If you limited your enlargement to a magnification that left you with
450 dpi at output on an imagesetter with 3600 dpi, then if everything
else was perfect I'm not surprized you got convincing platinum prints
from the negs. 3600/450 is enough to get the right amount of data in
each pixel (though 4800/600 would be even better and result in smoother
tone). But it's real easy for everything else not to be perfect.
The problem is that imagesetters weren't designed to do this. Operators
think entirely in terms of dot percentage for prepress use, while as
Fokos points out carefully in his paper, you need the correct "optical
density" as well. They aren't used to dealing with precise optical
density. A good service bureau should of course still calibrate when
they switch rolls of film (they also have to maintain the
roller-processing chemistry properly: imagesetter film itself isn't
really that different from other silver materials). Banding is less
likely to come from mechanical/chemical problems than from improper
settings in their software to work with your files.
A final point is that you'd be amazed how much film is re-run at service
bureaus. Handing over six files and expecting back six perfect pieces of
film the first time just ain't the way things are done. It's part of the
workflow to have a big reject rate. Partly because designers are
constantly changing their minds about color, last-minute type changes
happen all the time, and I think it's even important that imagesetter
film is generally an emphemeral intermediate product between design and
printed piece. (The client in fact often never sees a piece of film
during production: dyelux and matchprint proofs are inspected for
placement and color: no one but the techs handles or views the film.)
The mind-set regarding "film" at a pre-press house is vastly different
from photographers mind-set about their negatives. You'd cringe at the
way they physically handle the stuff on the stripping tables (g).
---Carl