Glad it helped. Actually it was kind of a simplified version of what is probably going on. In all likelihood there isn't just one norm per poly, but rather one for each vertex of each poly. The general concept of what's happening is the same but the actual details get a bit more complicated than what I described.