HDMI 2.0 officially announced: 18Gbps bandwidth, 60fps 4K

[CountZero]

http://www.engadget.com/2013/09/04/hdmi ... nel-audio/

With this and Thunderbolt 2.0 coming out, Hi Rez Vr at 60-120 hz can become mainstreem

[Dungeonman VR]

http://www.engadget.com/2013/09/04/hdmi ... nel-audio/

With this and Thunderbolt 2.0 coming out, Hi Rez Vr at 60-120 hz can become mainstreem

Sure...now everyone...if it's not to much of a bother...please go out and buy $5000 PC's that can run 4k resolution at 60fps...

[Kamus]

http://www.engadget.com/2013/09/04/hdmi ... nel-audio/

With this and Thunderbolt 2.0 coming out, Hi Rez Vr at 60-120 hz can become mainstreem

Sure...now everyone...if it's not to much of a bother...please go out and buy $5000 PC's that can run 4k resolution at 60fps...

I've been running 1080p before the PS3 came out... And you are telling me that i have to spend 5,000 dollars for 4k (just 4 times the resolution) 7 years later?

Please...

4K isn't as big as a deal as people seem to think. They go read one website that ran Crysis 3 at Max settings, highest possible AA and AF and then say you need 3 titans to be able to play 4k...
Nonsense.
There's a point where going up in settings on a game that was developed for MUCH lesser hardware enters diminishing returns:
You get to lose a LOT of performance, for very little visual quality gains.
4K is doable today on even a midrange PC with sane quality settings.

[colocolo]

naahh, lets say two 780s.... 1300$ for a rig. but i wouldnt spent that much money too early. it isnt worth it.
even crysis 3 is already looking like crap to me. i will wait as long as i can. one highend maxwell at late 2014 will be the deal. if Brigade catches up a second one will be necessary but so so worth it.

[Kamus]

naahh, lets say two 780s.... 1300$ for a rig. but i wouldnt spent that much money too early. it isnt worth it.
even crysis 3 is already looking like crap to me. i will wait as long as i can. one highend maxwell at late 2014 will be the deal. if Brigade catches up a second one will be necessary but so so worth it.

Well, if by brigade you mean just the tech demos. Then sure, that might just be enough.
But running brigade like graphics on a game like Crysis 3? Good luck with that. I think we're still a few generations away from that. My guess is that in 7 performance doublings we'll start to see it in just about every high quality graphics engine out there.
But we may see it long before that in some limited capacity in other games.

But getting back on topic. I hope the rift has as many different inputs as possible:

Such as the newest display port, this new version of HDMI, and maybe even Wigig.

[colocolo]

naahh, lets say two 780s.... 1300$ for a rig. but i wouldnt spent that much money too early. it isnt worth it.
even crysis 3 is already looking like crap to me. i will wait as long as i can. one highend maxwell at late 2014 will be the deal. if Brigade catches up a second one will be necessary but so so worth it.

Well, if by brigade you mean just the tech demos. Then sure, that might just be enough.
But running brigade like graphics on a game like Crysis 3? Good luck with that. I think we're still a few generations away from that. My guess is that in 7 performance doublings we'll start to see it in just about every high quality graphics engine out there.
But we may see it long before that in some limited capacity in other games.

But getting back on topic. I hope the rift has as many different inputs as possible:

Such as the newest display port, this new version of HDMI, and maybe even Wigig.

No i mean Brigade 3 which will be well superior than the tech demos.
If i believe Sam Lapere's words for a second. (almost noisefree compared to Brigade 2 and overall quality like Octane Render) Crysis will look like Tetris compared to that.

[ss248]

But running brigade like graphics on a game like Crysis 3? Good luck with that. I think we're still a few generations away from that.

lol.
You don't understand how this engine works?

[Alejux]

No i mean Brigade 3 which will be well superior than the tech demos.
If i believe Sam Lapere's words for a second. (almost noisefree compared to Brigade 2 and overall quality like Octane Render) Crysis will look like Tetris compared to that.

I seriously doubt Brigade 3 will get rid of all the noise I've seen in the B2 tech demos. Sam seems too optimistic about that technology being ready for today. I would be glad to be proven wrong, but if I had to bet money, I would say we're still 3 GPU generations until Path Tracing can compete with other engines for actual high quality gaming.

