Doom 3 rendering paths difference

Bjorn

Veteran
I know what you're thinking, "Noo, not another Doom 3 rendering thread" :)

But, i have to :) , I just read this with regards to the different rendering paths (thanks to Uttar) in the Dawn demo:

There's very little difference even between FX12 and FP32 so in order to spot the differences I need to grab shots at the same position and right now I don't know how to do that.

Now, if it's hard to spot the difference between FX12 and FP32, how hard will it be to do the same for the different paths in Doom 3 ?

Will it be something like:

FP32 vs FP24: Not unless you magnify the screenshots 10X
FP24 vs FP 16: Now we only need to magnify the screenshots 2X :) to see the difference
FP 16 vs FX12: Noticeable difference though perhaps not while playing the game

Or what do you think ?
 
I think the point is that Doom3 only actually uses on shader for the lighting, and only about 9 instructions at that. With different kinds of shaders and different types of calculations precision will become a much greater issue. Doom3 isn't a game designed to make use of high precision, its a DX7 class multipass title with some frills on top.

Edit: Oh, and, errr...
http://www.beyond3d.com/interviews/jcnv30r300/index.php?p=2
 
DaveBaumann said:
I think the point is that Doom3 only actually uses on shader for the lighting, and only about 9 instructions at that. With different kinds of shaders and different types of calculations precision will become a much greater issue. Doom3 isn't a game designed to make use of high precision, its a DX7 class multipass title with some frills on top.

Edit: Oh, and, errr...
http://www.beyond3d.com/interviews/jcnv30r300/index.php?p=2

With regards to the interview, I had a disagreement with Joe where i thought that Carmack was more sensitive to those issues then the 3D community. We'll see who ends up right in that regard (i just hope that i can admit that i was wrong if it turns out that way :)).

Anyway, that's part of the reason why i brought this up, if it's difficult (can't judge that for myself until i see some screenshots) to spot the difference in the Dawn demo, even between FX12 and FP 32, then how hard will it be in Doom 3 which is much less (at least AFAIK) shader intensive then the Dawn demo ?

Edit: The Dawn demo is made by Nvidia though so it's of course hand tuned for the NV3X series which might not make it the best thing for comparing these things. But i would still be interested in seeing the difference with some screenshots.
 
K.I.L.E.R said:
Doom3 isn't a game designed to make use of high precision, its a DX7 class multipass title with some frills on top.

You are joking right?
I think it's a matter of perspective.

What, in your opinion KILER, would you disagree with Dave's opinion? What are the features of Doom3 that you know, KILER?
 
I would have thought that the use even simple shaders would have made it more of a DX8 class product, even though it relies more on traditional - read old fashioned!? - rendering techniques.
 
Reverend said:
K.I.L.E.R said:
Doom3 isn't a game designed to make use of high precision, its a DX7 class multipass title with some frills on top.

You are joking right?
I think it's a matter of perspective.

What, in your opinion KILER, would you disagree with Dave's opinion? What are the features of Doom3 that you know, KILER?

NURBS?
Being completely done on videocard, not CPU. I have yet to see a card capable of doing it 100% in hardware. Doom 3 supports it.
 
K.I.L.E.R said:
Doom3 isn't a game designed to make use of high precision, its a DX7 class multipass title with some frills on top.

You are joking right?

Not to speak for Dave B, but no, he's not. Carmack made it very clear when announcing the rendering model for D3 that the baseline functionality was GeForce1 level. It takes five passes, but it outputs the same result (minus slight loss of precision due to so much multipassing) any other card does, except for the floating-point precision you get on a DX9-level card.
 
Thought about my above post and felt that it may come out sorta like undermining KILER's knowledge of Doom3, so I thought I'd just say what I think Doom3 -- the game and the engine -- is about.

First of all, I tried to look for links to what Carmack said during QuakeCon 2002 but I couldn't (or didn't have the patience, really... I originally read it at Gamespy's) but I suggest you try to look for it as it contains relevant things wrt what he said about Doom3.

Doom3 is originally based on the original GeForce technology. That's DX7. The hardware scene progressed but Carmack already knows in what direction that progress is moving towards. He implements additional stuff. FP support, for one, which is a definite DX9 feature, and probably the most relative one in Doom3 when it comes to "DX9 features". Two-sided stencil is another.

IMO Doom3 the game will be mostly impressive looking due to its per-pixel (DP3 BM) lighting, clever usage of vertex programs to set up fragment programs, good model skinning and good looking textures that "falls in" with the game's story. That E3 gameplay movie had many gushing accolades -- I was personally underwhelmed but I will not make any definite judgement based on a movie... differences in fidelity between that movie and actual game experience (with far more pixels) makes sure of that. Stencil shadows? It's definitely cool and I have no doubt it adds to the immersion factor... but I don't look at it as a great 3D feature per se. It is probably, however, an important factor in his engine configuration (wrt performance).

