PDA

View Full Version : John Carmack on DOOM 3 and Game Development Trends



Brevity
05-15-2003, 08:19 AM
The Wizard of id chats about the state of the industry
By Dave "Fargo" Kosak | 5/15/2003

Developer: id Software
Publisher: Activision
Release Date: 2002
Genre: Action



Amidst the noise and bustle of the Expo, it was a relief to get into the relative quiet of a conference room. On stage sat John Carmack, id Software co-founder and arguably the most influential gaming graphics programmer in the world. He was there to talk about DOOM 3, as part of a public interview hosted by Newsweek technology editor N'gal Croal, but DOOM 3 was only part of the discussion. During his hour-and-a-half talk, Carmack touched on the problems associated with building graphically intense titles, the time needed to create modern content, challenges faced by small developers, and even eclectic topics such as rocketry or creating an online "metaverse."

Of course, it all started with DOOM 3. The lights dimmed and once more crowds were awed with footage. Whereas last year we got a taste of what's to come, this year we got a gulp. DOOM 3's darkened, flickering corridors were now populated with 36 flavors of horror. Along with zombies, we had zombies with chainsaws. Strange spidery creatures with bony limbs scurried in and out of the shadows. Imps clambered down pillars, moving not unlike Gollum from Lord of the Rings. Monstrous demons hurled around smaller zombies like throw pillows. There were creatures that resembled baby-heads with moth wings. (!) Room after room was filled with cinematic 'scare' moments, such as when the half-human figure submerged in a tube of bubbling fluid suddenly jerked itself against the glass and screamed.


Carmack (right) takes questions from the audience.The world was jammed with pixel-perfect detail, showing off what Carmack described as "the pores and veins and crevices" of each individual monster and surface. Screenshots look okay, but the interplay of the lights and shadows as the light sources move and the creatures move through them is absolutely incredible when seen in motion.

Scared yet? DOOM 3 is a far cry from the original DOOM, with its hordes of monsters and blazing fast action, and that shift in design priorities was one of the topics that anchored John's talk.

Designing on a Six-Year Time Frame

How did it all come together?> "We made good strategic decisions two and a half years ago," Carmack explained. He went on to describe the problem content creators face when they're trying to push the graphical envelope: predicting where the videocard market will be three or even six years down the line. That's three years to bring a state-of-the-art game engine to market, and then another three years (hopefully) of the engine being relevant for gamers. Granted, Carmack has an easier time than most: He can influence the hardware manufacturers, making his designs somewhat self-fulfilling.

When Carmack started drawing up his plans for the DOOM 3 engine, he wanted it to break new ground graphically -- and then he wanted to design a game around it. It was about getting what he called "a real 'WOW' impact," and it meant sacrifices elsewhere. "You have to make compromises," he explained. A complex lighting renderer meant that players wouldn't be able to enter a room full of a hundred monsters and mow them down like grass. It defines a different experience; one where the player progresses from elaborate set piece to elaborate set piece engaged in brutal up-close combat with small groups of extremely detailed monsters.

Next: The battle for design time

Carmack is a purist. He doesn't incorporate graphical gimmicks for any one particular effect on any one piece of hardware. He extolled the value of "universalness and consistency" across hardware, and the technology he wanted to design was built on similar principles. Current technology renders different types of surfaces differently, and it renders light sources differently based on whether they're dynamic or static. Not so in DOOM 3: "Lights and surfaces are all treated as first-class citizens," he said, meaning they were rendered by the same processes. "It's not about being spectacular," he explained. "It's about being consistent."

The result is the tightness of DOOM's world, where light plays off of every surface and shadows move with the same fluidity as the creatures that cast them. This, too, leads to different design decisions. Quake 3 was a game about action and movement; richer graphics would be wasted on a game like Quake 3, where nobody has time to look at stuff. But DOOM 3 is about being rooted in an environment, being trapped in a far away place where each room is a custom crafted set piece. You're meant to linger.

Besides, modern games are so detailed that crafting environments takes a lot of time. This is wasted development time if the game encourages people to rush through. And that brings us to the next topic...

The Battle for Design Time


Bring out the heads!Carmack spent a great deal of time talking about how advances in game rendering technology are causing game creation time cycles to spiral out of control. id Software has always maintained a very small team, and Carmack is constantly fighting a battle to keep design cycles short. "It's a losing battle," he concedes. Doom 3 is pushing three years of development time. The new engine allows game mappers to see the lighting of a level in real time, just as it will be rendered in-game (instead of waiting for a time-consuming compile), but the time savings are eaten away by the level of detail.