[LukaszMMt1]

@up

If the noise was at level it was in the last video they showed I would be ok with it.
Just look for screenshots with noise (so you know they aren't static), yt compression really kills the effect.

[colocolo]

i have been following the blog for a very long time. And i am sure that the demos they are showing to us isnt the top. Sam Lapere always also told that.
The thing about Brigade 3 is that the lighting system and material system is now almost the same as in Octane. Apparently this would take more resources, but Sam Lapere said they have found a really clever way to greatly reduce the noise in postprocessing without effecting the image quality in any term. Maxwell might be a bit too early perhaps if it was only a midrange and not THE top notch card.
But Volta (2016)with a bandwith of 1 Terabytes....and thats probably midrange IMO since they will use Hybrid memory cubes and those chips are already around 320GB/s-480GB/s per 4GB stack.
We will see already 12GB graphics card at the end of this year according to Jules Urbach at GTC OTOY talk.
by the way i dont see how actual high end game engines could compete in any way with a path tracer.
humans have a very good sense for detecting false images. and even path traced images dont look like real photographs say 'photorealistic'. this term by the way is so misused. Its a long way till photorealism and especially a challenge for artists to make a world look like reality. no way around photogrammetry here.

[Dungeonman VR]

http://www.engadget.com/2013/09/04/hdmi ... nel-audio/

With this and Thunderbolt 2.0 coming out, Hi Rez Vr at 60-120 hz can become mainstreem

Sure...now everyone...if it's not to much of a bother...please go out and buy $5000 PC's that can run 4k resolution at 60fps...

I've been running 1080p before the PS3 came out... And you are telling me that i have to spend 5,000 dollars for 4k (just 4 times the resolution) 7 years later?

Please...

4K isn't as big as a deal as people seem to think. They go read one website that ran Crysis 3 at Max settings, highest possible AA and AF and then say you need 3 titans to be able to play 4k...
Nonsense.
There's a point where going up in settings on a game that was developed for MUCH lesser hardware enters diminishing returns:
You get to lose a LOT of performance, for very little visual quality gains.
4K is doable today on even a midrange PC with sane quality settings.

Actually...it is a big deal when you add the Oculus into the equation. you're forgetting that everything has to be rendered twice.

[xhonzi]

Did I miss it, or was there no mention of 48fps or really anything new for Frame Packed 3d?

[ss248]

I seriously doubt Brigade 3 will get rid of all the noise I've seen in the B2 tech demos. Sam seems too optimistic about that technology being ready for today. I would be glad to be proven wrong, but if I had to bet money, I would say we're still 3 GPU generations until Path Tracing can compete with other engines for actual high quality gaming.

That noise is not so bad.
It's like noise you see in the dark room IRL.
You can get used to it pretty fast.

[Kamus]

But running brigade like graphics on a game like Crysis 3? Good luck with that. I think we're still a few generations away from that.

lol.
You don't understand how this engine works?

Yes i do... I think you are the one that probably didn't understand my analogy.

What i meant was:
Running a tech demo at good frame-rates is one thing; running a full fledged game with physics, animations, AI, multiplayer, etc, etc. is on a whole different ballpark of computing power.

Basically, what i'm trying to say is: If you think that because you can run one of those tech demos at 1080p@60 FPS you'll be able to play a full fledged game like Crysis but with the Brigade like lightning on top of it, instead of the fake shadows and lights Crysis 3 uses, you are in for a rude awakening.
There is no way in hell that's going to happen by next year.

No i mean Brigade 3 which will be well superior than the tech demos.
If i believe Sam Lapere's words for a second. (almost noisefree compared to Brigade 2 and overall quality like Octane Render) Crysis will look like Tetris compared to that.

I understand what you mean, but you seem to be grossly underestimating the amount of computing power we would need to run a game that has lights that "look like tetris" (crysis 3) with brigade like lightning. We are nowhere near being able to do that.

It's one thing to run a tech demo that looks very good at good resolution and high framerate. But running something that looks like the tech demo in an actual game requires much more CPU power.

Of course the beauty about path tracing is how parallel it is. All it takes to get more performance is more transistors on a GPU basically. So it's very easy to guesstimate when we'll have the compute power to have it in games. My guess is around 7 performance doublings from where we are currently at.

[Kamus]

Actually...it is a big deal when you add the Oculus into the equation. you're forgetting that everything has to be rendered twice.

