I used to follow this list, but have been away from it for a while.
Now I've got a problem that is driving me nuts, so could really use
some help from the real scanner savvy members. The question is
simple, but the background is a little involved.
I scan extremely large (up to 10"x60") panoramic negatives in pieces,
then stitch them back together in Photoshop. I've done this for some
time now with generally good results, but have had more trouble than
expected when using the used Umax Mirage IIse (a 12x17 flatbed with
transparency unit) I bought a year or so ago. I'd hoped the bigger
scan area would make the job less tedious since I could do fewer
stitches, but the stitched areas are much harder to match with this
scanner than some others I've had.
I thought the density was just drifting between scans, but finally
started noticing some trends and did a couple tests to confirm what I
was seeing. I've posted a file that shows the problem:
This is a a sheet of paper laying across the 12" width of the
scanner, but scanned as a transparency (i.e., using the light in the
scanner lid shining through the paper). I've pumped the contrast up
to better show the problem. So, the dark side is one end of the light
tube and the light side is on the other. Evenness along the 17"
length of travel is fine, as you would expect. Scans of prints have a
similar problem and tests show that the result is almost identical,
so it does not seem to be the light output itself, but maybe falloff
in the optical system.
All that brings me to the calibration question. Shouldn't calibration
adjust for this? Since the calibration is done with light going
through the whole system from light source to sensor, it seems like
it should balance for whatever is going on here.
So, is this thing really calibrating or not? I normally use VueScan
to drive it, but have used and tested with the Umax Magicscan driver.
This issue is unaffected. If I choose "calibrate" in VueScan, nothing
happens, but with either driver, every time I start a "preview" I get
a dialogue saying "calibrating". The first calibration run of a
session (the first time you preview) is quite long (maybe a couple
minutes), later ones are faster.
Otherwise, the combination of this scanner and VueScan works fine for
these large negs since I don't need a lot of resolution anyway (at
least it is fine if the negs aren't excessively dense). Here's one I
did recently:
Mark
----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or
body