Now, Doom3 as an engine is an entirely different animal. Depending on the schedule of the engine's licensees' games, we may yet see better looking games than Doom3 (the game). The engine itself is about the best possible way (wrt current DX9 hardware + OGL) to generate shadow volumes within a world that changes dynamically, about BSP building vis-a-vis vertices, about overcoming problems associated with static lighting scenes when we have shadow volumes, about using 9 (!) textures. None of these are the concerns of OGL and vertex/fragment program innovations but very smart engine creation with a set of priorities in mind.

Doom3, the engine, is mostly about register combiners (plus vertex programs) at the hardware level and what you do with them. I think Dave's comments refers to the use of DP3 BM and stencil shadows in the game, which aren't >DX7 and which I agree is considered "frills" in presence of DX9.
 
Kiler:

Doom3 doesn't support NURBS. No hardware today (or tomorrow) supports it either, so having it in the game would be pointless.

Doom3 *IS* for all intents and purposes a DX7-class title. All that you need to render its graphics is available in DX7. Stencils, dot3 bumpmapping, multitexturing, cube maps. That's it, man.

Pixel shaders just mean you get more speed and higher quality, but you'd get full functionality on a DX7-compliant card.


*G*
 
IIRC They use NURBS in level design only, with everything being converted to triangles on the way out. Even if it wasn't, all the NURBS tessellation and triangle conversions would have to be done on the CPU side for all cards, thus still making it a DX7 game/engine (but I don't believe specular highlights stay on for the DX7 class render path).
 
Ilfirin said:
IIRC They use NURBS in level design only, with everything being converted to triangles on the way out. Even if it wasn't, all the NURBS tessellation and triangle conversions would have to be done on the CPU side for all cards, thus still making it a DX7 game/engine (but I don't believe specular highlights stay on for the DX7 class render path).

Ahh, finally an explanation. :)
Thanks.
 
Reverend said:
Stencil shadows? It's definitely cool and I have no doubt it adds to the immersion factor... but I don't look at it as a great 3D feature per se.
Having played a lot of Splinter Cell lately ( on xbox .. modded of course ;) ) im gonna have to say that realistic shadows indeed _are_ so very uttely cool. Its a completely new experience.
Previous game that i felt so immersed with was System Shock 2 due to very clever use of 3D sounds.

Ahem, btw, what was it again that were supposed to be the "new groundbreaking" engine features that D3 will bring to table ? Shadows ? Done already .. physics ? done already...
 
I think the use of normal maps to project so much detail onto the charachters, and the unified lighting model are the two "big" things.
 
Grall said:
Doom3 doesn't support NURBS. No hardware today (or tomorrow) supports it either, so having it in the game would be pointless.

The fact that no current neither next gen hardware supported NURBS don't prevent anyone from using them in an engine. Tesselation would be done on the CPU, that's all.

I know one game engine that uses what I would call high order surfaces (by contradiction to tesselated surfaces), that works well and is impressive.
(it uses fractals and some HOS AFAIK)

[And no it's not the DOOM III engine]
 
Ingenu said:
I know one game engine that uses what I would call high order surfaces (by contradiction to tesselated surfaces), that works well and is impressive.
(it uses fractals and some HOS AFAIK)

You are not thinking about the Quake3 engine that uses bezier curves to create curved surfaces are you?
 
Josiah said:
I think the use of normal maps to project so much detail onto the charachters, and the unified lighting model are the two "big" things.
Judging by current published media, more like unified darkness model with couple of spotlights. Imo Splinter Cell for instance does the lighting/shadows very well, and by current indications DoomIII will be adding nothing radically new.
As for normal maps, yeah, would be cool to have them in more widespread use, but they are no substitute for real geometry detail. Plus, properly normal-mapped games are there already.

Im just thinking that JC might have been right when he said that he sees his next true challenges in rocketry. He helped kickstart and revolutionize the PC graphics industry, now his trying to do the same with public space transportation. More power to him in his new endeavours.
 
no_way said:
Judging by current published media, more like unified darkness model with couple of spotlights. Imo Splinter Cell for instance does the lighting/shadows very well, and by current indications DoomIII will be adding nothing radically new.

:?:

As for normal maps, yeah, would be cool to have them in more widespread use, but they are no substitute for real geometry detail. Plus, properly normal-mapped games are there already.

Name me some of these properly normal mapped games please, just curious.
 
Back
Top