Next Gen Renderer Xbox720/PS4

[Zelex]

Hi All,

 

Some stuff I presented at siggraph this year. vvv

 

 

Let me know whatcha think! ( btw, yes I know the music cuts out, but it was too good to not use )

 

Best,

Jon

[AlbertA]

You did that? Daaayuuummmm..

 

Is there a technical paper I can look at?

[AlbertA]

Was this the talk?

http://s08.idav.ucdavis.edu/

Edited December 26, 2008 by AlbertA

[Ryan FB]

We got some 280's in for doing on-GPU micro-CT reconstruction with CUDA. Smokes the hell out of our 48-core minicluster.

[foogledricks]

My four-year old did something like that on her etcher sketch.

 

Actually, I don't understand the context. You developed a rendering engine. Created a detailed model that it could render. And you're showing off how well you can zoom in, pan, and dolly around the model, with excellent performance, using your engine? I know very little about much of an achievement this example is because I'd have to see an A/B comparison. Like if you showed how choppy the Unreal 3 engine would render that same model.

 

Perhaps this thread is not for the likes of me anyway. But humor me if you can. Thanks.

[kelley]

There is no way in hell UE3 could do that. What your seeing is the future, not the current. I've seen the demo in person and its super impressive.

[Zelex]

Was this the talk?

http://s08.idav.ucdavis.edu/

 

yup

[Zelex]

My four-year old did something like that on her etcher sketch.

 

Actually, I don't understand the context. You developed a rendering engine. Created a detailed model that it could render. And you're showing off how well you can zoom in, pan, and dolly around the model, with excellent performance, using your engine? I know very little about much of an achievement this example is because I'd have to see an A/B comparison. Like if you showed how choppy the Unreal 3 engine would render that same model.

 

Perhaps this thread is not for the likes of me anyway. But humor me if you can. Thanks.

 

Without SVO tech, IIRC it can run at something like 10 fps raw rendering the triangles. With SVO, 60 fps. SVO is also technically rendering many times the detail than the triangles actually represent. Meaning that really if you rendered all the detail in triangles instead of the ray casting approach, your probably talking on the order of 10s or 100s of millions of triangles. The GPU would maybe get 1 fps with traditional rendering. But I digress, if you strictly rendered only the exact same thing in triangles and SVO, in that example data set, the difference would be from 10 fps to 60 fps. The real power of the technology is the unique "nearly infinite" geometry and texturing. Its nearly infinite because your bound by disk size or streaming speeds. However if you do procedural content generation it really is infinite. The net result of the tech is a generational skip in geometric complexity. Meaning, it can display the kind of detail you would see on a PS5 or a XBOX1080, but on a PS4 or XBOX 720.

Edited December 26, 2008 by Zelex

[Union Carbine]

(detailed explanation snipped)

Wow. Thanks for the explanation. Any downsides to that technology?

[Christopher]

Considering the success of the Wii, what if MS goes with a 360.5 and Sony does a PS3.5? Will it still work, just scaled down?

[Zelex]

The 360 should have enough processing power to do it at about 30 fps. Theory doesn't always meet practice there though. Also, the runtime memory requirements might be prohibitive there on an existing consoles. I don't think the PS3 has the pixel processing power required for it. So, yeah, absolutely. Next gen consoles should have about 4gb of ram, which is more than enough. To meet the lower processing requirements you would just have to scale down on the extras like real motion blur, depth of field, etc..

[Camp]

Considering the success of the Wii, what if MS goes with a 360.5 and Sony does a PS3.5?

 

What does that even mean?

Why would either company put forth the development money to deliver a ".5" update? Talk about opening a door for your competition to leapfrog you.

[Starhawk]

What does that even mean?

Why would either company put forth the development money to deliver a ".5" update? Talk about opening a door for your competition to leapfrog you.

 

I think he means the success of the Wii showed that the most technically underpowered console can kick everyone else's ass, and could Microsoft and Sony implement a form of this technology if they decided that the next Xbox/PS would be a slight hardware update with, say, a strange new controller.

