John Carmacks Quakecon keynote

IMHO it is very important. For an example, imagine that someone designed a level which put some horrible monsters in shadow to scare the player. If there is no shadow on some slower machines (or it can be turned off), the desired effect would be gone. Other examples including using bump mapping to carve texts on walls, etc.

There could be some visual difference between newer and older hardwares (such as precision). However, the overall feeling should be the same.
 
Thowllly said:
What I'm wondering, is what the hell is that goat doing there? (at -4:11 and -2:40)

TsN (Team Sportscast Network) who was there streaming live. They were starting to get bored with his speech (I believe why the Q&A about an hour was cut off). There was around 1100 people having a fit that TsN wasn't coving the games going on at QuakeCon instead of the Keynote. The goat is "The Axis Goat" for one of the shoutcasters that handle RTCW matches. It was a request by several of the people several times, wanting the keynote coverage ended so the games could continue.

I asked TsN about making use of their second server next year. This way, they would be able to cover the other events at QuakeCon and still keep the gaming coverage going. Guess we will have to see what they do next year. The Q&A IMO had probably more info then the entire speech and was cut short. Ahh well, maybe next time.

To whoever recorded this. I'm glad to see you caught the beginning of the cast, I missed the very beginning because I was arguing with the people at TsN in their irc channel :D
 
Dave B(TotalVR) said:
Skip to five exactly minutes before the end...

What the hell is that whine he coms out with? what the hell is wrong with his voice? its baffling!

More TsN Humor. People were ripping on JC pretty bad. Same people playing his games and the reason they were at Quakecon. Shortly before the keynote, they (the IRC'ers some at QuakeCon on IRC) were like "All Hail The Carmack" etc. Until 10 minutes into the speech when their attention spans came to a halt and JC became the target of ignorance. He would be proud of his followers, I'm sure.
 
Did anybody catch the part where he talked about the next project? He point blank stated thathe would _definitely_ use a high-level Shading language for the next project.

He is also (and has been) calling on the hardware/graphics community to develop a Graphics API so he doesn't have to spend time optimizing for each card...

He did say that, for all intents and purposes, that 3DLabs/Cg are identical... but didn't specify/endorse anything right there and then...

So, this is almost surely going to be something we hear about in the not too distant future...

Given the fact that we now understand that NV30 is the platform he will be using for his next project...AND it would appear that Cg does provide what he's been calling for, I've got to believe that he will end up leaning towards CG...
 
I like that he is adding backends for all the different levels of hardware around. Not only will it help the game engine run optimally on said hardware but it will also speed the usage of the advanced features of some of this new DX9 stuff.

I wouldn't be surprised if Carmack added a few goodies for the R300 and NV30. :)

Sounds Like he's tempted already...

"I've got pretty good ideas of where I'm going with the next generation technology... for the most part my work on Doom, the significant contribution to it, is done. Set in stone. The renderer decisions are all made. We're adding a few minor tweaks here and there... I've had to add control for slightly different levels of shadow... the console shadow tweak... It's that level of tweaking and then of a bunch of optimization that 's left to do, and I'm getting a little tempted now to start peeling off and working on some next-generation technology generations to research some of the things."
 
Typedef Enum said:
Given the fact that we now understand that NV30 is the platform he will be using for his next project...AND it would appear that Cg does provide what he's been calling for, I've got to believe that he will end up leaning towards CG...

That's not my reading of it at all. His statement on "using the next gen cards" was very much that.... note cards.

He currently has a pro-nvidia bias for research due to historically much higher driver quality, so if they can hold that up through nv30 (maybe a big if given it is a new design) then nv30 will likely be his main work platform. However, the R9700 drivers are very fine, JC is pleased with them, and we may see a shift to ATI.

My interpretation is that he was calling for a consensus - he says that the difference between them are minor and implies that they should go away. I'm certain he won't go with anything that's not ARB-ratified, but that might not happen soon enough. Cg might get some early research done if it and nv30 are ready for the bigtime in time, but since he wants to start work soon, the only card he can use is R9700.

In the longer term I can't imagine him doing anything other than OpenGL 2.0 and whatever the ARB approve for the HLSL. This may be Cg or a modified version of it, of course.
 
That's not my reading of it at all. His statement on "using the next gen cards" was very much that.... note cards.

Didn't he state somewhere that the NV30 will be the basic platform for his next gen engine ?

In the longer term I can't imagine him doing anything other than OpenGL 2.0 and whatever the ARB approve for the HLSL. This may be Cg or a modified version of it, of course.

That would be my guess too. Carmack has never been a fan of propriety stuff so i don't think he would consider using Cg unless it became a standard. And (as you mentioned) he stated that the difference between them was so small that it wouldn't really matter which one the ARB choosed.

