was able to come up with a new method for texturing that would assure no repetition at all: the spontaneous creation of completely randomized textures.
Harts method utilizes the random number generator in a computer to generate a set of random numbers, which is then transformed into a texture fitting the general specifications of the program requesting it (Hart 89). If developers use this method, instead of specifying a pre-generated texture for each piece of geometry in a game level, they will simply set what approximate color the texture should be, and the random texture generator would create something totally new every time the level was loaded. Detail oriented textures would still need to be hard-set and hand-placed by developers, but this method could make a big difference in both the realism of a scene, and also the amount of time it takes to create a level; as the old methods required creation by hand of numerous textures and painstaking, time consuming placement of those textures. Using this method, time to market and cost of production for any project can be decreased, and the quality of the graphics in the game only gets better. The flexibility of OpenGL makes this new process available to OpenGL based developing firms already, while developers using DirectX will have to wait until Microsoft gets around to adding it to the feature list.
This flexibility of use is one of the main reasons that OpenGL is a superior choice for developers over DirectX. The truth of the matter is that because technology changes so fast, any technological advances made in one API will be mimicked by the other API on the next release. However, the design philosophy behind the two APIs does not change. The ability for programmers to take OpenGL and make it do exactly what they need it to do and add in new features just conceived and not yet supported by either of the APIs available is unparalleled by DirectX. This distinct advantage gives OpenGL a very real reason to be considered the technologically superior API.
As clear as night and day
Lighting and shadows, just as in Hollywood movies and photography, add a great deal of mood and dimension to the appearance of any scene in a video game. One of the hot spots of videogame development at this moment is the use of real time lighting and shadows, the ability to have the lighting on a level altered and effected by the movement of the models and other objects present in the level. Games of the past had serious limitations to the type of environmental lighting that could be shown to the user: it would have to be defined when the game level was created and could not be changed at all after the level was completed. All moving objects inside the level were unable to cast a realistic shadow. Both DirectX and OpenGL are capable of doing much better than this, in fact they are both capable of calculating out the path light will take around an object, a process known as ray tracing, and thus calculating where a shadow will fall given the position of both the light source and the object in question. The current method for creating realistic shadows is known as volumetric lighting and shadowing, where the model that is being shadowed actually has geometry attached to it with the outer edges defined by the formulas returned by the ray tracing algorithms. However, this is where the similarities between the two APIs methods end. In order for DirectX to create the extra geometry, it must pass the object though the geometry unit on the video card between three and five times, depending on the complexity of the object and the number of lights being cast upon it (Microsoft Developer Network). OpenGL, on the other hand, simply loads this task onto the processor, allowing it to ray trace the shadow and then passes the formulas back to the video card for rendering (Segal and Akeley 43). OpenGLs method uses much less system time and has a lower impact on the number of frames per second than DirectXs brute force method.
The other method commonly used to generate shadows in video game engines is the manipulation of the stencil buffer. The stencil buffer is an area in the memory of the video card that stores a simplified picture of what the user is seeing at the momentsort of an outline of all the objects on the screen. One innovative technique for shadowing involves taking the stencil buffer and locating the object to be shaded on it, then this data, basically the outline of the object from the users point of view, is skewed and stretched to match a single ray trace from the light flowing thought he center of the object to the geometry behind which is having shadows cast upon it. This data is then fed back though the video card and rendered into the next frame as a shadow on the wall (Segal and Akeley 38). Although this process might sound more complicated, the incredibly complex and time consuming methodology for ray tracing is only utilized once per light source per object with this technique, which makes it much easier on a computer than volumetric shadowing. Here, again, OpenGL has a distinct advantage, offering support for much more detail in the stencil buffer than DirectX. In DirectX, the outline of an object can only be outline or blank space (Microsoft Developer Network); in contrast, OpenGL supports soft edges, allowing the shadows cast by furry or other organic objects which dont generate a hard shadow outline to be soft and realistic in appearance (Segal and Akeley 40). This advantage is another example of how OpenGL is technologically superior to DirectX.
