The future of PC games - Engines And Engineering

[apoppin]

Gamespy (& Intel)'s Article.

(1st of 4 parts) Not too technical (it's Gamespy) . . . but still well worth reading!





I hope this is not a repost. At least I haven't seen it here (it is quite current).

 

[apoppin]

Really worth reading . . .

. . . is this in the wrong forum?

 

[DaveSimmons]

Yes, just not getting any replies

I saw the blurb on Gamespy's page but the "intel" kept me from reading. intel is so far out of the loop for game graphics / engines compared to ATI and nVidia (and Carmack) I expected it to be lousy and/or biased towards CPU rendering in engines.

I wish I had the spare time and energy to play around with engines but after a long day of writing Windows C++ code for work I'm usually more in the mood to play games or watch a movie.

Edit - just read it, it's a pretty good overview of current engine & AI thinking by developers, and intel-free except for the ads.

 

[singh]

Sorry, it's not technical enough

 

[apoppin]

Originally posted by: singh
Sorry, it's not technical enough

That's why it's here and not highly technical.

And I found it most interesting that Doom III will use "unified" lighting:

"The problem with unified lighting is that you end up sacrificing unless you have a very specific look that you are going for. I think id will get away with unified lighting in Doom III because they are going for a dark, creepy, very atmospheric look. But there are only so many dark, creepy games that you can make."

Physics will improve:

Epic took a big step toward realistic physics by integrating "Rag Doll" technology into its Unreal engine. Here, pre-canned death animations used in previous first-person shooters are replaced with physics-based body falls.

"In Unreal 2003, we have 'Rag Doll' physics for player deaths so that when you shoot a player in the shoulder and you kill him, he goes flying shoulder-first, maybe spins a few times and falls down a stairway, and ends up in some realistic pose," says Sweeney.

"It was a big improvement over Unreal Tournament," says Sweeney. "In Unreal, if you shot some guy on a ledge, he died right there with half of his body hanging off.

"The next big step will be getting Rag Doll working in situations besides player deaths, such as when you shoot a player in the shoulder and he doesn't die, but goes flying back, then recovers and starts fighting you. The physics system we have now is capable of these kinds of things, but there is still some R and D work to be done."

"In the future, I see frightening realism," adds Taylor.


And AI looks to be improving also:

"A.I. isn't just about making characters that can play well. It's about making them play believably."

 

[nirgis]

I had an interesting idea, but I'm not sure if it is already used.

If we had a few companies solely designed to making engines, then surely they could make several extremely high quality ones, say, like a fps one, or a rpg/simulation one, etc. I know that most designers use pre-existing engines like the Quake 3 engine, etc. I would have to think however, that the quality of games would have to increase when game designers did not have to worry about what engine to use, and instead could focus more on gameplay, graphics, etc., given that excellent engines existed already for sale by liscense.

 

[apoppin]

Originally posted by: nirgis
I had an interesting idea, but I'm not sure if it is already used.

If we had a few companies solely designed to making engines, then surely they could make several extremely high quality ones, say, like a fps one, or a rpg/simulation one, etc. I know that most designers use pre-existing engines like the Quake 3 engine, etc. I would have to think however, that the quality of games would have to increase when game designers did not have to worry about what engine to use, and instead could focus more on gameplay, graphics, etc., given that excellent engines existed already for sale by liscense.

It already is being used. The Quake III and Unreal engines (as well as LithTech and others) power many different games. The Sims uses another type of engine (found in other games). They are licensed (for around $100,000 each; more or less - this may be "old" info). I know licensing engines is a significant part of developer's profits. The Doom III engine is also available for licencing (but the games cannot be marketed until DIII is out).

What I'd like to see is two "versions" of each game released - one for "high-end" systems and the other for "regular" systems so compromises in programming don't have to be made.

 

[nirgis]

It already is being used. The Quake III and Unreal engines (as well as LithTech and others) power many different games. The Sims uses another type of engine (found in other games). They are licensed (for around $100,000 each; more or less - this may be "old" info). I know licensing engines is a significant part of developer's profits. The Doom III engine is also available for licencing (but the games cannot be marketed until DIII is out).

What I'd like to see is two "versions" of each game released - one for "high-end" systems and the other for "regular" systems so compromises in programming don't have to be made.

Right, but the Quake 3 designers had to design both the quake 3 game and its engine-- imagine if they had spent all of their time on just the engine. As for your second point, I think that fits perfectly with my idea, there would obviousbly be low-tech engine companies, high-tech engine companies, and a continuum of engines in between

 

[DaveSimmons]

What I'd like to see is two "versions" of each game released - one for "high-end" systems and the other for "regular" systems so compromises in programming don't have to be made.

