Doom III, a step in the right direction...or not?

[Nagorak]

There is something that's been bothering me for a while about the new Doom 3 engine. Anyone who has read my posts, probably knows I'm not a big fan of id's games, but let's just put that aside, since it's neither here nor there. In this case I'm talking about just the engine technology itself.

As we all know, Doom 3 uses a new advanced lighting system which is supposed to be more realistic. Unfortunately, I'm starting to wonder what trade-offs are being made in order to support this new lighting engine, and whether they are worth it. Obviously the models in Doom3 have noticeably lower poly counts, than those found in other games (UT2k3, etc).

But in addition to that I've noticed that ALL the screenshots I've seen for Doom 3 are completely indoors (small hallways and rooms). The question is, is this due to game design, or whether it's to cover up a serious limitation in the engine. Obviously id knows their engine inside out, so it's not surprise that they'd design their game around the engine's capabilties.

It's only been in the last few years that games have finally advanced to the point where we can have large outdoor environments, while still getting decent performance. Is it possible that by pushing the envelope of 3D technology too much, the Doom3 engine is going to thrust us back into the bad old days of "hallway crawls"? Apparently the Doom3 engine also has problems with trees and stuff, due to the lighting just being far too slow with complex objects.

So my question is, are these new lighting effects really worth it? Yeah, it may look better indoors, but the second you look out a window will your FPS drop through the floor (of course in D3 there won't be windows in that case). The fact that id had to use such ridiculously low poly models has me very concerned about whether the engine could run at all decently with large open spaces.

I know developers can license a different engine, or make their own, but considering the many development groups that just license id by default, might the move to the Doom3 engine not have a rather adverse affect upon game design in the coming years? Is this really a step in the right direction-- the direction of more lighting realism-- or is it a step in the wrong direction-- the direction of more enclosed and less involving environments.

What do you think, am I just paranoid?

 

[jvd]

First off have you ever played a doom game ? It takes place inside. Second i believe the engine can switch to another lighting mode for outside shadows and lighting . Third well , in the future when better cards come out you can up the polygon count in the engine .

 

[Mulciber]

yes, doom has always been an indoors sport for the most part. though there are some outdoor levels in the original dooms. but since i havent played doomIII how am I to know how it will work with outdoor environments.

and no, games devs dont just chose an engine by default. those things costs them tens of thousands of dollars to purchase and get trained on. they have to look at all the current engines and evaluate which one is best for them...its a HUGE investment

 

[Reverend]

You have to look beyond your preconceptions.

- When you talk about lighting, the majority of its importance has to do with indoor scenarios - almost anyone can make outdoor scenes look "fantastic" due to the way games are made.

- "low poly models" complaints... my opinion is that I absolutely agree with Carmack's decision about this vis-a-vis his engine and hardware limitations. I prefer realistic environments versus curvier-looking models (and, please, don't bring in N-Patches into this!)

Finally, for someone as concerned with graphics as Carmack, the most visually impressive game has as much to do with how a game (and its engine) is designed to take advantage of the available hardware and its features and limitations, as it has to do with what you want.

But in addition to that I've noticed that ALL the screenshots I've seen for Doom 3 are completely indoors (small hallways and rooms). The question is, is this due to game design, or whether it's to cover up a serious limitation in the engine.

There are lots of priorities in making a game. For someone like Cramack, my opinion is that this has more to do with what he wants and what graphics companies & current technology limitations (process tech limitations, so on, so forth) can't give him. He wants to push the graphics envelope... but given the current limitations, he has to make sacrifices. Is that news to you??

Maybe it has to do with pushing the graphics envelope given what Carmack perceives as the the way the (video) market will be.

Whether a game is predominantly indoors or outdoors shouldn't be worth discussing here - it's what sort of graphics technologies (that are being taken advantage of) that are being used that should be.

 

[Cyborg]

You ask, you get!

http://www.quake.de/pictures/doom3/fullsize/doom3eatme.jpg

Whether you can go outside or if its just a window view, that i do not know.

