I've been playing video games for a long time, and I mean a long time. The first such game I ever played was the original Space Invaders. It's one of the rare, pleasant memories I have of childhood, in fact, my father and I sinking way too many quarters into that thing at a hotel during one of our few family vacations. He and I both loved that game and never liked any of its successors nearly as much. The original was a classic and remains so even today as far as I'm concerned.
From there I progressed to the inevitable Pong console that a friend owned, my own Atari 2600, and eventually to a handful of games for my TI-99/4A personal computer—does anyone else remember Tunnels of Doom as fondly as I do? On the PC I was an active gamer with the "Gold Box" series of AD&D adventures, became still more active with the Id Software revolution of Wolfenstein 3D, DOOM, etc., and finally turned into a full blown fanatic about the time the original X-COM: UFO Defense was released. Starcraft simply sealed my fate; I've been an addict ever since.
Having dated myself rather unflatteringly with that retrospective, I think game development has reached a significant turning point in light of its prior history. A very large portion of the effort expended to develop games was previously aimed at making images on the screen, at making even the most basic audio play back properly, at providing the most rudimentary sort of interface capabilities for the gamer, at just getting enemies to walk toward the player and shoot without killing themselves or doing something obviously absurd. Technology has come so far that there is a fundamental shift occurring in the way game worlds are modeled, a shift that will not merely change how games are developed but how they're played as well.
First of all, I should divide games along two lines: (1) those that try to model some sort of relatively consistent, familiar world, and (2) everything else. The former category includes games such as Battlefield 1942, BloodRayne, Baldur's Gate, Command & Conquer: Generals, Diablo II, etc. Despite how radically divergent these games are in their settings, content, and faithfulness to the real world, they are all nevertheless set in worlds with gravity, land, sky, water, and so forth. This distinguishes them as a class of games from other favorites like Risk, Tetris, etc. The turning point of which I speak refers only to games in the former category not the latter.
The reason for this is because all of those games, to one degree or another, are merely static and roughly skin deep in terms of the degree to which they model a world. As big a step forward as the original Quake was, what with its true three dimensional world instead of the so-called "two and a half dimensional" world of the DOOM series, it still didn't even begin to reach beyond the merely static and the roughly skin deep in its approach to modeling a world.
To put it differently, all the great lighting effects were just that, effects; i.e., they were the result of applying pre-calculated lightmaps to textures in the game. Similarly, though any given surface might look more or less like wood, it was painfully obvious that it wasn't wood. At least, I don't know many wooden beams that can survive getting hit by rockets; the structures in the game could easily outlast one's patience and ammunition.
To be clear, this isn't a complaint against Quake, DOOM, or any other game. Such games pushed the technology of their times right to the limit. The greatness and successes of all such games are the very reasons we today enjoy such an embarrassment of riches in contemporary games. They pushed the envelope, and they pushed it hard. Capitalism made sure that the winners were rewarded and, thus, that hardware and software only continued to grow more powerful to keep wow-ing the paying customer.
The pushing is still going on today, but its direction is changing is shifting. Whereas the bulk of the effort spent developing games in the past was always focused on faking lighting, faking sound, faking destructible environments, and so forth, the future of game development will be focused squarely on using true lighting, genuine sound, and treating things as things in and of themselves, not merely as an impenetrable bunch of polygonal surfaces. This is the difference between the past and the present, between the static and skin deep games of old and the games to come.
Games like Far Cry (FC), DOOM 3 (D3), and Half-Life 2 (HL2) all demonstrate this in varying degrees. For example, FC is the first game I've played with substantial use of per-pixel, dynamic lighting. As I said in my review of the game:
I couldn't believe what I was seeing the first time I accidentally shot an overhead light fixture in a darkened building. Shadows played crazily about the room in real time as the light swung back and forth. I'm accustomed to lightmaps and other such trickery, which provide the static illusion of proper lighting; it's fair to say that the future is here today with FC, insofar as it does some amazing things with real-time, genuinely dynamic lighting.
