See post 84.
OCR technology embedded in the scanning process could very easily produce layers as it looks for text of the same shade and separates it out from the background “noise” in order to pass it through an OCR algorithm.
So, it depends on the scanning technology used, I think.