I've been running stuff "rendered twice" on my 3 year old computer using 3d vision (GTX 580) just fine at 1080p.
As long as developers take good care to optimize for framerate; you'll be very surprised on how good looking results we could get on even a midrange PC

[cybereality]

So can we finally get 3D at more than 720P? Can't believe they don't even mention 3D.

[ss248]

Yes i do... I think you are the one that probably didn't understand my analogy.

What i meant was:
Running a tech demo at good frame-rates is one thing; running a full fledged game with physics, animations, AI, multiplayer, etc, etc. is on a whole different ballpark of computing power.

Basically, what i'm trying to say is: If you think that because you can run one of those tech demos at 1080p@60 FPS you'll be able to play a full fledged game like Crysis but with the Brigade like lightning on top of it, instead of the fake shadows and lights Crysis 3 uses, you are in for a rude awakening.
There is no way in hell that's going to happen by next year.

I think, I get you right.
The beauty of this engine - it's doesn't really matter how much polygons on screen you have.
If it's already working, you can add as much polygons as you want, it's not so performance dependent.

https://www.youtube.com/watch?v=huvbQuQnlq8

Each jolly dancer contains over 60k triangles, totalling over 300 million dynamic triangles in real-time.

Physics

This engine can even do realistic water physics real time. It's not a problem.

animations, multiplayer

Doesn't really matter, because it's not performance dependent.

AI

It's another question. I don't think developers will start making good AI before they cap graphics.
So it's a big problem, but it's problem of whole industry.

Anyway, I expect even better graphics than brigade 2 in games in 3-4 years on high-end PC.

P.S. Sorry for my english, I'm reading most of the time. But I hope you get me.

[Kamus]

I think, I get you right.
The beauty of this engine - it's doesn't really matter how much polygons on screen you have.
If it's already working, you can add as much polygons as you want, it's not so performance dependent.

Oh, i'm familiar with the blog and with what the engine is capable off. There is little doubt that this is the way all games will render once we have the compute power. But i really believe we are about 7 performance doublings away from being able to completely ditch the old way of rendering in favor of path tracing. or around 2 orders of magnitude. (powers of 10)

If this was feasible today, that's the way we'd be playing at least some of our games. There's a reason you can't play games that look like that just yet. And that reason is performance.

[colocolo]

I think, I get you right.
The beauty of this engine - it's doesn't really matter how much polygons on screen you have.
If it's already working, you can add as much polygons as you want, it's not so performance dependent.

Oh, i'm familiar with the blog and with what the engine is capable off. There is little doubt that this is the way all games will render once we have the compute power. But i really believe we are about 7 performance doublings away from being able to completely ditch the old way of rendering in favor of path tracing. or around 2 orders of magnitude. (powers of 10)

If this was feasible today, that's the way we'd be playing at least some of our games. There's a reason you can't play games that look like that just yet. And that reason is performance.

i can understand your pessimism. It would be too good to be true.

quote from http://raytracey.blogspot.de/2011/01/ar ... l-gpu.html


Arnold render is a unidirectional path tracer, so it makes a perfect fit for acceleration by GPUs. "Maybe in a few years it will" could be a reference to Project Denver. When Project Denver materializes in future high-end GPUs from Nvidia, there will be a massive speed-up for production renderers like Arnold and other biased and unbiased renderers. The implications for rendering companies will be huge: all renderers will become greatly accelerated and there will no longer be a CPU rendering camp and a GPU rendering camp. Everyone will want to run their renderer on this super-Denver-chip. GPU renderers like Octane, V-Ray RT GPU and iray will have a headstart on this new platform. Real-time rendering (e.g. CryEngine 4) and offline rendering (e.g. Arnold) will converge much faster since they will be using the same hardware.

ive read somewhere that those Maxwell cards (Denver) will house 4-16 ARM cores.

[Kazioo]

I see discussion about Brigade, so I think it's worth posting some recent quotes from one of the authors of the engine:

This is still Brigade 2 with some features from Brigade 3 (if you will). Brigade 3 is currently under heavy development and its quality of lighting and materials will blow you away, it's of a completely different level compared to what we have now and is offers the same quality of Octane (minus Oxtane powerful material system).

