Is Doom3's renderer revolutionary?

[Scali]

Scali, give us an example of a renderer that you consider a revolutionary point in the history of real-time (w/some reasons). A developer's idea of revolutionary may defer from that of an end consumer or enthusiast.

Well, I think Wolf3D is the best example. It uses raycasting, which ofcourse wasn't new... but it uses it in a way that was never done before, as far as I know... Namely, raycasting 'columns' of walls, with textures on them.

Another thing that could be considered revolutionary is the approach to bumpmapped radiosity in HalfLife2, or the offset-mapping in UE3.0, again because those were never done before, as far as I know.

 

[Scali]

Do you know what's common of all those four programs?

It's pretty much a certainty NONE of them would have been designed the way they are had not the people behind them seen the Doom3/Geforce3 presentation at that macworld fair in 2001.

While we never know for sure ofcourse, it is highly likely. I think the term 'trendsetter', or perhaps 'milestone' would fit Doom3 nicely (although I found this link: http://www.arscreat.com/cryptichon/html/shadowvolumes.html, was this an engine which applied shadowvolumes in a game before Doom3? It even had the projected flashlight too).
But that's not the same as 'revolutionary', I think.
That it is just a trend, should also already be obvious. The next generation will not be using stencil shadows anymore. Just look at 3DMark05 and UE3.0. This will probably be much of the same story. By the time UE3.0 is released, many others will have implemented a lot of the ideas from UE3.0.

That's right. 2001. You speak about this and that beating D3 to market, SO WHAT. D3 was still first. The D3 renderer existed pretty much in its final form THREE AND A HALF YEARS AGO, in the sense it used reverse stencils, bumpmapping, cubemaps, all that stuff. Later it was of course optimized, updated for DX9 etc. But the thing is, without that presentation, the smattering of per-pixel lighting/stencil shadow games we have would most likely have been ZILCH. Witness the power of "me-too"...

I think that may be pretty much the whole issue. Doom3's renderer was 'done' three and a half years ago, and never really evolved from there anymore, which allowed others to catch up, or even supercede Doom3, before its release (think about the offset-mapping mod for Doom3, this could have been a standard option).

 

[Diplo]

http://thesaurus.reference.com/search?q=obsessive

 

[DemoCoder]

Well, if Doom3 really kicked off the revolution, than I'd say the engine is revolutionary. Otherwise, you're just playing word games.

Look, we all know that if you do a SIGGRAPH search, you can find people who wrote papers on individual techniques well before Carmack and other game industry people. But sometimes it takes someone to build something exciting to convince the rest of the world that an idea is worthwhile.

 

[davepermen]

after the first public images of doom3, and thats now longtime ago, people started to analyze, and adopt the featureset they got seen, as it was, at this time, revolutionary for games.

since then, a lot of time passed by, and by today, about everyone has it's own little doom3 bumpmap effects made, and some stencil shadows.. this is logical, as everyone wants to copy a bit of the good stuff.

anyways, when you start doom3, and know it all yet, as you've seen it all yet, in demos, in games, in the videos, as everywhere everyone implemented it more or less successfully by himself, you notice one thing:

doom3 is still the one. it performs exceptional for the thing we saw that often yet crawling (tenebrae, etc..). it looks amazing even on low end systems at lowest resolutions (unlike those demos, wich plain suck), and it shows another revolutionary part of itself. the great visual impression it delivers, no mather that the effekts all are known since quite some while, it delivers them in a great way, high performing.

doom3 brought up a revolution. starting with doom3, people started to care about more than just polygon count. people started to investigate much more into pixelshaders, and they first got into the generic gamers mouth. without doom3, that whole graphics hype with shaders, bumpmapping, shadows, image quality, global illumination, per pixel lighting, lighting in general, etc, wouldn't have started off.

doom3 started a graphical revolution since it's official release. nowadays, it shows an awesome implementation of what it started.

 

[Scali]

Well, if Doom3 really kicked off the revolution, than I'd say the engine is revolutionary. Otherwise, you're just playing word games.

