For a look at ss4k vs. 5000 dpi from a Heidelberg S3900
with Dmax 4.5 see
http://www.amber.ucsf.edu/~ross/gallery/tech/banding3.html
(which starts out as an investigation of banding on the ss4k
that I suspect may have been due to dust on the sensor).
My favorite 'look' of the numerous things I tried, including
with the Heidelberg scan, was based on Vuescan w/ the ss4k
that had the original banding problem ("adjusted Vuescan raw").
At 5000 dpi (last picture), you can see differently colored
dye clouds.
Bill Ross