[Bunny demo] ran on 1 GTX Titan

Brigade 3 will be a lot faster and almost noisefree compared to Brigade 2. We're implementing some techniques that will greatly reduce the noise without affecting the image quality. More importantly, the old material and lighting system has been thrown out completely and is replaced by a new one, which is essentially identical to that of Octane Render, so you won't be able to tell the difference between the two. So if you're not yet familiar with Octane's quality, believe me it will blow you away.

Right now, we're also working on tools to make it easier for game devs to start mucking around with Brigade.

Brigade scales better when you render at higher resolutions due to having more coherency between pixels, e.g. 4x the resolution requires about 3-3.5x the compute power. Another nice benefit of rendering at high resolutions is that noise is much less noticeable at higher res. Because of that (and recent speed increases), I'm often running Brigade at 1080p and rarely go below 720p.

To be honest, I couldn't care less about MS or Sony or any other next gen console manufacturer. A Brigade powered game will probably not run on PS4 or Xbox and cloud gaming will make consoles obsolete in a few years, so your best bet will be Brigade games in the cloud.

but.. but... VR

hehe thanks, yep Brigade wiil let you enter the matrix soon enough, especially combined with the rift

[Kamus]

To be honest, I couldn't care less about MS or Sony or any other next gen console manufacturer. A Brigade powered game will probably not run on PS4 or Xbox and cloud gaming will make consoles obsolete in a few years, so your best bet will be Brigade games in the cloud.

but.. but... VR

I've been thinking about ways we could "have our cake and eat it". (VR and cloud gaming)
I've thought of a few solutions that may not even be that feasable, but here they go:

I've thought that since they are already streaming video, why can't they just stream a really high resolution video that has a 360 degrees field of view?
That way head tracking would be as instant as it is right now. The latency would come with positional tracking though...

Another way i've thought this could be possible, is to render some content locally, and really complicated stuff, like the lightning on the cloud. That way the game should feel as fast as ever. (i think nvidia is doing something similar to this, and they claim latency isn't noticeable)

I don't know how well this would work, but figured it's food for thought...

[colocolo]

To be honest, I couldn't care less about MS or Sony or any other next gen console manufacturer. A Brigade powered game will probably not run on PS4 or Xbox and cloud gaming will make consoles obsolete in a few years, so your best bet will be Brigade games in the cloud.

but.. but... VR

I've been thinking about ways we could "have our cake and eat it". (VR and cloud gaming)
I've thought of a few solutions that may not even be that feasable, but here they go:

I've thought that since they are already streaming video, why can't they just stream a really high resolution video that has a 360 degrees field of view?
That way head tracking would be as instant as it is right now. The latency would come with positional tracking though...

Another way i've thought this could be possible, is to render some content locally, and really complicated stuff, like the lightning on the cloud. That way the game should feel as fast as ever. (i think nvidia is doing something similar to this, and they claim latency isn't noticeable)

I don't know how well this would work, but figured it's food for thought...

????

[Kamus]

????

He was pointing out that we can't have VR + cloud gaming at the same time (because of latency). Which is why he highlighted in yellow the part about brigade games in the cloud.
The latency cloud gaming would bring to the table would be a deal breaker for VR.

[ss248]

????

He was pointing out that we can't have VR + cloud gaming at the same time (because of latency). Which is why he highlighted in yellow the part about brigade games in the cloud.
The latency cloud gaming would bring to the table would be a deal breaker for VR.

GTX 980 in 2015 will be enough (I hope).
But I don't think big multiplat developers will use something like brigade, because of weak consoles.
The only hope is oculus exclusive games, but will 980 be enough for 120 fps ?
IDK, time will tell.

UPDATE:
I just rethink everything and based on trend I think we will have around 9.6 Teraflops (2xTITAN) in GTX 980.
GTX 580: 1.5 Teraflops
GTX 680: 2.5 Teraflops
GTX 780: 4.0 Teraflops
GTX 880: 6.25 Teraflops
GTX 980: 9.6 Teraflops
Bunny scene (I posted it above) was running at 720p 30fps on single Titan.
So based on wild speculation, in 2015 we will have 1080p 60fps, enough for bottom line VR.