Actually one version is good enough for a well-written engines, by reducing screen res, switching to using the lower-polygon modlels and lower-res textures that are already there for distant objects, dropping some animations, and turning off some of the lighting / shadow computations.

For example, Dungeon Siege ran just fine at 640x480 on a friend's Celeron 400 MHz, runs with more eye candy at 800x600 on my P3-933, and runs with everything maxed on slightly faster machines.

There isn't really any other way to do it. Q3 can't magically run on a 233 MHz machine except by downgrading its rendering to the level of an engine that did run well (Q2).

 

[apoppin]

Originally posted by: nirgis

Right, but the Quake 3 designers had to design both the quake 3 game and its engine-- imagine if they had spent all of their time on just the engine. As for your second point, I think that fits perfectly with my idea, there would obviousbly be low-tech engine companies, high-tech engine companies, and a continuum of engines in between

I think it's happening already . . . for example, if you are designing a game with more rudimentary graphics - i.e. "Deer Stalker" you can license a (cheaper) earlier build of the Unreal engine or Quake I. The Quake designers envisioned their engine powering other games. Look at Quake III and then compare the highly tweeked engines of Alice and RtCW.

Let me clarify what I meant about 2 versions with an example: Doom III comes out optimized for the GF3/Radeon 8500 class of cards. A year later a version more optimized for the 9700/NV30 comes out. Just as the Quake I engine has been rewrittern to enhance the graphics pretty significantly, ver. 2.0 would be much more impressive on the highest performing machines.

While I'm at it, I like to see a feature in a game that allows you to replay the "movie" of your game (minus all your failures, "deaths" and start-overs). It'd be pretty interesting to just sit and "watch" Max Payne" or "NOLF2" from start to finish.

 

[Adrian Tung]

Originally posted by: nirgis

It already is being used. The Quake III and Unreal engines (as well as LithTech and others) power many different games. The Sims uses another type of engine (found in other games). They are licensed (for around $100,000 each; more or less - this may be "old" info). I know licensing engines is a significant part of developer's profits. The Doom III engine is also available for licencing (but the games cannot be marketed until DIII is out).

What I'd like to see is two "versions" of each game released - one for "high-end" systems and the other for "regular" systems so compromises in programming don't have to be made.

Right, but the Quake 3 designers had to design both the quake 3 game and its engine-- imagine if they had spent all of their time on just the engine. As for your second point, I think that fits perfectly with my idea, there would obviousbly be low-tech engine companies, high-tech engine companies, and a continuum of engines in between

I don't think that marketting would buy this idea. In order to promote your engine, you need a good game. But if you don't develop a good together with the engine, no one will take your engine as it hasn't been proven in the market yet. Sort of chicken and egg thing... that's why most developers spend time developing both the engine and the game.


atwl

 

[apoppin]

Quake started out with ver. 1.0 and is currently working on ver. 4.0 . . . same thing with the Unreal engine. These specific game franchise design teams continually work to upgrade and refine their engines to support the games that will be taking advantage of them. I imagine paying at least $100,000 for a game license gets some serious tech support.

 

[DaveSimmons]

Let me clarify what I meant about 2 versions with an example: Doom III comes out optimized for the GF3/Radeon 8500 class of cards. A year later a version more optimized for the 9700/NV30 comes out. Just as the Quake I engine has been rewrittern to enhance the graphics pretty significantly, ver. 2.0 would be much more impressive on the highest performing machines.

So you'd really like to be able to drop-in an an "engine upgrade" to an existing game. It's nice in theory but some technical problems:

- engines are currently integrated at the source-code level and customized for each game, they don't remain cleanly separated from the rest of the game code the way that (for example) the video card drivers do.

- the way games are currently written the code would not be able to make use of any new features. For example the non-engine code in Q1 - Q3 doesn't have any way to take advantage of the unified lighting model in the Doom3 engine.

- the game also is missing most or all of the object data that the new engine wants, for example the Q2 textures wouldn't include bump-mapping data.

Until engines reach a plateau point the main upgrades that can be done to existing games are to allow higher resolutions, more polygons / hires textures, and anti-aliasing. You get those upgrades when you move to a better graphics card.

 

[Lonyo]

You could run old Unreal stuff in Unreal Tournament which had an engine upgrade, but it didn't make the Unreal stuff look any better really . They would have to design the stuff with an upgrade in mind, or it probably wouldn't work.

 

[apoppin]

I AM suggesting that they design game engines with upgrades in mind.

I think what was done with the Quake ONE mod was amazing. And even the old Half-Life engine got a decent (if subtle) upgrade from installing Blue Shift.