When fighting this battle, how do developers save time? Once again, they make sacrifices! The detail with which models can be rendered puts a lot of pressure on animators to make the models move as realistically as they look. That's a LOT of work. Game design can help skirt this issue: In DOOM 3, almost all the models players interact with will be demons or mutants. Nobody knows how a demon is supposed to move, so the animators have a freer reign. Similarly, most of the humans players meet are undead -- if their movements are stiff or lumbering or awkward, all the better. (For this reason, id Software used motion capture for the few non-zombie humans players interact with at the beginning of the game. The possible interactions are limited here, which made it easier.)

Carmack talked about "rendering" vs. "simulation" when tackling problems like these. Rendering technology has advanced by leaps and bounds over the last couple of decades, but realistically depicting a biped walking up or down stairs without pre-animating it is a relatively unsolved problem. And ultimately, designers have to make decisions: Will realistically simulating this movement impact gameplay? Or is there a way to design around it? For DOOM 3 the team opted to hand-animate the models and limited the situations where they would look out of place. The idea is not so much about looking good as it is to "avoid looking bad," as Carmack put it.

The topic of development time came up again during the question and answer session. "Can a garage developer still make it in the industry?" someone from the audience asked. Carmack admitted that there wasn't a lot of choice for a small developer -- they can't directly compete with an "incumbent" developer working in the same established genre with a bigger team, bigger budget, and longer development cycle. On the other hand, small developers can attack a profitable niche, work on a new genre of games, or develop for platforms that have shorter development cycles (like the Game Boy Advance.) The lengthening development cycle is definitely changing the composition of the industry.

Multiplayer, Teamplay, and "The Metaverse."

So why did Carmack and id Software, known for their love of furious competitive deathmatch, switch gears to what Carmack described as simply "a really really good linear single-player experience?"

The answer lies partially in the direction the market is moving. Team-based games are dominating the online action scene, for a number of reasons. The variety of gameplay is part of it, as well as the variety of play styles a team game can support. Carmack also figures that, in team games, half the players always win (as opposed to one player in deathmatch.) But this, personally, never interested Carmack. "I never really appreciated team activities ... in real life or in computer games," he said, to the amusement of the audience. Instead, id Software moved in another direction.

Brevity
05-15-2003, 08:20 AM
Later during the talk, the subject returned to multiplayer when an attendee asked about Carmack's thoughts on "The Metaverse," or a shared virtual environment as depicted in the classic novel Snowcrash. Before id Software decided to make DOOM 3, Carmack had toyed around with and pitched the idea of creating what he called a general purpose online environment. However (in part due to profitability considerations) the company opted not to pursue it.


If this is a virtual world, we don't wanna live there.But Carmack still wants to see it happen: "maybe someone will pick up on that," he stated, at the same time pointing out that several companies are trying. He went on to talk about how a new generation of input and output devices is overdue, enabling people to interact more fully with virtual environments -- much as the analog mouse helped make the last generation of 3D games possible. "I'm waiting for someone else to tackle VR goggles," he said, inferring that it's time to give this technology another try. Head-tracking software might come next. Or cameras allowing you to broadcast your facial expressions onto a digital avatar. It was clear from his lengthy answer that he was still fascinated by the idea, so don't interpret DOOM 3's single-player focus as a definitive or permanent direction.

A "Golden Age" for Graphics Rendering?

Tough choices for developers, a hard market to break into, and spiraling development times. Is the picture really so bleak? Not at all. In fact, when it comes to graphics technology, Carmack said that we're at a "Golden Point" in graphics API tech. He claims that graphics professionals like himself now have the ability to render any functional equation they need via the hardware cards. New APIs aren't needed every time someone wants to create a new special effect. He goes so far as to claim that he can create a next-generation engine using the current APIs.

Certainly we're seeing this all over the show floor, as games like Deus Ex 2, Half-Life 2, and especially Doom 3 push some gorgeous pixels onto monitors scattered around the expo. Time will tell if the graphics enhancements of DOOM 3 justified the gameplay decisions that were built around them, but there's no doubt that when the game hits shelves, it'll represent a major leap forward in presenting artificial spaces in real-time.

But before we let you go, you might be curious -- as was another audience member -- how Carmack's forays into space flight are proceeding. Carmack called rocketships his "other avocation." "Specifically," he added, "Rocketships to carry people." His own homegrown team of engineers hope to have a manned supersonic flight airborne by the end of the year, with a craft capable of space shots taking to the skies by the end of next. He raved about how much fun it is to return to a place where he can "learn something new in every page of every book you look at," and he says he "certainly recommends taking a hard right turn" whenever you feel you've mastered a field. From DOOM to the moon, Godspeed!

Liquid Gears
05-15-2003, 02:01 PM
DOOM III looks amazing, seriously, this game has the most stunning grafics I've ever seen. I want this game so bad, that I want it just as much as Halo 2. I LOVE the way the creatures crawl out of there hidding spots. That's really cool, my favorites the one that climbs out from under the floor in the E3 2003 Doom III Demo. That game's gonna be HUGE!! When does it come out anyway?