Tim Sweeney on Real-Time Radiosity

[XxStratoMasterXx]

Well, I sent an email to Tim asking whether real-time radiosity would be viable in games, and here's what I got from him

Realtime radiosity will eventually be possible. You could implement physically inaccurate single-bounce approximations right now just by precomputing per-object irradiance; 15 years from now there will certainly be enough CPU and GPU power to implement arbitrary unconstrained radiosity. In the meantime, I expect we'll see a variety of approximations emerge.

 

[Wunderchu]

cool

 

[MasterBaiter]

So who sez that dev's won't talk to someone out of the blue eh? Just don't start e-stalking the poor man now!

 

[Diplo]

15 years is a hell of a long time in game terms. It's the difference between PacMan and Half Life 2...

 

[XxStratoMasterXx]

Hmm, I guess I was proven wrong

At least I know that if I have a really pressing question, then I'll ask him about it.

 

[KimB]

15 years is a hell of a long time in game terms. It's the difference between PacMan and Half Life 2...

Not quite. More like the difference between Super Mario 3 and Half Life 2 (or, more relevantly, Paper Mario: the Thousand Year Door). But still, yes, quite a long time.

 

[Blade]

Actually, now that gaming is already in an advanced state.. 15 years could seem even longer than it is.

The leap from current-gen to 15 years from now will likely be a much smaller leap than.. say.. Donkey Kong to Quake.

 

[KimB]

Well, as I've said before, trying to predict what computing will look like in 15 years is fairly unfounded. Current technologies just cannot be extended at current rates for that long. Thus, computing trends will change dramatically over the next few years. I don't know if anybody can accurately predict whether or not it is realistic for one of a number of competing technologies for the next wave of computing will come to the forefront, and therefore whether or not we will experience an abrupt jump in computing, or a move to faster acceleration of computing speed, or whether advancements in manufacturing will simply come to a near-standstill, with further advancements being made in relatively small steps through better software and more efficient hardware.

So, in short, it may well be true that computer games won't change as much in the next 15 years as they did in the past 15, but who knows? Perhaps some new technologies will come to light that will again dramatically alter the face of gaming (as 3D acceleration has done, for instance).

 

[Megadrive1988]

15 years is like the difference between the late summer 1989 launches of TurboGrafx-16 & Sega Genesis, and the launch of Doom 3.

but 15 years is meaningless.

it could happen in 5 years. or 25 years.

 

[Reverend]

Hmm, I guess I was proven wrong

At least I know that if I have a really pressing question, then I'll ask him about it.

Screw that other famous developer who wont ever answer a technical question of mine.

Wow, from idolizing JC whom you had never privately corresponded with before, to damning him when he doesn't answer your emails! You should dish it to JC "Screw you John for never answering my emails... I've got Tim Sweeney now!".

 

[KimB]

Any of these guys will answer an e-mail if they find it to be interesting. Just keep in mind that some of them get tons and tons of e-mails, and thus yours might just get overlooked if you send it at the wrong time.

 

[XxStratoMasterXx]

Rev, editied my previous post. It was rather childish of me to make my previous statement. They're both really cool, it's just it's very exciting to have "The Carmack of D3D" reply to your email about 10 minutes after you sent it

Also, back on the topic of radiosity. Do you guys agree with Sweeney? Disagree?

He also mentioned approximations that will happen until we get the raw computing power, what do you guys think we'll see in terms of these approximations?

 

[KimB]

Well, I doubt that the radiosity algorithm used in offline rendering will ever be made realtime, but similar-looking approximations of global illumination will make it to realtime rendering before too long.

 

[Reverend]

"The Carmack of D3D"

ROTFL... I gotta tell Tim this!

 

[Cryect]

Does that mean we can call Carmack the "Sweeney of OpenGL"?

Quality realtime radiosity should definately be possible in the near future (5 years) as long as the environment is restricted with small spaces and restricted so most the environment isn't moving. Prolly with some more improvements with Monte Carlo Radiosity we could see it sooner.

 

[arhra]

I remember seeing a few papers on doing photon mapping on GPUs over the past year or so. Here's one i found with a quick google search. Of course, an academic paper with a proof-of-concept using a simplistic scene at not-really-interactive framerates is a long way from a full game, but it's a start.

 

[3dcgi]

"The Carmack of D3D"

ROTFL... I gotta tell Tim this!

I remember reading a story about GDC a couple years ago where Sweeny was poking fun at himself and Nvidia by wearing a name tag reading "Carmack MX". I got a good laugh out of the story.

 

[Reverend]

Apparently, when Tim gets drunk, he asks you "Hey, wanna go have more fun?" while if John gets drunk, John says to you "I appear to have lost the ability to have full control of the capabilities of all my bodily functions and thus I am inclined to tell you I will call it a day and return to my abode".

 

[XxStratoMasterXx]

Apparently, when Tim gets drunk, he asks you "Hey, wanna go have more fun?" while if John gets drunk, John says to you "I appear to have lost the ability to have full control of the capabilities of all my bodily functions and thus I am inclined to tell you I will call it a day and return to my abode".

How would John have the mental capacity to realize that he has lost control of his bodily functions if he was drunk?

Tim really asks you "Hey do you wanna go have more fun?"??

Also, what I meant by "Carmack of D3D" is the fact that Carmack is arguably the best in the industry and Sweeney is also very good at his job and has much weight in the industry, plus he codes in D3D rather than OGL primarily.

 

[AlNom]

You guys crack me up.