That's a huge step beyond other games that still use lightmaps, one that has to be seen to be truly appreciated. Lightmaps allow only the illusion of light and shadow. Various games have also dabbled with stencil buffers, and other such techniques, to provide some real-time shadowing effects, but these too have always been tricks of one sort or another. Technology is finally progressing to the point at which most of this sleight of hand will no longer be necessary. Game developers will soon have the freedom to light scenes in very complex and largely arbitrary ways, without having to spend years developing a custom engine to do it.
Those who have seen the first demonstration of the third iteration of the Unreal Engine will understand what I'm driving at. I'm talking about the development of gaming worlds with proper dynamic lighting, with penumbra effects, true light bloom, and characters (and everything else in the game) being self-shadowing as things are in the real world. In other words, we are now entering a new era in game development, one in which extremely realistic lighting will be largely taken for granted. FC is the first game I've played that takes big steps in that direction, and D3 should do even more along those lines.
But this is only one half of the major shift that's occurring. All that messing around with light and shadow may arguably take a back seat to the more fundamental shift: a move away from skin deep, polygonal surfaces toward substantial entities in the gaming world. I have been waiting for this shift for some time, wondering when hardware and software technology would allow it to occur. Seeing the HL2 demonstration at the E3 show back in 2003 convinced me that the turning point was coming soon; the release of FC along with the coming releases of D3/HL2 makes me believe that we've finally hit it. As I said of HL2 then:
More impressive than the facial technology, to me at least, was the conceptual shift in approach in handling objects in the world, and I think this is arguably going to be HL2's most important contribution to gaming. For the most part, game worlds treat objects as if they're made of polygons, which admittedly they are. HL2, in contrast, treats objects like they are real objects. That is, most things in the world are treated as substances that exemplify properties, to be philosophical about it, and those properties determine the way that the substances are affected by the laws that govern the world. To sum it up briefly, stuff acts like it should.
This may not seem like a big deal, but trust me when I say that it is. No longer will that wooden surface be nothing more than a texture-mapped bunch of polygons. Oh, it will still be rendered in that fashion, no doubt, but the game engine will treat it not as a hollow surface but rather as a solid object. The difference is the difference between illusion and reality, the difference between a mere facade and a substantial entity in the game world.
That's not to say that fully destructible, perfect worlds are going to show up on my desktop next week, next month, or even next year. What it is to say is that game developers will be forced to develop their content much differently going forward, just as players will be forced to approach said content differently while playing. The computing horsepower is just arriving to allow this approach to work, and I expect said approach will encompass larger and larger portions of the game content as time marches onward.
So why is this such a big deal? I can think of two obvious answers: (1) it's going to take the shackles off game play like no other innovation has to date, and (2) it's going to make games more consistent and more appealing to a larger audience. Regarding the first reason, what I mean is that game developers will have much more freedom in the situations and options that they present to players. One of the neat bits from the aforementioned HL2 presentation involved the player making his way past a series of traps. In one case, for example, he tossed a grenade under a dumpster, blowing out its supports and dropping it on the bad guys.
Such a thing could always be done previously, of course, but it would have required special-case code. Every such instance of special-case code requires attention during testing and makes the game more complex and more "breakable", which is but one good reason that developers normally try to avoid such special-case solutions to in-game problems. But when the game engine is designed to treat a grenade like any other explosive device, to treat the dumpster as a metal object with a certain mass, and to treat the supports on which it rests as wood with a certain mass and destructibility, there is no special-case code involved. Such specific behaviors "emerge", as some like to put it, from the more general properties of the entities being simulated.
Think about what this means and apply it to an example involving the protagonist of some future game, one who needs to gain access to a heavily-secured building. If he can acquire the uniform of a security guard, then perhaps he can walk right in under the cameras undetected. If he can't find a uniform, then perhaps he can come by the necessary tools to disable the cameras and pick the locks. If he can't manage that, then he might simply blow the hell out of the doors; even metal shatters if you pound at it long enough. Alternately, he might get in by crawling through the sewers and digging/blasting a hole up into the basement. Or, to provide yet one more option, maybe he's the type who will drive a car through one of the building's walls.
