Question about this Ghost Recon3 opening clip...

[ralexand]

http://www.gametrailers.com/player.php?id=6044&type=mov

What would it take to get real-time images of this quality. What is current or for that matter next gen hardware lacking?

 

[Megadrive1988]

well, I was about to blurt out Xbox-3 or PS4.....

but, maybe all that will be needed to do those visuals in realtime gameplay + cutscene is Ghost Recon 4 on Xbox-2 (360) but coming late in its lifecycle as a 3rd or 4th generation piece of software that maximizes the 360's architecture.

I don't know if that will be possible though.

however notice the expansive view down that street into Mexico City, a few seconds into the video -- I think that could easily be done on pre-Xbox360 hardware. it wasnt even smooth.

 

[pakotlar]

im pretty sure that this video clip is real-time running on the alpha dev kits. at least that's what I believe was said about it. True? Don't know.

 

[Geeforcer]

im pretty sure that this video clip is real-time running on the alpha dev kits. at least that's what I believe was said about it. True? Don't know.

No, the intro video was not realtime.

 

[ralexand]

however notice the expansive view down that street into Mexico City, a few seconds into the video -- I think that could easily be done on pre-Xbox360 hardware. it wasnt even smooth.

I was most impressed by the sharpness of things.

 

[Tacitblue]

Yeah, I don't think we'll see anything like that for another couple generations of hardware, PC's included.......you'd need real TFLOPS on the CPU alone plus enough other factors to push consumer level systems into what's now considered true supercomputing power. Still 10 years ago I had an old XT......I still chuckle when i think about it when i compare it to my modest machine i have today.

"640k out to be enough for anybody" attributable to Bill Gates, seems laughable today, but my P4 chugs on some programs even with half a gig of RAM.

Ahhh progress.

 

[Megadrive1988]

im pretty sure that this video clip is real-time running on the alpha dev kits. at least that's what I believe was said about it. True? Don't know.

No, the intro video was not realtime.

see, I knew it.

 

[Megadrive1988]

Yeah, I don't think we'll see anything like that for another couple generations of hardware, PC's included.......you'd need real TFLOPS on the CPU alone plus enough other factors to push consumer level systems into what's now considered true supercomputing power. Still 10 years ago I had an old XT......I still chuckle when i think about it when i compare it to my modest machine i have today.

"640k out to be enough for anybody" attributable to Bill Gates, seems laughable today, but my P4 chugs on some programs even with half a gig of RAM.

Ahhh progress.

a real TFLOP (of single precision floating point) of CPU performance should be no problem for Xbox-3's and PS4's CPU. actually we should have a few TFLOPs of real CPU power on those machines plus a few dozen "TFLOPs" on their graphics processors

 

[ralexand]

Anyone have an idea where these flops will go to in relationship to the video? Will the flops allow for a more advanced renderer?

 

[Bohdy]

You see, the thing with FLOPS is that they only come into effect when a promising game turns out to suck arse.

If it is bad enough, then it is known as a "TFlop".

Anyway, I don't think that trailer is realtime in the same way that the Killzone trailer is not currently realtime, but I do believe that both could well be realtime later on.

 

[Gollum]

LOL, so TFlop stands for Terrible Flop eh?

This is the first time I've seen this video, its obviously the pre-rendered intro of the game and I honestly don't think we're gonna see this quality this gen (just as I think we're not gonna see the Killzone concept's full visual quality). The character models and scenery are probably achievable in geometry and texture complexity, maybe even the animation, but lighting and surfacing just won't look that good in real-time anytime soon.

For years upon years we were stuck with simple phong-shaded single-layered textures! Pixelshading massively enhances real-time shading capabilities, but there's still a very distinctive gap between that and offline rendering. Don't even get me started on the complexity of the lighting in those scenes...

 

[mckmas8808]

I too think that its pre rendered. Its just the way it plays out it looks like an intro. But at the same time I think we are under playing what next generation games will look like. When Sony showed the Tekken and GT demo we would have never thought that games would have went from 'those demos' to what they are now. Just look at Splinter Cell, RE4, God of War, and GT4. They surpass those old demos by miles.