 

[DemoCoder]

There are outdoor games that use stencil shadow volumes, but it's not very effective since in the wide open outdoors, there is not much self-shadowing going on and the light source is typically the sun, ambient light, which doesn't move quick enough to demonstrate "real time shadows"

Typically, only the main character and other models, vehicles, etc get shadowed applied.


I personally am not really impressed with Unreal2. The 3dMark2003 Proxycon demo shows more of what I want.

 

[Doomtrooper]

The main problem with ID engines is their color palette, always dark greys and browns, very dreary and gloomy and as mentioned before 97 % inside.
I prefered the original UT engine over Quake 3 just for that, plus the ability to handle outdoor enviroments much better.
Quake 3 :Team Arena required a pretty good working of the engine to handle outdoor enviroments well, but what they did, they did well...team arena had some pretty good outdoor maps.
Overall ID Engines have spawned much better games than the original game that ID used the engine on:

Halflife
MOHAA
RTCW


Doom 3 will look good, but will probably be poor in the replayability and storyline department (Like Unreal 2). IMO ID has never delivered a single player experience that of Quake 2 (IMO their best game).

There is other engines coming that will probably be technically superior to Doom 3, Stalker is one that comes to mind..yet doesn't have near the Hype (another drawback to ID games, sort of like seeing a movie hyped to death then when you finally see it..it's like 'well that wasn't that great'.

Some good screenshots here of that engine:

http://www.stalker-game.com/index.php?t=gallery&rnd=1879

 

[ChrisK]

I think both things are needed: vast outdoor environments *and* better materials and lighting.

I'd suggest you to work with some 3d graphics package, if you have the opportunity.

Make some 3d scene. Then enable shadows for your light sources, add bump mapping, specular highlights or specularity maps (if that application supports them) and reflectivity to your materials.

Then turn all of these things off again, and see how dull and unrealistic all of the objects look.

If I'm looking at today's games (not considering technologies used, but from a purely visual standpoint), I can't see many advances in lighting since the days of the Voodoo2 - apart from some rare cases of hard shadows, bumpmapping or environment mapping.

If Doom3 *finally* changes that, all the better.

Chris

 

[Ostsol]

Well. . . the colour palette is obviously not an engine limitation, merely an art choice.

If Doom 3 supports some advanced outdoor terrain ala the Unreal Engine, I'd expect that they'd use lightmaps rather than shadow volumes. Shadow volumes will still, of course, be used for representing shadows cast by objects onto the terrain, but the terrain's own shadows are likely to be done via lightmaps. Fully dynamic lighting for terrain would be quite useless for a game such as Doom 3 unless a day-night cycle is to be implemented -- which isn't something necessary for Doom 3's gameplay.

 

[DemoCoder]

I haven't seen any movies that had convincing outdoor daytime CG scenes. Typically, they will be in space or at night or dusk. Global Illumination is a tought nut to crack.

The most realistic "outdoor" renders I have seen used Image Based Rendering tricks (e.g. real photography). See Matrix 1/2/3.

 

[BNA!]

If Doom 3 supports some advanced outdoor terrain ala the Unreal Engine, I'd expect that they'd use lightmaps rather than shadow volumes.

I highly doubt that.

//

I know that Doom will be able render some competent outdoor scenes (apart from the already known material seen in the E3 material).

But, of course, it is not what people will call a terrain engine like they have in a flight simulator. I still wonder why peopld don't complain about the indoor rendering capeabilites of racing and flight sims?

Did Unreal2 become any better because of outdoors?
Or is it that id software engine haters usually use the "outdoor" argument to stop the discussion.

Well, Doom will hold more than one surprise

 

[Ostsol]

If Doom 3 supports some advanced outdoor terrain ala the Unreal Engine, I'd expect that they'd use lightmaps rather than shadow volumes.

I highly doubt that.

Any particular reason?

 

[BNA!]

Any particular reason?

The primarily reason is, that Carmack shoots to 100% for a unified lighting model. Breaking this rule the very first second when creating content doesn't seem logical.

Can you tell me any particular reason why outdoors would need a lightmap? For faked shadows?
Hell, id software already fakes shadows behind alpha channel cutout textures, so it would be very easy to fake some nice terrain shadows the same way.