I think that brings up a parallel question: Is it a revolution that Doom3 kicked off, or is it a trend or milestone?

I see it as a trend, for various reasons... I can list a few:
- It is only temporary, and steered by the capabilities of hardware. We will have another trend in the next generation of games.
- It is not a completely new way of rendering things, but rather an extension of old concepts.

I suppose it's also a milestone because it's one of the first successful renderers with full selfshadowing and per-pixel lighting, which increases realism considerably.

As I said before, I think the visual impact skews the opinions, since the visual impact of applying shadows and bumpmapping on everything is far greater than the technological impact. As the no-option in the poll indicates, the ingredients for this visual impact have been around for decades.
I suppose that also explains why Doom3 has such a large impact on the gamers. To them it seems like a huge step, because they have no knowledge of the technology behind Doom3.

But sometimes it takes someone to build something exciting to convince the rest of the world that an idea is worthwhile.

Yes, but this brings up the question whether Doom3 is such a case or not. There have been a number of other realtime shadowvolume renderers before Doom3. So is it really Doom3 that convinced the rest of the world, or is it one of those earlier renderers that actually convinced Carmack (and others?) to do the same?
Or, the alternative option: is it so obvious that shadowvolumes are a worthwhile idea that no convincing is required at all?
I think at least the per-pixel lighting falls in the 'obvious' category.

 

[Goragoth]

Well, I think Wolf3D is the best example. It uses raycasting, which ofcourse wasn't new... but it uses it in a way that was never done before, as far as I know... Namely, raycasting 'columns' of walls, with textures on them.

Didn't id's own maze game (the name elludes me at the momement) do texture mapped ray casting? And what about the Ultima game that pioneered the technique at the same time? Besides, ray casting itself was an old technique by the time Wolf3d came out.

Another thing that could be considered revolutionary is the approach to bumpmapped radiosity in HalfLife2

I don't know a lot about this but what they've done sounds a lot like a simple extension of a SIGGRAPH paper from 1986: "A Radiosity Method for Non-Diffuse Environments". You're right though, I certainly would call what they've done revolutionary.

or the offset-mapping in UE3.0

UE3 isn't released yet and we already have many examples of offset mapping, it is a well known technique by now, and I would imagine that it has been implemented in software a long time ago. Humus' tech demo has demonstrated it certainly, so obviously UE3 won't be revolutionary, remember you have given credit to Doom3 only for when it was released, not when it first demonstrated the techniques (which was quite a while back).

What about Carmack's reverse? Was that revolutionary? It seems to me that this hasn't really been implemented before (I know, several people came up with it independantly around the same time but I would still give him some of the credit).

Otherwise, you're just playing word games.

I'm quite sure that this is all that is happening, in fact I have the feeling much of the arguing on these boards simply stems from the fact that people have different definitions for words they use. What is revolutionary? How unified does lighting have to be to be called unified? Oh well...

 

[Scali]

anyways, when you start doom3, and know it all yet, as you've seen it all yet, in demos, in games, in the videos, as everywhere everyone implemented it more or less successfully by himself, you notice one thing:

doom3 is still the one. it performs exceptional for the thing we saw that often yet crawling (tenebrae, etc..). it looks amazing even on low end systems at lowest resolutions (unlike those demos, wich plain suck), and it shows another revolutionary part of itself. the great visual impression it delivers, no mather that the effekts all are known since quite some while, it delivers them in a great way, high performing.

I suppose that's personal. On my PC, 3DMark03 runs a lot faster, even without any optimizations, and it also has more characters on screen, and the characters have higher polycount.
So to me, 3DMark03 is the one.

 

[Scali]

Didn't id's own maze game (the name elludes me at the momement) do texture mapped ray casting?

I don't know that game. But if there was such a game that started it, and wolf3d just improved that technology, then well, that game would be the revolutionary one, but the point remains. Whichever was the first, was revolutionary.

And what about the Ultima game that pioneered the technique at the same time?

From what I understood, Ultima doesn't use raycasting of wall 'columns' but generic polygon rasterization.

