Do you think we will get Killzone CG level graphics next gen?

posting from the Wii so i wont elaborate. my thinking is, YES, Xbox 2010 (Xbox720 or whatever) and PlayStation4 will most likely rival, even probably surpass the Killzone E3 2005 trailer. i'll write more later when i get back to a computer.


edited on PC.

it is technically impossible for either Xbox 360 or PS3 to reach MotorStorm or Killzone E3 2005 trailer graphics in realtime. even if portions of those trailers were realtime, and sped upto 60fps.

MotorStorm PS3 does not look nearly as good as the better parts of the E3 2005 trailer, not even close. the same will be true of Killzone. the RSX has severe limitations and even the Xenos does. therefore, even late-generation software on PS3/Xbox360 will not rival the Killzone trailer.

just like the Xbox1 never rivaled the Raven-Robot demo. heck, even Xbox360 and PS3 graphics are not rivaling the Raven-Robot demo in image quality (anti-aliasing, filtering, 3D motion blurr).

it's not just about polygon detail. or shaders. or any single thing. it's everything combined in those trailers that make it IMPOSSIBLE to achive on current consoles, even later in their lifetimes.

with a 5-10x increase in processing/graphics/RAM, the consoles of the early part of the next decade should be able to do the trailers we saw to hype the PS3 at E3 2005.

anyone who thinks Xbox 360 and PS3 will be able to reach Killzone trailer graphics anytime soon, or ever, are simply kidding themselves, or don't understand. it WILL take at least a 5x to 10x in performance to do it.

 

"What is so amazing about the KZ trailer that it can't be matched this gen?"

Nothing, save for the perfect IQ. People seem to think that the Killzone CG was so vastly revolutionary when in reality looking at GOW for the 360, the naughty dog PS3 game, and FF13 it seems to me that we're not too far off at all from the KZ trailer. Why do people think that there is little room for advancement with these consoles? Just wait until you see some of the stuff coming out in 2 years or so..

 

It's been pointed out a few times, but I think it's worth emphasizing:

There is a vast difference between the enthusiast and the professional's eye.

For example, I can't tell the difference in quality between the KZ smoke and Lost Planet smoke. According to the professionals here, there is a chasm between the two. And again, I tell the KZ CG has better lighting than most, but I can't readily pick out what technique was used.

Rmember, I'm not talking about casuals here. I know there's a country mile between Motorstorm and the short film of 2005.

 

im no professional about the smoke but i can clearly tell the difference between Killzone's smoke and lost planet's smoke, the former is way more complex compare to the latter.

 

Yes. And fans fit into two categories: One who, when introduced to the new techniques, can then perceive the differences and then those who cannot. It is like the gamer who says, "I cannot imagine graphics getting better than this game" only for 3 years later to say, "Wow, those graphics are so flat, static and uninteractive and have not aged well at all". The water in HL2 is a great example. A friend of mine still raves about it and thinks it is the same as real water because he cannot see a difference. Nevermind the large pixelation, he just cannot see a difference--even when pointed out. Other fans are able to see new, better techniques and say, "You know, I couldn't tell in the past, but I can now see why the old way wasn't very good".

When you reverse the paradigm, e.g. a CGI and then a game, people tend to look at the simularities. Character models, textures, etc all look about the same, and the general tone of lighting... it is all good. Right?

And yet the critical eye who understands the techniques and is looking for the known flaws and areas of disparity between offline and realtime rendering, well, the 'sore thumbs' stick out.

I think we are reaching a point where graphical jumps will become less and less important for users (but not the content creation pipeline) because a larger and larger number of consumers just cannot see the differences. They are huge, but the things that are different are not things the uninformed eye really notice. And most importantly most consumers are buying the games, not the graphics. We graphic hounds are the minority I would say, at least in the sense of the minutiea of graphical techniques and quality.

 

Actually, characters in killzone had 1 000000 + poly's while GoW characters has 10k.

 

I agree that GOW most def isn't more impressive than the Killzone vid, but I want to ad that level of detail consists of much more than poly count.

 

both of those numbers are very wrong.

 

The difference may be all in your mind.

 

not ragging on one particular console or the other, but I will say this, both Xbox 360 and PS3 got shortchanged in terms of GPU performance overall. even with Xeno's 256 GB/sec 10MB eDRAM.

they both have 4 Gpixel/s fillrate. without anti-aliasing. with AA, Xbox 360 fillrate remains the same, while RSX drops. but still 4 Gpixel/sec fillrate is not enough for this gen imo. I also disgaree that shaders are more important. they are important but can't make up for the lack of pixels or polygons.

Sony mentioned in one of its patents that realtime graphics would make use of several tens of Gpixels/s of fillrate, so that meant at least 20 Gpixels/s - that's where I had believed the current generation would be. that would've allowed HDTV resolutions with highly complex graphics at high framerates.

I believe that still applies to pixel-shader centric GPUs not just Graphics-Synthesizer renderers.

so we are now behind the curve in console performance IMO. they focused on increasing CPU power while making the GPU less important, even Microsoft, even though Xenos is pretty damn impressive.

that's why the leap in graphics from this current generation is so much smaller.

the PS3 and Xbox 360 are capable of doing some of the things that people expected PS2 and Xbox1 to do.

so I can only assume that Xbox 720 and PS4 will do the things we expected Xbox 360 and PS3 to do.

this might be an overly simple way of looking at things. well, I'm a simple minded guy, lol.

all the slick CG seen at E3 2005 should be the starting point of first-gen PS4 gameplay graphics.

graphical complexity, very high image quality/perfection, consistant 60fps for every game, 1080p resolution.

 

Why not?

