Ruby: Whiteout

[chrisATI]

Hello, I just thought I'd post a quick note to let people know that the Whiteout demo has been released: http://ati.amd.com/developer/demos/rhd2000.html

You'll need Windows Vista with a D3D10 capable GPU with 512MB video memory. You will also want to make sure that you have the latest graphics drivers installed.

Enjoy!

--Chris

 

[Rufus]

You'll need Windows Vista with a D3D10 capable GPU with 512MB video memory.

The site specifically say Radeon HD. Can someone with a G80 try it?

 

[BRiT]

I gave it a quick try but the screen looked like I was in a whiteout storm. Nothing showed up, but I could hear the audio.

 

[Tim Murray]

I gave it a quick try but the screen looked like I was in a whiteout storm. Nothing showed up, but I could hear the audio.

Demirug has commented in another thread that this is because of invalid shader bindings.

 

[Jawed]

Since the tessellation video features "the same" mountain range as featured in Whiteout, I'm puzzled how it is that Whiteout seemingly doesn't use tessellation

The Frostbite presentation:

http://developer.amd.com/assets/And...eRenderingArchitecture(GDC07_AMD_Session).pdf

on pages 44-52 covers geometry and snow accumulation.

Fractional Brownian Motion (fBm, what's the matter with all-capitalised TLAs?) is on pages 42-44 of:

http://developer.amd.com/assets/Tatarchuk-Noise(GDC07-D3D_Day).pdf

The implication appears to be that fBm is operating solely as VS/GS, but it would be nice to get a clear statement.

It seems, from those PDFs, that Chris has worked on the mountains/snow. I think they're a huge improvement over the usual geometry/landscape seen in games, so I'm looking forward to much more...

Jawed

 

[Rys]

There's no tesselator use in the Whiteout demo, and if I were to speculate it'd be an issue of timing (in terms of when it was developed vs when support would have been exposed in a driver). No current driver has tesselator support enabled, and won't for some time (so I hear anyway).

 

[AlexV]

There's no tesselator use in the Whiteout demo, and if I were to speculate it'd be an issue of timing (in terms of when it was developed vs when support would have been exposed in a driver). No current driver has tesselator support enabled, and won't for some time (so I hear anyway).

I`ll probably catch flak for saying this, but if anything has been dissapointing for me with the 2900, the drivers would be that thing. They`re simply crude, IMHO.

 

[Geo]

So? Has anyone run it? What kind of FPS are you getting?


And thanks for the headsup, Chris.

 

[AlexV]

Not impressed with the performance...but then again, I`m single-carded, maybe it`s aimed at Crossfire users. Some things are nice, but I think they didn`t get it quite right with the material shaders for Ruby`s outfit and the pilot`s outfit and quite a few others, as they`ve achieved a very plasticy look for those. The lips at the end are rendered nicely though.

 

[CarstenS]

I've done some insane 2560x1600-Shots. (If you like, you can take a look at them here: http://www.pcgameshardware.de/?article_id=603480)

Kinda kills the Fps in the scenes with motion blur - and in turn also kills the motion blur effect. (No, didn't take shots of that).

 

[Jawed]

So, is this a D3D10 demo or is there something in there that prevents non-ATI cards running it?

Jawed

 

[Demirug]

So, is this a D3D10 demo or is there something in there that prevents non-ATI cards running it?

Jawed

As already stated in the other thread about this demo it looks like that nvidia have trouble with the invalid shader semantic bindings that the ATI driver seems simply ignore.

 

[Jawed]

As already stated in the other thread about this demo it looks like that nvidia have trouble with the invalid shader semantic bindings that the ATI driver seems simply ignore.

So it should be a D3D10 application, but "sloppy" coding prevents it from being so?

Jawed

 

[Demirug]

So it should be a D3D10 application, but "sloppy" coding prevents it from being so?

Jawed

So far I was able to check this, yes.

 

[Suflex]

I've done some insane 2560x1600-Shots. (If you like, you can take a look at them here: http://www.pcgameshardware.de/?article_id=603480)

Kinda kills the Fps in the scenes with motion blur - and in turn also kills the motion blur effect. (No, didn't take shots of that).

Horrible dithering on Rubby's hair and coat fur.

Here

 

[Jawed]

Is that a transparency anti-aliasing artefact, a "screendoor effect"? Could this be one of the invalid shader semantic bindings that Demirug refers to. I presume AA techniques such as this aren't part of D3D10.

Jawed

 

[AlexV]

Is that a transparency anti-aliasing artefact, a "screendoor effect"? Could this be one of the invalid shader semantic bindings that Demirug refers to. I presume AA techniques such as this aren't part of D3D10.

Jawed

Yup, pretty much a screendoor artifact. You get the same stuff with the COJ DX10 demo, some of the DX10 SDK demos and probably many others. Upping the multisample samples to 8 instead of 4(can be done in sushi.ini) takes care of it(mostly), but it`s a costly solution.

 

[cadaveca]

I get no screendoor @ 4x samples here, but ".ini" is edited for both cards. Interestingly enough, the demo runs just as well when crossfire is enabled in ".ini" as when it's disabled.

I get the screendoor @ 2x sampling though...really bad in her hair over her face.

Same problem as in CoJ. Although, since d3d updated a wee bit from install of demo, CoJ seems to look a bit better?

 

[AlexV]

Beauty is in the eye of the beholder

 

[Silent_Buddha]

Odd, I don't get that screen door effect at 4x either when I run whiteout. Just ran it again to doublecheck.

Regards,
SB