DooM3 Graphics Engine

[BenSkywalker]

Jeff-

Try setting your AGP aperature to 512 and then tell us there's no difference...

I was pointing out that a lot of people showing 'Ultra' numbers couldn't just post numbers and assume that they were seeing 'Ultra' settings. We already covered the combined RAM issue in this thread.

You're creating the problem. Doom 3 is a dark game... you're not supposed to crank the gamma and look into every dark corner and see perfectly.

You have the in game brightness turned down? If not, you are raising the gamma yourself. As far as not being able to see in every dark corner, how about seeing in lit rooms?

Aliasing in every game is noticeable.

HardOCP was claiming it wasn't, I was pointing out that they were wrong.

Actually it's not. If it is for you, either you're in a bright room, or your monitor isn't displaying it correctly.

I'm playing in a dark room on a ~six month old NEC FP2141SB-BK- perhaps your monitor isn't displaying it properly, I have the best gaming monitor you can buy

BFG

We are comparing Doom III to FC, not to KOTOR. Please pay attention.

You brought up FC's foliage as being impressive, I'm pointing out it looks extremely poor in comparison to a title that predates it by some time. You were using Painkiller to compare to D3 also which I had no problem with, he!l I went out and bought the game just to see what you were talking about.

So where does D3 create a rippling water effect and then illuminate it?

Just show we on a map where the lake is on Mars and I'll get those screenshots right over to you.

I'll check this next time I'm in the game.

Go ahead and check it out, I'm not joking.

Reflections are now over the top?

Mirror finish on a sewer main? Absolutely that's over the top.

Even with a flashlight on them?

Of course, most tiles are extremely dull until they have something shiny put on top of them. Semi glosss paint actually reflects considerably more light then non waxed tiles.

And what about mirrors?

D3 doesn't have an engine capable of drawing the light back across the room, you are right about that.

No because Quake III not only ran at whip-lash speeds but it also basically delivered the highest eye candy around at the time.

On a Voodoo1? Give me a break. 39.9FPS on a V3 2K running 1024x768 is breakneck? And that was a board that came out months before the game, not years.

Point out some cheating and I'll complain, don't you worry about that.

The 'optimizations' they were running on the 9600 series boards for so long behind the users back and in direct contradiction to what the app requested. That was one of your criteria for why nVidia was 'cheating'.

Oh, if you only knew what my current hardware is.

R9700Pro is it not?

As a side note, JC commented that the NV3x is performing shader subsitution in D3. Have you seen me complaining about that? Not really because I really can't be bothered.

He stated that minor changes made to the code made it fall off the fast path. I don't know if that indicates compiler optimizations particularly for those shaders or substitution(but I've never really cared either way if it doesn't effect IQ, the same standard I use with ATi dating back to 'Quack').

So outside you'd expect no shadows? You are kidding right?

The shadows are there, I'm a bit confused here.

Yes, I did. The 32x0 configuration in zixel mode can stay when perfoming MSAA samples on the R420.

Can it or can't it, you are swapping back and forth here.

It's not outputting 32 while running MSAA.

I stated that it loses zixel fill when running MSAA, now you swap back and forth on if it does or not. If you disagree with that then it needs to be a 128x0 part, if not then you agree with my initial statement that you started off disagreeing with.

You said it can on the NV40 which is fine but I haven't heard about that so I wasn't commenting on it.

The NV40 loses zixel fill although it is still operating as a 32x0 part.

Eno-

Turn off ALL lights in the room, turn the gamma/brightness up till it starts to look grey, then lower back till its between grey and black , and thats the highest you should raise it.

Gamma you can crank to obscene levels before you lose true black, I wouldn't think of raising brightness to the point where it was anything resembling grey, the thought of it is absurd to me(much as the insanely overbright settings they use for displays by default in general are fairly sickening). If you are looking at a black screen you shouldn't be able to tell if the display is on or off, numerous calibration tests and software have backed this assertion for my particular display.