Hardware matters too
There is yet another aspect to the battle between DirectX and OpenGL that needs to be discussed: the battle for hardware support. It does not matter how many new features an API supports if there is no hardware on the market to support the new features! This is why both SGI and Microsoft spend lots of money and effort on wooing the big names in video hardware: nVidia, ATI, and Matrox. Up until last year, nVidia was one of OpenGLs best supporters, faithfully supporting all new features included on the latest releases of OpenGL before the other manufacturers even knew they existed and working in close partnership with SGIs engineers on new possibilities. However, two years ago, nVidia signed a deal with Microsoft to create the 3D brain for their new entertainment console, the X-Box. Since entering into this partnership, one can only wonder what other deals Microsoft has been offering nVidia in exchange for cooperation, and at the same time wonder how long nVidia will actively support OpenGL above the level needed to assure their products are compatible enough to sell.
In fact, nVidias latest generation of video cards are very blatantly designed to work best with new DirectX games. With the release of DirectX 8.0 at the end of 2000, Microsoft created a new method for rendering games in which the models and geometry are not defined by triangles, but rather as a collection of vertices. This new method of rendering, known as Vertex Based Rendering, is best executed on a current generation nVidia card with the patented User Programmable Vertex Engine (Lindholm, Kilgard, and Moreton 149). The Vertex Engine was designed from the ground up to work seamlessly with DirectX. The engine is designed to take the entire load of this mathematically challenging method of rendering off the processor and shoulder it entirely on the video card (Lindholm, Kilgard, and Moreton 153). Speed tests and benchmarks have shown that using one of the new nVidia video cards over an older model makes little difference for the speed at which OpenGL games are rendered, but makes a huge positive difference for the speed that DirectX games are rendered: clearly indicating an engineering bias at nVidia towards Microsoft over SGI (Bell). Rather than refining their rendering techniques above what was previously marketed for OpenGL, nVidia has spent their time working with Microsoft. If hardware support for OpenGL continues to fall behind that of DirectX, it will not be long before a game programmed for OpenGL is no longer marketable, since no consumer would purchase a game that they cannot run on their computer at home. Microsoft has SGI right where it wants it.
Two is better than one
Microsoft and nVidias X-Box also holds some more significance to the battle between DirectX and OpenGL. The X-Box actually runs on DirectX as an operating system, meaning that any game developed for DirectX on the computer can easily and very cheaply be reprogrammed to work on the X-Box. Given this advantage over OpenGL, it is no small wonder that the majority of the big video games released in the past year used the DirectX API, and were released for both the X-Box and the PC either in tandem or one right after the other. In effect, the X-Box makes every dollar spent on the development of a game for DirectX possibly worth twice as much on the return, since the possible audience for the game almost doubles when one takes into account the huge number of X-Boxes already installed in consumers homes. It appears that Microsoft has an incredible advantage here, and is poised to take the same monopolistic control of the video game market that it currently has over the Personal Computer market.
What does the future hold?
With the future of the videogame industry looking bright from a financial standpoint, many wonder who exactly will reap the most benefits from its large and growing revenue. Given the current situation and the strangle hold Microsoft is poised to place over both the PC and console video game markets, it seems very reasonable to predict that within the next couple of years, DirectX will become the overwhelmingly dominating standard and OpenGL will fade into the limelightreverting to its earlier state of being used only for research and development for Hollywoods special effects studios. Once Microsofts control over the market is complete, they can easily raise licensing costs and demand higher royalties for all games that use the DirectX API, allowing Microsoft to control and direct the video game market as the company sees fit. After all, in its heart any industry is more about making money than any other goal, and if using DirectX will guarantee the largest profits available for the time being, there is no logical reason for a business minded company make any other choice for an API. But, like all things, the future is yet uncertain. Perhaps something else will come along that will keep competition alive and assure a healthy distribution of control and revenue to all parties involved in the industry, but only time will tell.
quote:
Ferret obviously shouldn't have said:
That comes out to thirteen pages?
With Double spaced 12 point times new roman font, its 12 and a half pages actually.
Nice bootleg! Do you mind? 
quote:
When the babel fish was in place, it was apparent Tier the Genius said:
::steals Blindy's paper::Nice bootleg! Do you mind?
it's not too useful without that biblography to go along with it.
quote:
This one time, at Blind Swordsman camp:
it's not too useful without that biblography to go along with it.
I can make it up. Teachers are annoyed enough at having to read dozens of paper, they don't check the references! 
Seriously though, well written. But correct me if I'm wrong, but didn't Descent make use of 3d hardware before Quake did?
