Comment 3 for bug 1374594

Ah, now that you've posted it, it's obviously a better solution. Thanks.

I think there's still a problem, though: the computation of texcoords at lines 267-279 uses the names of the filters instead of the list of texcoords that the filters want. Similarly, the provided texpad inputs are linked to the filter names, not to the textures they refer to.

The pattern is pretty simple:

l_texcoordC=(vtx_position.xzxz * texpad_txcolor) + texpad_txcolor;

so if it is known that l_texcoordC corresponds to txcolor, these lines could be automatically generated.

I could do this unless you want to?