How to transform point normals into camera space

[magneto]

I have a geometry with 2 points (positions = both {0,0,0}, normals = {0,0,1} and {0,1,0}). If I parent this geometry to a camera, I get the points and normals correctly transformed:

But if I transform them myself without parenting, using optransform(/obj/cam1) in VOPs by multiplying with both point positions and normals, I get this:

The geometry is not transformed, only camera is.

Any ideas on how to fix this?

Thanks

[magneto]

I figured it out. I had to multiply the normal by the inverse transpose of the same matrix

[tjeeds]

I think you just need to convert your M4 to M3, you don't want to apply translation to vectors, only rotation.

[magneto]

Thanks tjeeds, how do I do that in VOPs? I looked at the Convert VOP but it doesn't have matrices.

Also you are talking about the normals, right?