What must not be missed is that none of these options requires any special-case code; they all flow naturally from the game engine's ability to simulate a rich and detailed world of substantial objects. Past games like Deus Ex have done much to provide the player with different options, but even the best such game pales in comparison to the possibilities in the next generation of games. When game worlds are built from substances that exemplify properties, rather than as a skin deep bunch of polygonal surfaces, gamers will reap a whirlwind of benefits.
Regarding the second reason, that the shift toward substantial objects will make games more consistent and more appealing to a larger audience, one must put himself in the shoes of the non-gamer to understand. I've been playing video games for so long that it would be largely impossible for me to do so, were it not for my wife. She likes games like Scrabble, Tetris, Bejeweled, etc., not the kind of games that I like to play. Thus, when she drops occasional comments about what she sees me playing, they're like a kind of gold to me for they are the unvarnished impressions of the non-gamer—a.k.a. the unwashed heathen (just kidding, honey!).
Having said that, let me point out that my wife hasn't been impressed with game graphics since I first got my old Monster 3D II accelerator card. Prior to that she had pretty well lumped all the other games I'd played in the same visual category, largely because I played them at 320 x 200 x 8 bpp (i.e., the "X Mode" of the VGA specification). After installing the Monster 3D II card, however, I could suddenly play Quake II at 640 x 480 x 16 bpp at great frame rates. The difference was astonishing. Even my wife thought it was amazing and commented that Quake II was the best looking game she'd ever seen.
But whereas I had noticed the big differences between Wolfenstein 3D and DOOM, between DOOM and Quake, between Quake and Quake II, etc., my wife didn't. What's more, she hasn't seen much to get excited about since. Sure, resolutions have gone up, texture quality has gone up, and various little things have improved here and there, but it's all still the same basic look: textured polygons with fake lighting at one resolution or another. We gamers are an excitable lot. We get geeked if a game introduces a neat flame effect, a fact which has apparently led to a competition amongst game developers to see who can implement the coolest exploding/flaming barrels. It's not hard to impress us; it's a lot harder to impress the non-gamer.
Today's games also still share the same absurd inconsistencies regarding objects in the world. It's annoying, to say the least, that I can shoot people through wooden doors in Raven Shield, but I can't hit them at all if they hide behind a cardboard box. Such inconsistencies are so common that gamers no longer think about them, but they stick out like a sore thumb to non-gamers like my wife. Whenever she sees something like that in a game I'm playing, she just shakes her head and finds it confusing and/or silly. After all, to the non-gamer, why should it make any sense that I can shoot through wooden doors but not wooden rails on stairs or even cardboard boxes for that matter? If you don't get it trust me on this one: it doesn't make much sense unless you're a game developer, faced with the alternative of creating a completely destructible world.
Game development has reached a turning point because technology has finally advanced far enough to allow modeling of a consistent, dynamic world. The games that will be coming to market within the next decade will continue to approach higher and higher levels of verisimilitude with lighting, physics, and modeling substances and their properties. The result is predictable: by the end of the next decade gamers will have photo-realistic, believable, consistent worlds to explore.
I should also point out that the artificial intelligence (AI) exhibited by characters in those worlds is going to be similarly impressive compared to games of the past. If there's anything that FC does, it convinces me that FC has set a new standard for gaming AI. I can't imagine that the advances in FC will be lost on the rest of the game-developing and/or game-playing communities. Its AI is simply wonderful and will surely be emulated in plenty of other games.
To get back to the visuals, however, jaggies will be a thing of the past, for we'll all be running our games at absurdly high resolutions. Gone will be static lighting and goofy tricks to approximate shadows, for we'll all be playing games with truly dynamic lighting. And gone will be many (if not most) of the ridiculous inconsistencies that plague games, some of which provide a confusing barrier to entry for the non-gamer.
Video games have come a long way in a short time, and they have a much longer road to travel yet. But the industry has clearly hit an important turning point, rounding a corner beyond which some truly amazing things will soon be seen. In light of my future predictions I'll say it yet again: it's a great time to be gamer!
04/20/2004