It wouldn't help in shading quality, lighting complexity, or vertex effects at all though. Fillrate is only a factor. It was big with PS2 that used masses of overdraw to render effects. It's far less important with single-pass shaders. eg. Masses of fillrate is needed when you render four textures in four passes, but not when you render four textures in one pass.

I think people's disappointment comes mostly from unrealistic expectations and impatience. It took PS2 devs quite a while to master that fillrate. It'll take devs quite a while to master these SM3 GPUs in a closed box environment instead of their heavy API driven PC experience. And for expectations, what exactly are people after? KZ type graphics from this gen wouldn't have come with 100 GPixel fillrates! All that AA and smooth models some are after just requires far more prcoessing power than we can fit onto a cost effective chip. Neither MS nor Sony held back on the GPU. It's the laws of physics that are preventing better!

 

4GPixels/s means you can render each pixel 32 times every frame for 60 1920x1080pixels frames per second.

Ought to be enough, IMO.

Cheers

 

KZ is 500K-1M polygons in the closeups, subdivision levels are adjustable in Lightwave. This info is straight from the source.

GOW hero characters seem to be around 10-12K triangles in their highest LODs. This one is based on my own modeling experience (can't tell you witch title, sorry).

 

Can you stop putting words that i didn't write when you quote my comment and don't take my comments out of context since I just replying to a post that i believe is invalid.

 

In other words. You can see the difference because THERE IS A DIFFERENCE. If there wasn't a difference, GG would have wasted a lot of money on that movie.

 

When people saw a launch PS2 game like Tekken Tag Tournament I doubt anyone imagined a game like God of War would ever be made for the same system 4-5 years down the line. But when you play a game like Resistance, there is already a glimmer of hope that what we saw in the KZ2 trailer is achievable.

And there is a considerable difference between being impossible with straightforward techniques and being impossible period. The "oh 100k poly per character mechs are technically impossible so visually this is impossible" and the "oh PCs have had more than 9GB/s of fill rate for a long time and PC games don't look this good ergo this is impossible" conclusions have to be taken with a huge grain of salt.

Anyway people who flat out say something can't be done are boring.

 

The difference, of course, is that most of the PS2 tech demos were running on PS2 hardware whereas Killzone was rendered offline with CGI level assets.

What people with your position have yet to do, though, is explain how the quality of the KZ CGI will be attained. A number of us pointed out many, many details in the video (e.g. hair) that are extremely computationally expensive to get the same fluid, dynamic look that the KZ trailer demonstrated.

No one is saying they cannot scale down the 1M poly models to something more reasonable and add detail "back in" with normal maps and virtual displacement maps and that they cannot use precomputed radiosity tricks and fake the hair, etc.

But all of these are noticable to the trained eye -- and will stick out to people in 2012 when the PS4 is kicking out games that use 4GB of memory and over 2TFLOPs of computational power. All these hacks will be exactly what they are and you will be able to see the distinct difference between the CGI and a hack (albiet very good one).

The bottom line is that the general feel and theme and artistic approach is surely capturable. Is the CGI quality achievable?

If so I am waiting to here a series of explainations about how

a) they can reach the same quality level
b) within the constraints of realtime rendering

Texture, edge, and shader aliasing alone are issues enough to cast doubt on any suggestions. No need to mention poly budget, particle budget, AI, animation, draw distance, gameplay physics, vehicles and vehicle physics, lighting and shadowing model, LOD system, and so forth.

Little Johnny: Can I travel faster than the speed of light in my VW Beattle and through the sun and live to tell about it?
Albert, the Physicist: No Johnny, you cannot.
inefficient: Albert, you are boring.

. . .

 

I don't recall Albert Einstein ever saying "man would never be able to fully explore our galaxy because the speed of light makes travel at those distances impossible." But if he did he would be boring!

It's one thing to say X limitation exists and another to say X will prevent Y.

 

Not really. Take the Beattle example. The fact it cannot fly or operate in space isn't just a limitation, it prevents little Johnny from exceeding the speed of light and traveling through the sun. The limitations of the Beattle prevent it from achieving those things.

Likewise the limitations of the NDS prevent the NDS from create a game of same quality as MGS3 on the PS2.

And by memory there are scenes with a dozen or so people on screen at a time in KZ CGI. The base models are all weighing in at 1M polygons. The sheer size of such assets is insane, and at peak RSX doesn't have the processing power to even generate that many triangles (especially all those small ones, which would choke the GPU), let alone run all the advanced shaders and lighting & shadowing techniques KZ CGI had going on.

If offline rendering could be done in realtime -- or near realtime -- with 512MB of memory and standard PC GPUs they would do it. VFX houses heavily invest in technology and techniques to shave percentages off of render time.

So the question to you, and others who think KZ CGI is achievable on the PS3, is how do you go about getting around the memory size and bandwidth constraints? Texture cache sizes and poly budget for rendering and storage? And the entire host of issues.

Just because PS2 Tech Demos were exceeded by real PS2 games doesn't mean Offline CGI Demos will be exceeded by real PS3 games. The logic doesn't follow.

And no one has made an arguement about how a developer is supposed to go about resolving all the quality and quantity issues seen in CGI and the limitations found in the current consoles.

 

My eyes are not "trained", i just play games but i can see a difference between the fluidity and dynamics of the 500k-1M models cut and 10-12k models of GOW. That is with camera taken footage against gameplay on my HDTV. A full HD cut will help even more i can bet on it.

Must be me being the pragmatic person not wanting to kid myself out of personal affliations or something.

Is there a VW Beetle now that can beat the speed of light? I love to have one.
People beating a dead horse with wishful thinking is not boring but irritating btw.