Besides, ray casting itself was an old technique by the time Wolf3d came out.

Obviously, but ray casting is just a primitive operation which can be applied to rendering. The same goes for triangle rasterization. But both can be applied in novel ways. For example, Wolf3D's way of raycasting walls was novel, as was the idea to use triangle rasterization of shadowvolumes.

I don't know a lot about this but what they've done sounds a lot like a simple extension of a SIGGRAPH paper from 1986: "A Radiosity Method for Non-Diffuse Environments".

I'm not familiar with that paper, so I cannot say if it's the same as what is done in HalfLife2 or not.

UE3 isn't released yet and we already have many examples of offset mapping, it is a well known technique by now, and I would imagine that it has been implemented in software a long time ago.

What I said was under the assumption that UE3.0 was the first, and demos such as Humus' one were based on what they saw in a UE3.0 demonstration.

Humus' tech demo has demonstrated it certainly, so obviously UE3 won't be revolutionary, remember you have given credit to Doom3 only for when it was released, not when it first demonstrated the techniques (which was quite a while back).

The difference being that per-pixel lighting and shadowvolumes were not new when Doom3 first showed them, while HL2's radiosity and UE3.0's offset mapping were (well, that is, assuming they were. I at least know that I've never seen them before).

What about Carmack's reverse? Was that revolutionary? It seems to me that this hasn't really been implemented before (I know, several people came up with it independantly around the same time but I would still give him some of the credit).

That is very subjective I suppose. It is such a simple, logical thing... on the other hand, it has been overlooked for years... And it does have quite an effect on the implementation and usefulness of shadowvolumes.
Strictly speaking it's a non-issue, because as Creative's patent made painfully obvious, Carmack wasn't the first.
If not revolutionary, the discovery of the reverse and the zfar-plane at infinity was certainly a milestone, because it was the last bit of the puzzle for robust and elegant shadowvolumes.

 

[Sxotty]

I suppose it's also a milestone because it's one of the first successful renderers with full selfshadowing and per-pixel lighting, which increases realism considerably.

As I said before, I think the visual impact skews the opinions, since the visual impact of applying shadows and bumpmapping on everything is far greater than the technological impact.
...
I suppose that also explains why Doom3 has such a large impact on the gamers. To them it seems like a huge step, because they have no knowledge of the technology behind Doom3.

I am not trying to flame read and consider a bit.

Which is more revolutionary.

A large increase in visual impact that has a profound effect on gamers

OR

A large increase in technology that doesn't have an impact on gamers.

You see a person playing a game doesn't care if everything has been done before, or if what is happening is stupendously complex, they care about how it looks. BTW your poll is obviously biased, though I did vote for no, but I still think it is silly and wont prove much.

 

[arjan de lumens]

As far as I can find, offset mapping in its current form was first shown in a tech demo by OpenGL.org forum member Mogumbo around mid-January this year, and spread rapidly after that when it became apparent how easy it was to implement (just 3 pixel shader instructions for a visually rather striking effect).

 

[Goragoth]

From what I understood, Ultima doesn't use raycasting of wall 'columns' but generic polygon rasterization.

Looks like I got my facts wrong on that one. I'm pretty sure that Catacomb 3-D was id's first game to do texture mapping + raycasting but obviously your point still stands and I completely agree that it was revolutionary.

What I said was under the assumption that UE3.0 was the first, and demos such as Humus' one were based on what they saw in a UE3.0 demonstration.

You might be right. I did a quick search to find out the origins of offset mapping and couldn't find anything. First I heard about it was to do with FarCry and by now there are several tech demos which is why I was so quick to dismiss it. It would be interesting to know who was first to describe it.

 

[DeanoC]

Revolutionary is subjective and based on each person own experiance and/or bias.

For the record, as I understand it (based on my biases ;-) )

HL2's radiosity light maps are cartesian formulation of 1st order diffuse irradiance maps which spherical harmonics were original used (actually HL2 are closer to the original cube map approximation). SH is generally perfered because it scales to higher frequencies easier than the cartesian formulation. But in HL2 case, its faster than the SH formulation due to the low order.