 

[VirtualLarry]

Originally posted by: eno

Originally posted by: BenSkywalker

As far as turnin up your gamma... the game was meant to be dark... youre supposed to run around conserving your flashlight and getting scared. :/

I assume you haven't played the game yet........ conserve the flashlight? Cranking the gamma is required to see pretty much anything in the game, even when there is lighting and/or you are using your flashlight(which the batteries don't die in btw).

Not sure why you can't see anything, I don't turn up my gamma or brightness, once it starts looking grey instead of black I back it off, thats how the game is supposed to look, dark as hell. Of course the game is getting washed out in color if you crank up all the gamma and brightness/contrast. Turn off ALL lights in the room, turn the gamma/brightness up till it starts to look grey, then lower back till its between grey and black , and thats the highest you should raise it. I love the game and I do have a hard time seeing, thats why theres a flashlight. Just replaced my 9700pro with a 6800GT, WOW, huge difference, feels like the jump that was made when I first got the 9700pro to replace a GF4Ti4400, huge increase in performance.

I wonder if that's the same issue that I was experiencing with UT (patch 4.36), on my then-new Radeon 9200, with Cat 4.6 (ocfaq.org soft-mod version, to allow OC'ing). Rarely (usually on the first game start after boot), the game would be fine, but most of the time, the gamma was so dark, the game was virtually unplayable. Turning things up, just made the game kind of washed-out and grey. IMHO, ATI's recent Cats DO have a serious in-game gamma-ramp problem, for whatever reason. Possibly that is what he was seeing here?

I know that UT isn't exactly the most brightly-colored game in the world, especially on some of the user-made sniper levels, but what I saw was definately a driver gamma-table bug, I practically couldn't see *anything* in front of me.

 

[Jeff7181]

I'm playing in a dark room on a ~six month old NEC FP2141SB-BK- perhaps your monitor isn't displaying it properly, I have the best gaming monitor you can buy

I beg to differ if you can't see the game

And the "in game brightness" is whatever it was by default... my monitor's brightness is at 35 (goes up to 100), and the contrast is at 70 (also goes up to 100). I adjust my brightness by turning it all the way down, then raising it until I start to see black parts of the screen significantly differ from the little black border around the viewable area. I also haven't touched the gamma... in game or otherwise.

Again I say, you are not supposed to be able to see well in Doom 3. That is what the game is about... you can't see... the compound on Mars is damaged... if you listen to all the dialog with NPC's in the game, you learn that they've been having power problems for a long time... so it it such a surprise that after a gate to hell is opened and parts of the compound are destroyed that some of the lights may not work? And is it such a surprise that in a compound with hardly any windows, when the lights are out it's dark???

I don't understand why people are complaining about the darkness of Doom 3... EVEN after they're told repeatedly it's supposed to be dark. It just blows my mind that people are told this over and over and over and over and over and still don't understand it...

*EDIT* I'm going to use my digital camera, fix the shutter speed and aperature and ISO and all that crap... take a picture of these forums, then take a picture in game and show you the difference... It is not too dark to see... nor is it too bright.

here

I adjusted it so it's as close as possible to what I'm seeing while playing... then used the same settings to take a picture of these forums... Doom 3 is quite a bit darker if you notice, lol. It's SUPPOSED to be that way. Only thing I can't take into consideration is how your eyes would adjust for a mostly black screen as opposed to a mostly white screen.

 

[CVSiN]

yes realistically speaking a flashlight does lighten up the room a bit more, realistically you won't find demons from Hell hiding in the dark waiting to skin you alive.

And you have contacted the devil about this and gotten a list of approved commando zombie/demon tactics? is that like the spec war doctrine of the demon community?

LOL
If I was a demon id wait in the dark for some idiot to pass by =P

 

[Jeff7181]

Originally posted by: CVSiN
yes realistically speaking a flashlight does lighten up the room a bit more, realistically you won't find demons from Hell hiding in the dark waiting to skin you alive.