We are underestimating next gen games.
________
Hashish

 

[London Geezer]

I too think that its pre rendered. Its just the way it plays out it looks like an intro. But at the same time I think we are under playing what next generation games will look like. When Sony showed the Tekken and GT demo we would have never thought that games would have went from 'those demos' to what they are now. Just look at Splinter Cell, RE4, God of War, and GT4. They surpass those old demos by miles.

We are underestimating next gen games.

Nothing this generation surpassed the prerendered Raven demo showed before the Xbox was launched.
Or any of the obviously shamelessly pre-rendered material showed.

So yes, if these companies push it a little too far with the pre-release CGI, the final games will not surpass them.

The only demos that were surpassed by PS2 were the pre-release realtime demos, which looked like crap since not even Sony knew how the hell to code for the platform. And all the Tekken, GT2000 etc demos were all realtime, and later vastly surpassed by the real games.

I'm not too confident of what was shown prior to Xbox launch, apart from that prerendered Raven demo (the girl fighting with the big robot) and that still hasn't been surpassed, it was just too much.

So, we don't know yet, but there is the possibility that both the graphics from the KZ and this GR3 prerendered scenes will not be replicated by the final hardware, if Sony and MS have pushed it too far with the CGI.

We'll have to wait and see when we have the real machines.

 

[_phil_]

Most people (particulary on the technical side ) haven't the slightest idea how to weight art contribution to wow/impressive factor.So they give it all to tech.

Just take RE4 ,double texture size ,*4 the polygon count ,and output it 1080p 2xAA .This should run easily on x360 or ps3 at 60fps.

 

[aneep]

about the Raven demo
if somebody made a "game" with just a butch girl and a robot (i don't remember the background in the demo, prolly very much empty), i think they would replicate that demo pretty easily, don't u think?

-aneep-

 

[London Geezer]

about the Raven demo
if somebody made a "game" with just a butch girl and a robot (i don't remember the background in the demo, prolly very much empty), i think they would replicate that demo pretty easily, don't u think?

-aneep-

No.

Go watch the video again (the prerendered version, not the Geforce2 "adaptation") and if you know graphics, you'll understand why.

There are lots of games like you describe. They're called DOA, VF, Tekken, Soul Calibur etc.
And none of them, even the best looking, comes close to the Raven demo.

 

[Laa-Yosh]

Another important thing - besides lighting - that's going to remain unique to offline rendering is the superior quality of the images: shader and geometry and texture antialiasing. The very simple reason is that there's no room for compromise in CGI, so the artists will go back and push the sampling up again and again until it looks good and is free of any noise and aliasing. Studios can always buy more rendering machines if the client wants the job done fast. But in the hardware, you're limited by your existing resources to what you can do in 1/60th of a second...

 

[ralexand]

Another important thing - besides lighting - that's going to remain unique to offline rendering is the superior quality of the images: shader and geometry and texture antialiasing. The very simple reason is that there's no room for compromise in CGI, so the artists will go back and push the sampling up again and again until it looks good and is free of any noise and aliasing. Studios can always buy more rendering machines if the client wants the job done fast. But in the hardware, you're limited by your existing resources to what you can do in 1/60th of a second...

Interesting...so do you thing the biggest difference between rendered and real-time is the quality of the AA.

 

[Laa-Yosh]

Probably not the biggest difference, but a certain one. And it's not just edge aliasing, but a lot of shader aliasing and texture aliasing (the two usually walk hand in hand). Almost everything related to sampling is going to be a problem, because the hardware is always limited in how many samples it can take in a given timeframe.
It's also why I think that realtime global illumination, "real" subsurface scattering and other ray-tracing based techniques will most likely be too much for this generation as well. Whereas offline rendering is slowly moving on to capitalize on one or another implementation of the above - ambient occlusion for example is almost a must have by now.

 

[ralexand]

Probably not the biggest difference, but a certain one. And it's not just edge aliasing, but a lot of shader aliasing and texture aliasing (the two usually walk hand in hand). Almost everything related to sampling is going to be a problem, because the hardware is always limited in how many samples it can take in a given timeframe.
It's also why I think that realtime global illumination, "real" subsurface scattering and other ray-tracing based techniques will most likely be too much for this generation as well. Whereas offline rendering is slowly moving on to capitalize on one or another implementation of the above - ambient occlusion for example is almost a must have by now.

Thanks for the reply, I'll have to look up subsurface scattering since I don't know much about it.