On the other hand - terrain looks very good when rendered in Doom using one single directional lightsource. Just don*'t expect it to have a million trees with two billion leaves each dropping dynamic shadows on top of polygon modelled self shadowed grass.

 

[Dave H]

To me the bigger question regards scenes that are indoors yet adequately lit. Sure a D3-like engine is perfect for when the entire room/hallway is lit by a single light bulb placed behind a giant ventilation fan or on the end of a kinetic sculpture. But normal rooms have lots of light sources. I just went into my (parents') living room and counted: 6 can lights in the ceiling, 3 table lamps, a floor lamp and a lamp on the upright piano. Imagine rendering that in the DoomIII engine! (Only 57 passes per frame for the GF 1. )

Almost forgot: plus a TV and a fireplace. And ambiant light from a large window and other rooms.

Meanwhile, Carmack has said the average poly in DoomIII is lit by a grand total of 2 light sources. Perfect for the atmosphere they're trying to create, but not for simulating any place you'd actually want to be.

 

[Hellbinder]

(and, please, don't bring in N-Patches into this!)

*snif* *snif*...

/walks away with head hung low kicking some dirt...

 

[DemoCoder]

Increasing the number of realtime lightsources won't bump up D3's quality very much. In dark rooms, you can get away with no real global ambient/diffuse calculation, and hard shadows. However, in a room with lots of lights, I expect correct diffuse illumination, I don't expect dark contrasty shadows except in corners. It just won't look like what your brain "expects" it to look like with that many lights, as a result, it will look plain phony.

Brightly lit Cg, especially in an indoor environment with lots of diffusivity, are hard to get correct.

The # of lights is not the problem. Besides, in photography and cinematography, you mostly work with 2-3 lights max in most scenes.

 

[LeStoffer]

Brightly lit Cg, especially in an indoor environment with lots of diffusivity, are hard to get correct.

The # of lights is not the problem. Besides, in photography and cinematography, you mostly work with 2-3 lights max in most scenes.

Indeed. I haven't been doing this much myself, but getting something close to looking realistic in apps like 3ds max/Maya in the good ol' days involved a lot of trial and error on lightning setup.

I wonder if Doom III will be easier for the game artists in this respect. It might have been time consuming to setup good looking light with the oldtimer lightmaps, but at least you had super precise control over the light.

 

[BNA!]

I wonder if Doom III will be easier for the game artists in this respect.

The realtime editor render window makes light tweaking very easy.
A good eye for key / fill lights doesn't hurt either

 

[DemoCoder]

Maybe level design will get much better if hollywood cinematographers and lighting specialists get involve. Alot of effort is spent on getting correct lighting in every scene in a movie. Even if some guy just has to use the toilet, you can bet they planned how to setup the lights, what film stock to use, correct exposure, and camera position for that scene.

 

[Nagorak]

Finally, for someone as concerned with graphics as Carmack, the most visually impressive game has as much to do with how a game (and its engine) is designed to take advantage of the available hardware and its features and limitations, as it has to do with what you want.

The first Unreal was visually impressive...Would it have been so if the whole thing had taken place inside the ISV-Kran? That's the question. Just because Doom3 has a better lighting engine, does not necessarily correlate with it being visually impressive. That's what I'm asking.

It seems like some of you feel that an "all indoor" game is just fine, but I really disagree. It actually sounds reasonable to me to use different rendering methods for inside and outside, but it seems some of you are skeptical that Carmack will do that.

I'm just wondering if this move might not be too soon and too limiting. NOLF2, RTCW, Jedi Knight II, UT2k3, BF1942, Tribes 2, etc, they all had outside areas (some more than others). You can count the number of modern games that are just inside on one hand (if that, I actually can't think of any off the top of my head).

Maybe the Doom3 engine will fare fine outside and there will be no problems, but if not it could be bad for id, if not the entire PC gaming industry (depending on how much effort really goes into engine selection). I mean, realistically, it would be best to use nothing but ray tracing in games, but it would be so slow that it would be pointless.