And you have contacted the devil about this and gotten a list of approved commando zombie/demon tactics? is that like the spec war doctrine of the demon community?

LOL
If I was a demon id wait in the dark for some idiot to pass by =P

Did you know you can hide in the dark from them too? There's one part near the beginning where you enter a dark room, there's only a couple light sources on one side that are pretty dim... if you turn on your flash light and get them to come out of hiding, then turn it off and go run to a corner... you hear them walking around groaning, but they don't come straight for you... if you flick the flashlight back on, you see them turn toward you again and start coming after you. Imps may be able to find you in the dark... but zombies can't.

 

[Jeff7181]

Here's some examples of Far Cry's ragdoll physics... now... I've never shot a dead body laying on the ground... but I don't think this is what would happen...

Doing 180's in the sand from being shot in the leg???

Hmmm... possibly the dock is a bit icy? Maybe it has a fresh coat of Thompsons on it so it's real slippery...

Ok... this looks pretty real, the ragdoll effects... but what's up with the gun sliding down the sandy hill?

Hmmmm... the body slides uphill in the sand when being shot...

 

[stardust]

Well Jeff, I kinda liked Farcry's almost silly ragdoll physics.. since a game isn't reality, breaking a few laws of physics here and there could be amusing..

I'm sure you enjoyed shooting the arms or legs of one of them dead mercs and watching them fly

 

[Jeff7181]

Originally posted by: stardust
Well Jeff, I kinda liked Farcry's almost silly ragdoll physics.. since a game isn't reality, breaking a few laws of physics here and there could be amusing..

I'm sure you enjoyed shooting the arms or legs of one of them dead mercs and watching them fly

Of course... but my point is that they're no more realistic than Doom 3, even though Doom 3 is less of a simulation than Far Cry. Can you imagine how people would bitch if they used physics based on the gravity of Mars? My god...

 

[stardust]

You could already jump on a kitchen desk with considerably heavy armor... maybe we need someone to write another Doom3 bible so everyone will stop their biznatching

I don't know if this was brought up already, but when I played Doom3 Alpha, the game was considerably brighter with no "absolutely" black areas and the performance was horrible especially when you used your flashlight. Maybe the "darkness" in Doom3 is made to enhance performance so many people could play the game at an above 30fps average. (meaning not as many rendered surfaces)

 

[Runamile]

Can you imagine how people would bitch if they used physics based on the gravity of Mars? My god...

D4mn, never thought of that! Mars has 1/3 the gravity of Earth. That means TECHNICALLY if the game had a realisic physics engine, you should be able to jump over 10 foot high walls and across 20 foot chasms. But I'm glad they didn't.

Edit: AND the bodies really WOULD fly 20 feet back after being blown away by a shotgun.

 

[BenSkywalker]

Jeff

And the "in game brightness" is whatever it was by default... my monitor's brightness is at 35 (goes up to 100), and the contrast is at 70 (also goes up to 100). I adjust my brightness by turning it all the way down, then raising it until I start to see black parts of the screen significantly differ from the little black border around the viewable area. I also haven't touched the gamma... in game or otherwise.

That explains a lot, your monitor is way overbright by default(your screenshots really back this up). Which calibration software are you using? With the screen as incredibly overbright as you have it it's no wonder you don't have an issue with how dark D3 is. On that screenshot you have in game on my setup you see nothing outside of a very small area in the back by the lights. I don't mean it's hard to see, I mean you don't see anything. This is on a properly calibrated setup though, not insanely overbright. Actually, I have to back off of that a little as I am using SuperBright settings in game which does puch my setup to brighter levels then it is supposed to be.

 

[Jeff7181]

Also... the flashlight actually DOES seem to reflect in mirrors... but only for a fraction of a second. Maybe this is a bug in the game that once you get the flashlight out for some reason it doesn't get reflected into the rest of the room like it does while you're raising it up?