[foogledricks]

I would not be surprised if the A/V leap is less aggressive in the next generation. I think what the Wiimote is to Nintendo is what software and services is to Xbox and Playstation. I think Sony and Microsoft perhaps should be less aggressive with A/V advancement in favor of these things. Cater their hardware toward enabling amazing dashboard/XMB features and services. Scale back on the super duper graphics advancement.

 

Just give me 1080p for all games at good framerates with a marginal increase in graphic fidelity. I really don't see me thinking that Gears of War 2 looks like shit in three years.

 

That being said, I do think that Wii games look like shit. But that's only because they mostly do. I don't think God of War looks like shit on the PS2.

 

 

For the first time in videogames I'm thinking that software and services are more important than A/V advancement. I'd be more excited about crazy new dashboard features that integrate with games better, integration with a game playing Zune where XBLA games can be played on, really aggressive digital distribution, etc...

[Brandon H]

If you're betting on the long haul, a half-step is not the way to go. Especially when you consider the economics of the Wii -- the fact that are probably tens of thousands of Wii owners out there who have never bought another title and are perfectly happy with WiiSports. And maybe WiiFit.

 

As innovative as it is, I don't know that there isn't some truth to the line of thought that the Wii is "the world's most expensive board game." I would be surprised if the legions of Wii owners out there are standing in line en masse for a "Wii 2" in a few years. They're satisfied with what they've got. So you've either got to catch lightning in a bottle again or push the industry forward on more powerful hardware and better storytelling and gameplay experiences. A half-step won't cover it.

[Zelex]

If you're betting on the long haul, a half-step is not the way to go. Especially when you consider the economics of the Wii -- the fact that are probably tens of thousands of Wii owners out there who have never bought another title and are perfectly happy with WiiSports. And maybe WiiFit.

 

As innovative as it is, I don't know that there isn't some truth to the line of thought that the Wii is "the world's most expensive board game." I would be surprised if the legions of Wii owners out there are standing in line en masse for a "Wii 2" in a few years. They're satisfied with what they've got. So you've either got to catch lightning in a bottle again or push the industry forward on more powerful hardware and better storytelling and gameplay experiences. A half-step won't cover it.

 

Not to be a nay-sayer, cause I just might eat these words, but I don't think Nintendo will be able to have the same effect for Wii 2 as it did for Wii 1. The reasoning being, is that I don't see any more power gloves to reinvent which would totally rock peoples worlds. Unless the Nintendo can really pull something amazing out of their butt, I just don't see them doing anything other than upping the graphics capabilities of the system. Perhaps doing the same as everybody else, just less cores. Lets take a survey of the various forms of input a human can give a computer:

1) Hand motion (aka power glove, wii wand)

2) Finger motion (aka controller,power glove)

3) Foot motion (aka ddr)

4) Toe motion (untapped potential! lol! )

5) Body motion (aka wii fit, camera, etc)

6) Head motion (untapped, awesome?

)

7) Facial motion (untapped via cameras)

8) Talking (mostly untapped singstar/rockband... Tea, Earl Grey, Hot! )

 

There is just not that many ways to go without upping processing power. Anything involving a camera will require lots of processing power. Anything involving toe motion is kind of ridiculous. I just don't see a direction they could go which would be dramatically different than what people are used to, and still be something that everybody wants to play with, without also using cutting edge technology. I think the next generation, they will either push the performance envelope again with all the other competitors as to enable these more computationally intensive innovations, or they will lag behind and ride on their previous successes, but a bit diminished over time. Content is surely a powerful driver of sales, but for how long without any technical innovation? I for one am very interested in seeing just how long Nintendo can keep this up A good next move for them would be head motion or finger motion. I want to see head motion without glasses, which does require much more computation than with glasses. Its a win-win for everybody if they can keep up the innovation. Go Miyamoto!

Edited January 5, 2009 by Zelex