gl_LightSource[0].position

[jeickmann]

Hi,

I'm trying to implement some kind of simple lighting in a GLSL vertex shader.
Currently I'm using a directional light and I'm trying to get the L-Vector for the diffuse calculation like this:

vec3 lightDir = normalize(gl_LightSource[0].position.xyz)

The only problem is, this position doesn't change if I rotate the scene, etc.
(I tried this by simple setting the vertex-color to that value)
The GL-Spec says, that the position gets stored after having been transformed by the modelview-matrix, so it should change when I rotate the view.
If I use fixed-function lighting, it works.
Has anybody else experienced this issue or can tell me what I'm doing wrong?
I'm using an ATI Radeon 9800 with Catalyst 4.1

Thanks,

Jan Eickmann

[Corrail]

I had a similar problem. Seems like the ATI driver doesn't multiply the light position with the modelview matrix if you are using GLslang.

[This message has been edited by Corrail (edited 02-08-2004).]

[jeickmann]

Thanks for the info. At least I'm happy that I didn't do something stupid in my shader. Well for now, I'll just have to pass it in as a uniform.

Jan

Btw: is there any list for the current limitations in the ATI driver concerning GLSL?

[This message has been edited by jeickmann (edited 02-09-2004).]

[dginsburg]

Hi there,

I believe the issue with gl_LightSource[n].position has been resolved and will be fixed in a future release. I will double check to make sure.

There is a list of known GLSL caveats that ATI maintains. Please contact 'devrel@ati.com' to receive it.

Thanks.

-- Dan (ATI)

[jeickmann]

> There is a list of known GLSL caveats that ATI maintains. Please contact 'devrel@ati.com' to receive it.

Already did that last thursday, thank you. No answer yet, though. Thats why I was asking here.

Thanks for clearing it up though.

Jan

P.S: please don't misunderstand this as being unpatient.

[This message has been edited by jeickmann (edited 02-09-2004).]

[dginsburg]

Hi Jan,

Thanks for your patience, I spoke with devrel and you should hear back today.

Also, I was incorrect, this issue has not been fixed. We will try to reproduce it. If you have a reproducer app, please send it to 'devrel@ati.com' as this will expediate the issue being fixed.

Thanks.

-- Dan (ATI)

[Marty666]

Hi all. I'm totally new to shading languages and i'd like to learn. I've seen some examples and this thread seemed to explain my problem. It doesn't, but here it is anyways:

vec3 LightDir = normalize(gl_LightSource[0].position.xyz);
LightDir -= v;

I get the error :
gl_LightSource left of '[' is not of type array, matrix or vector. Did i forget something? Please help...

Thanx,
Marty

[jeickmann]

what graphics card are you using? In 3dlabs drivers, gl-state based on arrays or structs is not yet supported. So you have to supply this info as a uniform.

If it's an ATI, it "should" work. At least you shouldn't get a compile error. As stated above, the implementation is buggy insofar, as that you have to manually multiply the light-pos with the modelviewmatrix.
I mailed a test-shader to ATI devrel, so hopefully, it will be fixed in the next driver release.

Regards,
Jan

[Marty666]

Ok, thanx... Suddenly it works I downloaded the new ATI drivers (for my radeon 9600 xt, 256MB) and it works fine now. I get a link error when I use noise functions though. This is what i did:

vec4 SomeRandSeed = vec4(1,1,1,1);
vec4 MyNoise = noise4(SomeRandSeed);

Ow. I read some 'tutorials' but I still can't find out how to use the uniforms. I can declare them in the shader, but how to I communicate from my application (in cpp)?

Thanx,
Marty

[Marty666]

Ok, I got it now... Thanx anyways for the help. My first shader works, per fragment lighting and my own dot3 bumpmapping