I think the argument that it would be detrimental to frame rates can be dispelled by these pictures as well...

Flashlight being raised
Flashlight in use after being raised

 

[Jeff7181]

Originally posted by: BenSkywalker
Jeff

And the "in game brightness" is whatever it was by default... my monitor's brightness is at 35 (goes up to 100), and the contrast is at 70 (also goes up to 100). I adjust my brightness by turning it all the way down, then raising it until I start to see black parts of the screen significantly differ from the little black border around the viewable area. I also haven't touched the gamma... in game or otherwise.

That explains a lot, your monitor is way overbright by default(your screenshots really back this up). Which calibration software are you using? With the screen as incredibly overbright as you have it it's no wonder you don't have an issue with how dark D3 is. On that screenshot you have in game on my setup you see nothing outside of a very small area in the back by the lights. I don't mean it's hard to see, I mean you don't see anything. This is on a properly calibrated setup though, not insanely overbright. Actually, I have to back off of that a little as I am using SuperBright settings in game which does puch my setup to brighter levels then it is supposed to be.

I don't use any calibration software, I go by what looks right to me. Every other game looks right to me... Far Cry... Morrowind... Soldier of Fortune 2... Painkiller... Madden 2004... Ghost Recon... BF1942 (before my boycott started)... Counter-Strike... RBS3:RS... Serious Sam... etc. etc.

None of them look overbright, nor does Doom 3. That being said... if what you say is true, then EVERY PC game is designed wrong so that it looks fine on my monitor, but not on your's, which is calibrated "correctly."

 

[BenSkywalker]

I don't use any calibration software, I go by what looks right to me. Every other game looks right to me... Far Cry... Morrowind... Soldier of Fortune 2... Painkiller... Madden 2004... Ghost Recon... BF1942 (before my boycott started)... Counter-Strike... RBS3:RS... Serious Sam... etc. etc.

What looks right to you is very overbright, that is likely why you don't have an issue with D3. On a properly calibrated display you don't see anything at all throughout a good chunk of the game. Of course the game is supposed to be dark, everyone knows that. The issue a lot of people are having is the inability to see anything.

None of them look overbright, nor does Doom 3.

Of course they don't if you like them overbright anyway.

That being said... if what you say is true, then EVERY PC game is designed wrong so that it looks fine on my monitor, but not on your's, which is calibrated "correctly."

Nothing remotely close to that actually. I don't like things overbright so I don't ever have them or want them overbright. As far as putting "correctly" in quotes, there are industry standards for calibration and there are numerous different calibration applications- it is not MY definition of what the correct way to calibrate a display is.

 

[Shinei]

Maybe the "darkness" in Doom3 is made to enhance performance so many people could play the game at an above 30fps average. (meaning not as many rendered surfaces)

Actually, this is true; there are console commands in Doom 3 that cull polygons that are entirely shadowed so you don't waste time rendering what you can't see. It's why your framerates will jump from 15 to 60 when you walk into a completely darkened room (assuming you have low-end hardware like I do). Of course, light interactions are also a factor in the game, given the fact that if the game decides to set r_skipInteractions to 0 you get a constant 60fps in any environment (though you LITERALLY can't see anything, because there are no lights, including the flashlight).

And that screenshot of the flashlight shows what you're looking at; the flashlight is illuminating the space behind you while you're pulling the flashlight out. Look carefully at the marine's hand and which way the flashlight is facing in the first screenshot. On the other hand, even self-illuminating items like the plasma rifle fail to light the marine model, even though it will do so with the environment and other models (along with generating your shadow if you enable it).

Lastly, the ragdoll physics are fine, it's a lot less slippery than FarCry/Max Payne 2, but it includes a more realistic interaction system. Clearly, if you're not applying a tremendous force to a heavy mass, the mass will not be moving very quickly anywhere; the only weapons in Doom 3 that apply massive force with tiny impulse times are the grenades and rockets, which is why they can literally bowl a Mancubus over, while if you just shoot them with the chaingun they kind of fall on their backs disappointingly. gg high school physics
Of course, there isn't much environmental interaction, such as being able to blow doors off their hinges with rockets, but that's what mods and games based on the engine are for.