[colocolo]

i am optimistic that it could happen with Volta. end 2015?2016
the brigade team at least has been asked about it by big game developers.
why isnt there any method to implement eyetracking into a HMD?
with fovea rendering it could mean a quantum leap for realtime graphics
it would give us all the power we need to run a path traced matrix at UltraHD res in a few years.
without it, it would be still a verylong way till we run Brigade at 4k or 8k.
i dont want to wait again this long.
Please Oculus VR, research this, i am begging you. i am desperate.
You are our only hope and last chance that this could happen as long as you are independent and free from business primates.

[ss248]

the brigade team at least has been asked about it by big game developers.

Really? Source?

why isnt there any method to implement eyetracking into a HMD?

There is actually. Just place two electrode on skin around eye and you can have it.
Pretty simple, but not consumer friendly.

without it, it would be still a verylong way till we run Brigade at 4k or 8k.

In ten years we will be able to. I'm sure.

[colocolo]

thats not enough.
also, a experience is not only about static graphics, so this resource should be used.
there are always very cool things you could compute. we will never have enough computational power.
basically what i want to exprience in 10 years is that scene from 2012 as the family drove with a limousine through that collapsing skyscraper and the whole city also collapsing around them. questions?

[ss248]

questions?

Yep.

we will never have enough computational power.

Don't think so. We already have enough power to do simple tasks, but we always will want more.
It's human nature.

from 2012 as the family drove with a limousine through that collapsing skyscraper and the whole city also collapsing around them.

We are able to script it right now. It's pretty simple.

And now a question:

the brigade team at least has been asked about it by big game developers.

Where did you get this information?

[colocolo]

http://raytracey.blogspot.de/2013/03/pr ... asted.html


Sam Lapere said...

Mark: yes, we've been approached by some of the largest game developers in the industry to use Brigade. Unfortunately I can't tell which ones.

[Kamus]

i am optimistic that it could happen with Volta. end 2015?2016
the brigade team at least has been asked about it by big game developers.
why isnt there any method to implement eyetracking into a HMD?
with fovea rendering it could mean a quantum leap for realtime graphics
it would give us all the power we need to run a path traced matrix at UltraHD res in a few years.
without it, it would be still a verylong way till we run Brigade at 4k or 8k.
i dont want to wait again this long.
Please Oculus VR, research this, i am begging you. i am desperate.
You are our only hope and last chance that this could happen as long as you are independent and free from business primates.

Eyetracking doesn't seem like something that hard to implement.
They could use infrared lights aimed at your eyes along with two small cameras to take care of the tracking. It's very easy to imagine the performance gains we would get out of it too. And since this is an HMD and nobody else can look at the screen, there would be no worries about eye tracking being a "one guy experience" since it already is. (and lets be honest, it would be amazing even on just a regular monitor since most of the time it's a 1 person experience, or better yet on a multi-monitor setup)

Humans are only able to focus a very small area of our FOV with high detail, and we are basically forced to move our eyes even to read on something as small as an iPhone, or we wouldn't be able to make out the details.

So yeah, I'm thinking this is something that should be on the top list of priorities for Oculus, since it doesn't seem that hard to implement, and the performance gains would be gigantic if we only rendered at high resolution and high polygon counts where our eyes are focusing on. (and maybe even half framerate if they figured out a way to make frame interpolation work with out any noticeable side effects)

I don't think this is even on the table at the moment though, or we would have read at least something about it. But let's hope they can implement a feature like this for the second consumer version. And it wouldn't hurt to ask them, maybe they are at least researching it.

[TheHolyChicken]

So yeah, I'm thinking this is something that should be on the top list of priorities for Oculus, since it doesn't seem that hard to implement, and the performance gains would be gigantic if we only rendered at high resolution and high polygon counts where our eyes are focusing on. (and maybe even half framerate if they figure dout a way to make frame interpolation work with out any noticeable side effects)

Getting eye tracking to the stage where it would be able to be used for performance gains (by rendering with high fidelity only where the user is looking) would require mastering PREDICTIVE eye tracking; that is, predicting where the user is going to look with his eyes before he actually looks. This "lead" time is required so that a game engine has time to prepare the frame that will be displayed in front of the eye, immediately BEFORE he looks. This is substantially harder than headtracking, and definitely does not fall into the category of "not hard to implement".

More basic eye tracking might be nice for things like better interaction with NPCs, eye contact with other players, aiming, etc, but nothing particularly groundbreaking (unless I'm overlooking something). For most of these purposes I'm sure that a simple estimate of where the user is looking would be more than sufficient.

