Game Developer's Conference 2004: Carmack Addresses the Masses
by Derek Wilson on March 30, 2004 3:34 PM EST- Posted in
- Trade Shows
Developing for the Future
In the beginning, there was John and a few friends who wanted to make a cool game while sitting around and programming. The fate of id Software was to grow into the position of a giant software development firm. In describing the transition, John joked about essentially becoming The Man against which all indie game developers fight. While (of course) very happy with the position of id Software in the industry, John realizes that it takes those small teams of friends writing games to innovate the next breakthrough in video games, as the large firms are almost locked into a type of game based on their current assets.John Carmack will always be the creator of the first person shooter genre, and id Software will be one of the best developers of first person shooter technology on the market because that's what the people that work there do. Of course, though the next breakthrough in game play may come from your neighbor's garage, the next big breakthrough in game engine technology will come from companies like id who have the money, man power, and influence to develop software for hardware whose concept barely exists.
With expanding complexity and difficulty, it takes a lot to keep up with hardware. Game development cycles are generally much longer than a generation of graphics hardware. Doom III, for example, has been in development for four years, while every six months we get a new set of graphics cards. In the early days, John would be involved with every bit of code that went into his games. Quake III was the first game where John didn't have control over every section of code, as bot AI was developed by someone else. With Doom III, id planned from the beginning to have four lead developers on the project. It just takes more than one person being in charge to push the envelope to the max (though John still feels guilty that there are some files in the Doom III source tree that he hasn't even opened).
Obviously, the gap between software and hardware is of the utmost interest to us. We rely on those on the bleeding edge to help us understand the actual performance of new hardware. Even with all the resources John Carmack has at his disposal, and eventhough he had a very good idea of where the hardware was going, he let us know that it would still take about two and a half years from now for game developers to come out with games that took full advantage of NV4x/R4xx generation hardware. Developers are just starting to get these parts into their hands (John mentioned that he just plugged an NV40 into his personal development system), and getting the most out of this technology and putting it into the hands of consumers is something that John has been struggling with.
To combat the problems with long development cycles, John is looking at ideas like tweaking an older engine to take advantage of current hardware and doing something like a Quake II remix where it's the exact same game but with awesome graphics and targeted at a bit of a niche audience. The cost and development time are lower on such a project since the assets and structure for the game already exist and just the core engine code would need to be reworked. Even this solution would take some time, but sooner is definitely better than later.
Facebook
Twitter
RSS
37 Comments
View All Comments
BenSkywalker - Wednesday, April 7, 2004 - link
"**if I have to do any sort of math involoving a distance from the viewpoint calculation or a projection, its realtime 3d.**"By that standard SuperMarioWorld for the SNES was was realtime 3D, as was FZero for the SNES(along with a slew of other games). I'm also looking at this from a graphics engine standpoint, but pretty much any definition based on your stance(any 3rd dimension factor) will include dozens if not hundreds of games prior to DooM(ignoring actual 3D titles like the original StarFox).
IIRC Carmack himself calls DooM's engine 2.5D, a real 3D engine has objects that take up space, DooM does not have this.
TrogdorJW - Monday, April 5, 2004 - link
#34 - Coming on strong as in the Ziplock distribution method was common in about 1980-1982. By 1984 there were plenty of places that sold boxed, retail games with nice graphics, etc. By the time the Nintendo console debuted, you would have been hard pressed to find anything other than Shareware shipping in plastic bags.Back in the 80s, *everything* was advancing quickly. I mean, what was the first arcade game? Gun Fight? Amazing Maze? Something like that, I'm sure. Released in the late 70s, I believe. The quality of those titles really wasn't any better than the plastic bag games like Akalabeth, and arguably worse.
Anyway, I stand by my statements as well. Computers would have come to be with or without games, and once they came to exist in the home, it was inevitable that someone would make games for them. The stuff done by early hackers on main frames is testament to that fact. Without Miyamoto, about the only thing I am willing to guarantee is that Nintendo wouldn't be around today. Someone else would have filled the gap in hardware and software, though, as the existence of electronic toys was just a magnet drawing all geeks to it.
In 1984, Origin and Electronic Arts were already producing games and making money, and other companies were entering the computer gaming market. Miyamoto and Nintendo had nothing at all to do with PC gaming surviving. You can stand by your statements all you like, but I'm stepping aside, Mr. Zax, and going happily on my way.
DerekWilson - Sunday, April 4, 2004 - link
first off, I'm gonna appologize if I came off insulting when i said that the games mentioned weren't 3d. bad day plus passion (and different perspective) about subject equals bad combination :-) sorry.#33 - KF -- I'm not sure where john came up with the factor of 100 number, but I do know that he was including the added processing power of the GPU in his factor of a million statement. I know he wasn't talking about pure MIPS or FLOPS performance, so its going to be a question of percieved usable power available to a game engine ... it might be a bit subjective, but we'll give John the benefit of the doubt.
Back to the 2d vs 3d question ... last visit for me :-)
I think we've got a definition issue. I'm coming at this from a graphics engine standpoint (and what I used to do in my spare time). From a graphics engine standpoint, whether i'm working with geometry, sprites or an array that holds wall data, this is what it comes down to:
**if I have to do any sort of math involoving a distance from the viewpoint calculation or a projection, its realtime 3d.**
even raycasting (orig wolf3d engine) involves taking some data and creating a 2d projection based on tracing rays from the viewer to an object (causeing depth to affect the final rendering). there's that 3rd dimension :-)
just because the guys at id represented their 3d world with two dimensions of data doesn't mean that their graphics engine was 2d. it just means they didn't store data they didn't need.
from an interface perspective, you only had 2 degrees of freedom in early games, but this still doesn't say anything about the graphics engine.
