How long before a X800 wrapper (Ruby demo) appears?

[Ostsol]

Tried it with the Geforce FX 5900, and this is the error I get, Absolutely no idea what it means.

//=====================================================
// ATI Sushi Error Log Created 5/6/2004 7:09 pm
//=====================================================
[AwFn.cpp] (line 3402): D3DAw Error: AwCreateRenderableTexture - Unable to create color texture object
[StartEnd.cpp] (line 2044): Error creating color buffer "cBlurPingBuffer"!
[Main.cpp] (line 880): Normal Application Exit

It manages to get to the load up screen, But when it actually finished loading the demo it crashes.

yea i get the same exact error with both dll's

Probably needs floating point render targets. AFAIK, NVidia still doesn't support them on the GeforceFX under Direct3D.

 

[LeGreg]

Wow, that is pretty sweet. What are the performance implications of this technique? Will it be usable in upcoming games?

It's getting pretty old already

the technique is called precomputed radiance transfer
encoded in spherical harmonics. (shprt)
The fact is that the technique is quite cheap in this form and actually it doesn't stress the hardware too much. Even so it can produce some impressive "real time" images (but it's more like an holograph with interchangeable lighting). I know at least a demo that runs on a geforce 3/4 class of graphics hardware (and if executed in a vertex shader virtually every class of hardware can do it).

There is a lot of issues associated with the technique as well: the simplest form cannot handle moving parts on objects (that's why it's a statue and not a moving character). But it's probably nice for static props and it can be of course combined with all sort of other lighting calculations.

The latest dx9 sdk comes with a sample and a framework to calculate such preradiance transfer (including scattering) and bake them into vertices or into a texture.

There are more elaborate forms that deals with different issues. What comes most of the time is the level of time spent in precomputation for complex scenes. Reducing this time is key actually.

LeGreg

 

[Dave Baumann]

Looking at the shaders, ATI probably could have gotten this to work in the R9700 and above since it looks like all the shaders could have been split up into multiple passes. They are applying 3 lights in a single pass, which just pushes the instruction counts to near 100 (or higher) for most of the problem shaders. The other problem ones are things like doing blurs and glows, which also 'could' be multipassed fairly easily too.

The demo took a year to develop all told, and the shaders were multipassed for development on 9800. It wasn't until R420 hardware came back that they made the shaders single pass for performance optimisation on R420.

 

[Bambers]

There is a lot of issues associated with the technique as well: the simplest form cannot handle moving parts on objects (that's why it's a statue and not a moving character). But it's probably nice for static props and it can be of course combined with all sort of other lighting calculations.

atis blurb says that rubys skin is supposed to use subsufrace scattering. how do they cope with the movement on that?

 

[tb]

There is a lot of issues associated with the technique as well: the simplest form cannot handle moving parts on objects (that's why it's a statue and not a moving character). But it's probably nice for static props and it can be of course combined with all sort of other lighting calculations.

atis blurb says that rubys skin is supposed to use subsufrace scattering. how do they cope with the movement on that?

LeGreg meant shprt, not subsurface scattering. shprt can also be used on "moving parts", I think there was a gdc 2004 paper about that...

Thomas

 

[THe_KELRaTH]

............ 3 weeks later in a forum somewhere about here ..........


Holy Smokes Batman Ruby's now running faster on R3x0 than a R420.......





Nice going Colourless

 

[g__day]

Colourless - huge thanks - once again - you're the man!!!

Why two files with the d3d9.dll - I tried both in my ATi demo directory and they looked teh same to me on a 9800 Pro.

Cheers, I didn't imagine you'd do this so fast!!!

 

[StarGazer]

Wow... 8500+ views in 2 days.
I hope ATi didn't castrate Colourless because we'd like to see how the wrapper is doin'

 

[Ante P]

NV40 + 61.11 + Ruby + someruby.zip =

[AwFn.cpp] (line 3402): D3DAw Error: AwCreateRenderableTexture - Unable to create color texture object
[StartEnd.cpp] (line 2044): Error creating color buffer "cBlurPingBuffer"!
[Main.cpp] (line 880): Normal Application Exit

 

[Colourless]

I said hopefully today.... and here we are!

Download it here
and read the ReadMe here

I STRONGLY RECOMMEND YOU READ THE README!

And here's some screenshots

The rendering isn't entirely correct, not be a long shot. But IMO it's pretty good for 2 and a bit days work. Next thing I'm going to attempt to multipass some of the shaders so the effects actually work properly

 

[StarGazer]

Thanks, and thanks again you Ultima VII freak!

 

[Monrad]

This version doesn't work for me

This is what I get:

//=====================================================
// ATI Sushi Error Log Created 5/7/2004 2:19 pm
//=====================================================
[Main.cpp] (line 894): Exception caught - BAD NEWS!


I had no problems running the other versions.

Thanks 8)

 

[Colourless]

Try grabbing the DLL from someruby.zip and see if that works.

 

[Monrad]

Yes Now it works like charm again

Congratulations

 

[LeGreg]

atis blurb says that rubys skin is supposed to use subsufrace scattering. how do they cope with the movement on that?

see their gdc paper about skin rendering...

LeGreg

 

[cheezemurda]

//=====================================================
// ATI Sushi Error Log Created 5/7/2004 12:36 pm
damn...

//=====================================================
[SSGenericAPI_D3D.cpp] (line 1312): No D3D Error message. You might be specifying more than one component on a source register.[qGlow.ssh] (line 1015): Error creating Pixel Shader
[SSGenericAPI_D3D.cpp] (line 1312): No D3D Error message. You might be specifying more than one component on a source register.[qTextureLit.ssh] (line 224): Error creating Pixel Shader
[SSGenericAPI_D3D.cpp] (line 1312): No D3D Error message. You might be specifying more than one component on a source register.[qTextureLit.ssh] (line 279): Error creating Pixel Shader
[AwFn.cpp] (line 3402): D3DAw Error: AwCreateRenderableTexture - Unable to create color texture object
[StartEnd.cpp] (line 2044): Error creating color buffer "cBlurPingBuffer"!
[Main.cpp] (line 880): Normal Application Exit

 

[kkman2002]

http://bbs.gzeasy.com/uploads/post-1-1083954216.jpg
more：
http://bbs.gzeasy.com/index.php?showtopic=156253

 

[Colourless]

If people are having problems, tell me your system specs.

Also make sure your Antialiasing Settings in your Control Panel is set to Application.

 

[cheezemurda]

i think the only problem here is that i have an nvidia card

5900 non ultra

 

[JigenD]

I got a different error with my FX 5900 nu... I got the exception - bad news with both dlls... the new one and 'someruby', haven't tried the 2nd old one yet but I will. I'll tell you if it works. I'm using 60.80 beta drivers.