Offset mapping is an fairly old idea, the current version come to the public attention on the OpenGL forums (which is where the Unreal guys have said they first heard about it...). TomF used a similar idea in Startopia (a few years ago) to simulate bump-mapping on hardware that couldn't. His 'parallax mapping' shifted a texture to acheive some texture depth. The modern technique changes the parallax amount per-pixel but the idea is basically the same.

For me a revolution in CG will happen when we finally kick the vector habit and move to the tensor formulation. Blinn's only be banging on about it since 1977... (I moved recently and havenn't looked back!)

BTW Id's Catacomb3D predates Wolf3D

 

[Scali]

I am not trying to flame read and consider a bit.

Which is more revolutionary.

A large increase in visual impact that has a profound effect on gamers

OR

A large increase in technology that doesn't have an impact on gamers.

I think the first is a case of the GAME, the second is the case of the RENDERER.
We are not discussing the GAME, and this poll is not for gamers.

 

[Reverend]

UE3 isn't released yet and we already have many examples of offset mapping, it is a well known technique by now, and I would imagine that it has been implemented in software a long time ago.

What I said was under the assumption that UE3.0 was the first, and demos such as Humus' one were based on what they saw in a UE3.0 demonstration.

No, Sweeney didn't think of this first. When I first brought this up to him, he told me he saw [edit]something similar to[/edit] this in action when visiting another developer a few months before January 2004 (the time I emailed him about this) and Tim CC'ed Andrew Scheidecker (the guy handling UE3's renderer) in our email correspondence.

[edit] Link re post on OpenGL.org regarding offset mapping.

 

[KimB]

Well, fine, if you're going to talk about whether or not the renderer is revolutionary, you'd have to compare back when it was first shown to the public, back about three and a half years ago (since that hasn't changed much).

 

[Reverend]

HL2's radiosity light maps are cartesian formulation of 1st order diffuse irradiance maps which spherical harmonics were original used

Um, could you repeat that?

 

[Scali]

Well, fine, if you're going to talk about whether or not the renderer is revolutionary, you'd have to compare back when it was first shown to the public, back about three and a half years ago (since that hasn't changed much).

Does that make any kind of difference then?

 

[Luminescent]

Well, fine, if you're going to talk about whether or not the renderer is revolutionary, you'd have to compare back when it was first shown to the public, back about three and a half years ago (since that hasn't changed much).

Good point. Doom 3's renderer was ready many moons ago, but game content development, etc. slowed it. Are we going to penalize a renderer because it had to put up with the delays of voice acting, polish, content creation, etc. due to something aside from renderer development.

Finally, there is a major difference between the time a technique was discovered for offline rendering and when it was officially implemented in a complete (robust) solution, in-game. I figured that in this case we were referring to the first time a complete game renderer (not tech demo), implemented whatever technique in a fashion that made its technology readily acceassiable (playable with reasonble performance) and noticeable. I mean, we could render spectacular things with ray-tracing/offline techniques on current hardware, but whether it comes at an interactive framerate is another story.

Maybe we could agree on the prerequisite attributes that must be met for the consideration of a revolutionary game renderer (in this context).

I propose it mus be:

1. A solution that is the core rendering technology of a newly released game.

Argument: We can mod old games with rendering features, but a revolution will not commence if an old game has lost its impact. We can create tech demos, but they will suffer the same fate, and perhaps lack a robust interactive/playable solution. The whole point is to make the solution interactive for the purpose of a game.

2. A solution that creates a percievable visual impact and influences other engines

3. A solution that is accessiable (playable) and scaleable (a general expectation for pc game titles are expected to be).

 

[wzh100]

i dno't think doom3 is a renderer revolutionary of pc games ,but i am sure is revolutionary of pc hardware,bumpmapping,shadow volume......
we can use geforce 4 mx440 to shading it,but it too slow,we need nv40/r4xx,we need money too,doom3 made me poor,but it've never made my happy.i believe unreal3 will be a renderer revolutionary of
3d technology and hardware.this is my Standpoint