Btw, i thought that his talk about the next generation (or possible the one after that one :)) engine was pretty interesting. That graphics programmers might move towards being more of a "technical director" instead of a programmer because the engines will be so flexible ( i guess he was concentrating more on the rendering part here) that the reason to develop a new one will be very small. This is of course connected with the advances in the coming DX9 and beyond cards.
 
I remember him saying something along the lines of 'next gen of which the first is already here and then there's nv30'. But, I couldn't find it to remind myself. He did waffle on a lot!

More fundamentally he's saying that it doesn't matter what your hardware limits are, as long as your compiler can make it run on the hardware it doesn't matter - even nv30 will probably have some limits that make even long programs need multipass (e.g. will it support more than 16 textures?)

He's already said that the lighting in Doom3 has changed things a lot - in particular, the artists are using almost all spotlights to give them more flexibility, and the guys who came from movie CGI are pointing out this is a very similar process to what they are doing.
 
Didn't he state somewhere that the NV30 will be the basic platform for his next gen engine ?

I hear people say this over and over. But I've never heard JC say that. With that said, I think someone who misquoted him, ran with what they thought it was and then others started using it. Now, a lot of people thing it's NV30. But I don't believe that and of the thinking that he's going to use the 9700.

As for use of OGL 2.0, I think that's another misquote/summarization of the fact that he's making a P10 render path which will end up doing 7 passes rather than 10 per pixel, assuming 1 light and so on.
 
Regarding next gen engine and NV30:

The question is where do we go from here? The next logical step is to develop an engine that is designed exclusively around the technology that you will find in the new technology video cards such as the ATI Radeon 9700 and the Nvidia NV30 which has not been released yet. The sad part about this of course is that we might miss an intermediate step, but I think that is what is going to happen.

So, more like based on the DX9 featureset then the NV30 specifically.
On the other hand, it's going to be a completely HLSL made engine so
maybe that's obvious.

Another interesting quote:

From what I have seen so far the new 3D Labs cards are not going to be competitive with the highest end video cards. The card has good drivers however and the virtual texture mapper is good and I think other companies will adopt this. I don't think that it will be a card that will be adopted by consumers.

Not exactly what i wanted to hear regarding that card.
 
Is the intermediate step DX8?

The good side is someone very skilled will do a game with pervasive DX9 tech, the bad side is when will it be available? Dont hold your breadth :)
 
according to Dave Orton, next year Spring from ATI, and likely from NVIDIA as well. (regarding low cost DX9 cards)....
 
pascal said:
Is the intermediate step DX8?

I think he specifically mentioned that in his speech.
That there were some extra stuff that could be done for the R8500, GF3/4 class cards but that it required a lot of extra work for the artists so it was cancelled.
Although you'll still benefit by getting somewhat better quality lighting as i understand it. And of course speed.
 
The quote about Carmack/next-generation/NV30 is something you won't find because I _think_ it's going to end up being one of those NV30 launch/marketing deals.

Somewhere out there...there exists (I believe) some material containing this quote...It's not, however, public.
 
Somewhere out there...there exists (I believe) some material containing this quote...It's not, however, public.

Oh, no, thats public. The Quote goes:

My current work on Doom is designed around what was made possible on the original GeForce, and reaches an optimal implementation on the NV30. My next generation of work will be designed around what is made possible on the NV30."

And, yes, it is in with some of the NV30 details, however there is no reference as to when the quote was taken. Theres a fairly good chance that it was done at the NDA breifings last March or last year and we don't know in what context the quote was taken, or whether he was aware of what R300 consisted of.
 
Also, it states capabilities made available by NV30, that doesn't mean that these abilities won't be possessed or aren't possessed by current day cards.
 
I have to admit I was rather disappointed that the target platform for Doom3 was the Geforce1. I guess it make sense and I understand why, but I thought all this time the "target" was the Geforce3? I guess that's the optimal level? Well, at least now we might see hardwired TnL get some serious use.

I do find it weird that he's going straight to DX9 as a target platform, totally skipping over DX8. I really have to wonder about that from a business standpoint. I guess his viewpoint is that it's a waste of time to make an engine for DX 8.1 and then have to do another one for DX 9 which is more complete. On the other hand, does that imply that the engine will not support DX 8/8.1 cards? If so, there might be a period where id will take a hit in engine licensing. Not that they don't have the money to afford it, by any means.
 
My guess from technology standpoint the target is the GF1 but from performance standpoint the GF3 (acting like a fast GF1) is more realistic. See this interview:http://www.tomshardware.com/business/02q3/020817/quakecon-01.html
THG - What platform does Doom III target?

TW - Doom III is targeted mainly at the GeForce 3 series cards and the Radeon 8500 cards. Other cards will be able to run the software, but of course the graphics will be scaled down from those cards on the target platform.

THG - When looking at Doom III is the game more video card or CPU dependent for the best experience?

TW - Due to the complex nature of the graphics fidelity that we are trying to present in Doom III, the game is more video card dependent than CPU dependent.
And it is more GPU dependent.
 
Back
Top