 

[BFG10K]

You brought up FC's foliage as being impressive, I'm pointing out it looks extremely poor in comparison to a title that predates it by some time.

I haven't seen KOTOR so I can't comment. I have seen D3 and FC and I can comment that FC blows D3 away in terms of outside areas.

Of course, most tiles are extremely dull until they have something shiny put on top of them.

Yeah, like a flashlight. To suggest that smooth surfaces like tiles are no more reflective than rough surfaces like wallpaper is simply ludicrous.

On a Voodoo1? Give me a break.

No, not on a Voodoo1. And I fail to see how obsolete hardware is related to the fact that Q3 was still basically the best looking game out there, not to mention that it demolished a lot of uglier games in the performance department.

The 'optimizations' they were running on the 9600 series boards for so long behind the users back and in direct contradiction to what the app requested.

I don't mind the optimizations but ATi should've told people about them and also put in a switch to control them. Anyway, my main gripe with ATi is their post 3.10 drivers.

R9700Pro is it not?

6800U.

He stated that minor changes made to the code made it fall off the fast path.

I've read your comment but I'm not responding because I don't want to rehash the arguments of old.

The NV40 loses zixel fill although it is still operating as a 32x0 part.

If it stays in 32x0 mode I don't see any reason why some shaders can't do zixel operations while others are doing MSAA operations. If the two sub-units in each shaders can both perform either operation then it's basically a given.

 

[CaiNaM]

Originally posted by: BFG10K
6800U.

you have a 6800u and you turn shadows off for performance?

 

[BFG10K]

I haven't tried as I've only had the card for a few days. My performance comments were related to the 9700 Pro.

 

[CaiNaM]

Originally posted by: BFG10K
I haven't tried as I've only had the card for a few days. My performance comments were related to the 9700 Pro.

ahh.. ok, that makes sense

i just popped in a GT last night; you should be able to run 1600 with all effects, 2xaa 8xaf and still easily double the performance of your 9700pro!

 

[BenSkywalker]

Yeah, like a flashlight. To suggest that smooth surfaces like tiles are no more reflective than rough surfaces like wallpaper is simply ludicrous.

Something shiny like wax, the tiles themselves have no reflective properties. I'm not the only one who has stated this in this thread, tiles themselves are not shiny.

No, not on a Voodoo1. And I fail to see how obsolete hardware

Fastest hardware two years prior to Quake3's release- Voodoo1. Fastest hardware two years prior to D3's release- R9700Pro which is what you were making comments based on. You were running obsolete hardware and complaining about the performance of the engine.

Q3 was still basically the best looking game out there, not to mention that it demolished a lot of uglier games in the performance department.

And numerous people back in the Q3 days claimed UT looked better, much as there are people like yourself claiming now that FC looks better. I really don't see how, but it is the same as it ever was.

I don't mind the optimizations but ATi should've told people about them and also put in a switch to control them. Anyway, my main gripe with ATi is their post 3.10 drivers.

None of them have worked properly with Sacrifice yet, my next part will not be ATi as they still really still can't hold a candle to nV's driver team. As far as a switch, it was previously your stance that doing exactly what they have been doing it is a cheat. I don't think it is, just makes their already poor filtering that much worse.

6800U

Nice, then turn the fvckin shadows back on When did you pick that up?

If it stays in 32x0 mode I don't see any reason why some shaders can't do zixel operations while others are doing MSAA operations. If the two sub-units in each shaders can both perform either operation then it's basically a given.

They are doing either zixel or MSAA functions, but enabling MSAA chews up zixel fillrate, that's what I've been saying. If you enable 2x MSAA you only leave yourself a 16x0 part as the other functional units are being used for z checks.