What I meant by the "environment [those games] are in" was that what we end up seeing is a 2d projection of a 3d rendering (as is true with all current 3d computer graphics from wolf3d to Doom 3).
If you want to restrict labeling 3d games to games with: true 3d physical interation with the world, true 3d levels characters and objects, and a true 3d "environment", then I don't think we can call any games 3d games until we have display technology that doesn't require a 2d projection. Unless you wanna look at the possibility of stereoscopic 3d with 2 2d projections for each frame (one for each eye) and some nice goggles...
But as long as we can recognize that those games had all had 3d engines, it can be acceptible for the first few games on the list to not be called 3d games, but rather games rendered in 3d.
But ... duke3d and the build engine? 6 degrees of freedom for motion, almost free looking (the viewer couldn't tilt his/her head sideways, but could look left right up and down at any angle). So levels are built with a bunch of "2.5d" shapes ... I don't think I could stop calling duke3d a 3d game :-)
again, I'm sorry if I got lines crossed and reacted less than optimally. I hope I've clarified where I was coming from though :-)
BenSkywalker - Sunday, April 4, 2004 - link
[q]Those games are 3d games because the environment they are in are 3d.[/q]The environment they are in is not 3D however, try looking up. He pulled off some extremely impressive tricks with those engines, but they are 2D based.
[q]However, C64 and Amiga were coming on strong, and the PC was just starting out.[/q]
Strong? Games distributed in zip lock bags is strong? The comments thread isn't a good place to have a discussion as lengthy as this one would require, but I was there for the early computer gaming days too. I'm not speaking as someone who is new to either end of the spectrum, and I stand by my statements.
KF - Saturday, April 3, 2004 - link
>the computing power we have now is about one>million times that of his first Apple machine.
Did John Carmack explain "a million."
Apple II's original 6502 processor was pobably 1 MHz, although it might have been more. The Apple IIc, with the "c" was later I think, and so a higher clock. 4G/1M=4000. Getting that up to 1 million is tricky. We need another factor of 250 Let see, 32 vs 8 bits is a factor of 4. I'd guess instructions per cycle was 1/4 for a 6502, and maybe 2 for a P4, for another factor of 8. We are up to a factor of 32. It's only short by a factor 8, but maybe John Carmack has one of those 8 processor computers.
>...with about 100 times more power, Carmack thinks we will be able
> to do in real time what it currently takes four and a half hours to render...
4.5 x 60 minutes x 60 seconds x 60 frames comes close to a factor of 1 million. So, did John Carmack explain why a factor of 100 going to be enough?
TrogdorJW - Thursday, April 1, 2004 - link
#27, I think you're entirely wrong. I remember the early 80s quite well. Consoles were dying in the wake of Atari problems, true. However, C64 and Amiga were coming on strong, and the PC was just starting out.My point isn't that Shigeru isn't influential, it's that he wasn't exactly innovative. He has had zero impact on the world of PC gaming, and in fact you could argue that if anything he has *hurt* the world of PC gaming. In the land of the consoles, however, it's a completely different story. He created/saved Nintendo, which then allowed Sega and later Sony to enter into the competition. With or without him, though, PC gaming would have continued.
The problem with console games is that they discourage innovation, because there is a much larger barrier to entry. PC games have shareware (although not as much these days), demos, betas, etc. You can also create mods, map packs, etc. There is no real way to enter into the console game creation world without a significant investment of capitol, and that means that crazy and innovative ideas are less likely to come from that realm of gaming. Yes, some titles are still innovative, but overall, consoles steal ideas that were successful elsewhere more than they create them.
DaveNCheez - Thursday, April 1, 2004 - link
I would love Quake2 Remix. Quake 2 is probably my favorite ID game... And I have been playing since the Commander Keen Days hehe...Just hard to get a game going.. Most friends don't want to play it or don't know about it. And there is no in game server browser...
Quake2 Remix would be a blast.
mvt3 - Thursday, April 1, 2004 - link
just cause it's personally interesting to me...define influence in the game industry...i think carmack has a hell of a lot, as did shigeryu
hmm ed logg though...no one mentioned him. also perhaps find this of interest:
http://www.arcade-museum.com/game_detail.php?lette...
there are many people that had incredible influence just a little more subtle than the overly publicized ones.
carmack's engines rule though. must say it.
CrystalBay - Thursday, April 1, 2004 - link
Q2 remix does sound interesting...DerekWilson - Thursday, April 1, 2004 - link
#26:"umm...wolfestein 3d, doom, heretic, hexen, duke Nukem 3d umm....AREN'T 3d"
Woah! do0d. That's the most wrong thing I've seen in a while.
In the grand scheme of things, we have as much left to render "correctly" now as we did then. Its all about approximations, optical illusions and doing whatever is good enough and fast at the same time.
billboarding sprites is a valid way to represent 3d objects in 3d virtual environment without sacrificing polygons. Its still done today in some cases.
Those games are 3d games because the environment they are in are 3d.