AMD Cinema 2.0 tech demo: real-time photo-realistic human models

Well:
Theory 1 is that the Cinema 2.0 tech is the OTOY stuff done by the guy in the Youtube videos where he describes how it works.

Theory 2 is that the Cinema 2.0 stuff is something else.

The last several posts have been the same as saying:
Theory 1 is the correct answer, here is some proof.

Theory 2, purple.

Then talking about how purple is cool.

 

Yeah I see what you are saying. Its this vid that puts me off.

http://www.youtube.com/watch?v=ROAJMfeRGD4&fmt=18

when he looks around the scene he only ever zooms in or looks around, the camera never leaves its location point. That was the only reason I put the second theory forward.

Its not quite as random or irrelevant as purple... thanks.

EDIT: Yeah ok theory 1 it is then, having checked out some more otoy footage especially of the transformers stuff, I now am just about believing that the ruby cinema 2.0 is the real deal; A fully interactive world. I guess I just found it too hard to believe, it is mind blowing stuff. Now I need to see it with my own eyes!

 

Well, it's, sort of, a fully interactive world, kind of...

I do believe the guy that presented that demo mentioned that like a movie set, all the building fronts are facades. So if you were to turn the corner at the wrong place, you'd see the back of the facade...

Still pretty damned impressive. And I'd love to see some adventure games made using cinema 2.0. Unfortunately until Nvidia can also do this, it's not going to gain much traction in the gaming space...

And I'd LOVE to see the return of voxels in games. I still remember the very first Commanche game back in the DOS days. And thinking of how much potential voxels had back then.

Regards,
SB

 

http://www.youtube.com/watch?v=bLiX...ly-cel-mai-real-personaj-creat-calculator.htm


Just WOW!

 

Ha, this is the video I wanted to see. It turns out there's very little CGI in it:

http://www.vfxworld.com/?atype=articles&id=3723&page=1

Still, damn nice.

Jawed

 

Truly amazing links nagus and Jawed. Thanks for sharing!

 

Just to make something clear - the video of Emily has the face rendered by Radeon HD's in realtime, but the rest is normal video feed from the original capture? (if it's not rendered by Radeon(s), why is it here?)

 

I'd noticed Paul Debevec being referenced before in connection with this but not quite clicked.

If you look at his site, looks like he's been developing a bunch of the science side of it over the last few years, then these OTOY guys have done a sort of rthdribl equivalent, realtiming the rendering theory.

http://www.debevec.org/

 

I believe I read somewhere that they licensed the Lightstage technology from Debevec, but I'm not 100% certain.

 

The tech is Lightstage indeed.


Parthenon is another SIGGRAPH paper IIRC. ATI must love 'em.

 

Key bits from the New York demo last week:

http://www.youtube.com/watch?v=x5yVjHaJ0PI

http://www.youtube.com/watch?v=XnihU4zCXe8

Jawed

 

And here is a link to TechCrunch shamelessly stolen from the forums at ompf.org with the best technical description I've seen so far.

http://www.techcrunch.com/2008/08/2...ealistic-3d-world-and-otoys-rendering-engine/

From what I can read is that reflections, among other parameters, are pre-baked which isn't very surprising.
Would be nice to know know how they handle dynamic reflections too, because there sure are reflections of Ruby in the Ruby demo.

Creds to ompf for digging it up:
http://ompf.org/forum/viewtopic.php?f=6&t=882&start=150#p9360

 

So, what's the definition of a static scene for the purposes of this set of upto 252 viewpoints?

Jawed

 

so whats the beef then? Hasnt the demo actually been released yet? I thought it was supposed to be out by now?