Since it does not compare high and low res meshes it does not even see
seams in uvws. It only produces normal maps from the light information
in 2d textures.
It can produce some seams around the edges of the image. To counter
this I simple increase the canvas size before the main algorithm runs
and then shrink it after, any of the seams gets clipped. however if
you run my script on a layer that has transparency of any kind it will
put some very ugly seams into the image that I can not control.
Does this answer your question?