[Alejux]

So yeah, I'm thinking this is something that should be on the top list of priorities for Oculus, since it doesn't seem that hard to implement, and the performance gains would be gigantic if we only rendered at high resolution and high polygon counts where our eyes are focusing on. (and maybe even half framerate if they figure dout a way to make frame interpolation work with out any noticeable side effects)

Getting eye tracking to the stage where it would be able to be used for performance gains (by rendering with high fidelity only where the user is looking) would require mastering PREDICTIVE eye tracking; that is, predicting where the user is going to look with his eyes before he actually looks. This "lead" time is required so that a game engine has time to prepare the frame that will be displayed in front of the eye, immediately BEFORE he looks. This is substantially harder than headtracking, and definitely does not fall into the category of "not hard to implement".

More basic eye tracking might be nice for things like better interaction with NPCs, eye contact with other players, aiming, etc, but nothing particularly groundbreaking (unless I'm overlooking something). For most of these purposes I'm sure that a simple estimate of where the user is looking would be more than sufficient.

Do you need predictive tracking algorithms when you can have latency of 4ms? Just wondering. There has already been much research done regards of doing eye-tracking using hot-mirrors. It's not exactly a new thing.

http://vision.arc.nasa.gov/personnel/le ... t_text.pdf
http://www.smivision.com/en/gaze-and-ey ... x-hed.html
http://gazeinteraction.blogspot.com.br/ ... acker.html
http://eyeseecam.com/

[colocolo]

So yeah, I'm thinking this is something that should be on the top list of priorities for Oculus, since it doesn't seem that hard to implement, and the performance gains would be gigantic if we only rendered at high resolution and high polygon counts where our eyes are focusing on. (and maybe even half framerate if they figure dout a way to make frame interpolation work with out any noticeable side effects)

Getting eye tracking to the stage where it would be able to be used for performance gains (by rendering with high fidelity only where the user is looking) would require mastering PREDICTIVE eye tracking; that is, predicting where the user is going to look with his eyes before he actually looks. This "lead" time is required so that a game engine has time to prepare the frame that will be displayed in front of the eye, immediately BEFORE he looks. This is substantially harder than headtracking, and definitely does not fall into the category of "not hard to implement".

More basic eye tracking might be nice for things like better interaction with NPCs, eye contact with other players, aiming, etc, but nothing particularly groundbreaking (unless I'm overlooking something). For most of these purposes I'm sure that a simple estimate of where the user is looking would be more than sufficient.

did you know that the Myo can do predictive muscle tracking?
it can detect the movement of the fingers even before they move. well it also should be this way.
thus a combination of electromyo oculugraphy and camera tracking could be a good combination.
Much like sensor fusion is done in the Rift right now.

cool, finally eye tracking coming for a reasonable price to market(99$):
http://www.youtube.com/watch?v=PL9cCi5zTzE

[Kamus]

Do you need predictive tracking algorithms when you can have latency of 4ms? Just wondering. There has already been much research done regards of doing eye-tracking using hot-mirrors. It's not exactly a new thing.

http://vision.arc.nasa.gov/personnel/le ... t_text.pdf
http://www.smivision.com/en/gaze-and-ey ... x-hed.html
http://gazeinteraction.blogspot.com.br/ ... acker.html
http://eyeseecam.com/

I had seen the last video posted on a thread somewhere already. And that is what convinced me that latency wouldn't be that hard to overcome. If we can have physical objects move almost at the same time our eyes do, why would latency on a 3D rendered world be any worse?

Getting eye tracking to the stage where it would be able to be used for performance gains (by rendering with high fidelity only where the user is looking) would require mastering PREDICTIVE eye tracking; that is, predicting where the user is going to look with his eyes before he actually looks.

I think the latency is already at the point where we could do with out any significant help from predictive algorithms.
But i'm just speculating here. It would be nice to see some proof of concept.

[colocolo]

SMI has integrated two eye tracking cameras into the frame of glasses. only 60Hz but one step closer.
he also talks about virtual reality recordings.....but for caves....
http://www.youtube.com/watch?v=kiFpMbfj_08
http://www.youtube.com/watch?v=J1OQJzDx_S4

[colocolo]

wow, check out this Murci!
it looks like it has as much polys as a next gen game in one scene and it has dark chrome as material for the wheels too. haha
http://raytracey.blogspot.de/