Tangent, Bitangent & Normal

[toneburst]

Hi,

please excuse another newbie question.

I’m working on converting some HLSL code to GLSL. I have an HLSL vertex shader that distorts a mesh via a number of different surface formulae. The formulae output vertex position (as a vec4), and bitangent and tangent vec3s (all presumably in Object coordinates).

I’ve been able to transform the vertex coordinates by multiplication with the gl_ModelViewProjectionMatrix. What I’m having trouble trying to do is figure out how to create the normal from the Bi and Tan variables, and transform it into the correct coordinate space, so I can use it to apply lighting effects.

I’m currently not applying any transformation to Bi and Tan, and using the following code to attempt to calculate the normal:

normal = normalize(gl_NormalMatrix * cross(Tan,Bi));

which doesn’t seem to do the job, as I get a black unlit surface with my Phong Directional lighting.
I’ve tried setting normal to the mesh’s original normal using

normal = normalize(gl_NormalMatrix * gl_Normal);

(where normal is a varying vec3)
This results in a lit surface (though obviously it’s not correctly-lit, as the normals from the original plane mesh are incorrect for the distorted mesh).
I infer from this that it’s definitely a problem with the way the normal is calculated.

Anyone know what might be going wrong?

Thanks in advance!

alx
http://machinesdontcare.wordpress.com

[NiCo1]

Maybe the resulting normal is facing the wrong way?

Try:

normal = normalize(gl_NormalMatrix * cross(Bi,Tan));

[toneburst]

Hi -NiCo-

Good thought!
Unfortunately, I just tried that, and it doesn’t fix it, sadly…

alx

[NiCo1]

Are you sure the Bi and Tan variables have the correct values?

[toneburst]

not at all, sadly. I’m pretty sure the original HLSL code worked.
I know the vertex position code works. Without actually seeing a visible effect though, I can’t really tell if the BI and Tan values are correct.

alx

[toneburst]

Incidentally, in the original HLSL code, Bi and Tan were multiplied by the
WORLDVIEWPROJECTION matrix before the normal calculation was performed with the code

n = normalize(cross(In.Tan,In.Bi));

I’ve tried multiplying Bi and Tan with the gl_NormalMatrix before calculating the normal, but this doesn’t do it, either.

Don’t know if this throws any light on the situation at all…

alx