Try making 5 scans over and over again. Do the bands show up in exactly the same place each time? Can the scans be stacked assembled and averaged?
This is one way that random noise can be reduced in difficult scans.
A slightly different approach was used with an Agfa Duoscan. I was able to produce banding in different areas by shoving the holder farther and rescanning. By averaging the banding was less problematic.
Bookmarks