MO,

Thanks for the "Unify Normals" option.

The problem with this method though, is that in forcing unification of the W component of the surfaces, the U & V components may be altered, even in cases where the faces are already aligned such as they with a polyLoft.
If you look at the image below; the magenta dots that are coincidental with the black circles are consistent with the original face alignment. Where they are shown in different locations, the new face has been rotated. Indeed the "showFrames" elements at the centre of each face are consistently aligned along blue axes, but not so with the red or green, as would be the case of the original polyLoft output.



Ideally the finding of surface normals should not alter the referent surfaces, but at the very least as a first step, I think it would be preferable to remove "Surface" output from the node, since it would be generally assumed that surface output would be the same as that input.

The other issue that the join/separate method has it that it reorders the faces, so that any regular indexing of rows and columns of faces in say a polyLoft is lost. This means though that if a user wants to address a normal vector for a particular surface, they are unfortunately not similarly indexed.

I see this problem stumped Wayne, and I'm am realising it must be very difficult. Thank you for persisting as long as you have. I would understand if you threw your arms in the air and said "Too Hard, I give up"

James
https://www.instagram.com/nodeology/