Apenheul said:
Point 1 has me confused.
Point 2 is bullshit.
Point 3, could do that on the gpu for some years already, this is just an alternative.
Point 4 is valid.
1 is something KK and others have discussed for a while. Using Cell-based STBs or TVs to allow you to manipulate something your watching in 3D. Hard to figure out exactly what they mean until they show something tangible. Of the 4 points, that's the one I find to be bs.
2 is most certainly not bullshit. They've already done this with the terrain renderer. There's no GPU. The Cell uses 7SPEs to raycast, and 1SPE to do JPEG compression on the image. It was then sent over a network connection to a G5, which put it to screen. The only difference being they used a wired connection instead of a wireless, but the concept is the same. This is why he mentions using H.264 instead, since you'll increase efficiency over a set connection. Believe.
3 can't be done on GPUs very well yet. The GP capabilities of most GPUs is paltry. Only very recently (like next-gen GPUs) have we approached the point where the FLOPS on a GPU can be practical for doing stuff like this. Even then, we're still not seeing it. He's talking about cloth physics, not cloth animation. Stuff like in SC and DOA isn't even scratching the surface computationally of what he's talking about. Then again, what he's talking about I don't see happening in any sports game. I'd like to be proven wrong, but I don't see it. But I think the lotsa ducks demo had some of this type of interaction when the cannonballs were being fired back and forth. But I'd be very suprised if those worthless hacks at EA ever tried anything that ambitious.
4 is related to 2, as I mentioned. I'd like it only for the possibilities of a next-gen flight sim that's connected to some satellite server that lets you fly anywhere in the world in a MMO atmosphere, with terrain looking as realistic as possible everywhere you go, and having accurate depth to boot. But...I don't think flight sims sell well enough for this to come to fruition though. Maybe Namco could do a limited version of this for the next-gen Ace Combat...maybe. PEACE.
EDIT: Not sure if you thought Point2 was bs b/c of latency reasons, or computational reasons. My bad if it's the former, I was referring to the latter. Yes, for latency reasons, it's gonna be limited to non-gaming apps, or games where responsiveness is not important.
EDIT2: Goddammit, gofreak already beat me to Point2.
Meh.
Whether it's more impressive visuals or more believable interaction between objects in the game world, anything that makes the experience more immersive can add intensity and contribute to the player's enjoyment. You can also build entertaining gameplay around a complex physics model, which is just as valid an experience as building a game around visual elements like lighting (as is the case in Doom 3). As far as I'm concerned, Carmack missed the mark with those comments, and you're doing the same when you echo them.