 

[BFG10K]

i just popped in a GT last night; you should be able to run 1600 with all effects, 2xaa 8xaf and still easily double the performance of your 9700pro!

The card is faster at 1600x1200 16xtri AF than my 9700 Pro was at 800x600 16xbi AF.

Something shiny like wax, the tiles themselves have no reflective properties. I'm not the only one who has stated this in this thread, tiles themselves are not shiny.

Smooth surfaces are inherently shinier than rough surfaces. It's a simple property of light & physics.

Fastest hardware two years prior to D3's release- R9700Pro which is what you were making comments based on..

I was also commenting on FC which is in exactly the same boat. It not only runs faster but its outdoor areas look better than what D3 can provide.

And numerous people back in the Q3 days claimed UT looked better, much as there are people like yourself claiming now that FC looks better.

(1) There were more games around that just UT that Q3 demolished.
(2) Even if UT looked better it couldn't hold a candle to Q3's performance.

As far as a switch, it was previously your stance that doing exactly what they have been doing it is a cheat.

Doing what was a cheat? AF optmizations? Like I said I'm fine with that (I use the tri & AF optimizations on my 6800U), I just take issue with ATi not telling anyone and not allowing control of them.

I don't think it is, just makes their already poor filtering that much worse

nV has both trilinear and AF optimizations.

When did you pick that up?

A few days ago.

They are doing either zixel or MSAA functions, but enabling MSAA chews up zixel fillrate,

Each shader can do two of them per clock so I fail to see how it then ceases to be a 32x0 part. If the shader had to do standard pixel operations then yeah, it can only do one of those and then it'd be a 16x1 part.

 

[BenSkywalker]

Smooth surfaces are inherently shinier than rough surfaces. It's a simple property of light & physics.

Most tiles here aren't all that smooth.

I was also commenting on FC which is in exactly the same boat. It not only runs faster but its outdoor areas look better than what D3 can provide.

Can provide...? You think FC's exterior levels look better then Hell? I sure don't.

(1) There were more games around that just UT that Q3 demolished.
(2) Even if UT looked better it couldn't hold a candle to Q3's performance.

You thought that back then though, a lot of people didn't agree with you at the time.

nV has both trilinear and AF optimizations.

One of the big reasons I'm not planning on upgrading to nV. I find it very sad that nV decided to go the ATi route of lowering quality for increased speed, they didn't fall into that with 3dfx but decided to now.

A few days ago.

Congrats, has it helped changed your perspective at all on D3?

Each shader can do two of them per clock so I fail to see how it then ceases to be a 32x0 part. If the shader had to do standard pixel operations then yeah, it can only do one of those and then it'd be a 16x1 part.

It chews up zixel fill has been what I've been saying. It can't perform 32 stencil ops per clock if AA is enabled.

 

[BFG10K]

You think FC's exterior levels look better then Hell?

I don't think Hell provides anything equal to FC's rippling water and lapping waves.

You thought that back then though, a lot of people didn't agree with you at the time.

Regardless of who thought what, Quake III ran far faster than UT did.

One of the big reasons I'm not planning on upgrading to nV.

There are options to control both in the drivers on a per-game basis if you like, but I leave them both. The IQ is still very good and the performance gain is nice.

Congrats, has it helped changed your perspective at all on D3?

I've been benchmarking the crap out of the card so I haven't really had a chance to play much of anything yet. D3 does look nice at 1600x1200 no doubt about it, and at 65 FPS it's faster than my 50 FPS @ 800x600 that my 9700 Pro used to get.

It can't perform 32 stencil ops per clock if AA is enabled

Ah, I see the confusion now. You took my comment to mean both but what I meant is that it simply stays as a 32x0 part (i.e each shader can do two zixel or two MSAA ops per cycle).

I did not mean that each shader can do two zixel + two MSAA operations for a total four operations per clock.