Scanning & TWAIN

["Michael Healy" <mjhealy@xxxxxxxxx>]

I wonder whether I might get some thoughts and feedback on a few problems
I'm having in the use of TWAIN.

I am scanning with an Epson 2450, TWAIN, and PS 5.5, driven by a Pentium 3
with 12-gigs and 96 megs of RAM. I am interested primarily in making
high-quality scans of b&w negs for digital enlargement to be used in contact
printing. My negs are 6x9 and 4x5. I am prepared to work my images in
PhotoShop, but I want very much to maximize the scan quality in the first
place so as to avoid excessive handling and "fixing" as a result of
incompetent scans. Right now I am experimenting with TWAIN's gray scale --
scanning, evaluating histograms, then rescanning with scale adjustments. I'm
not only trying to master the options, but I want to nail down the variables
so I know what I'm doing when I think I need to adjust one.

That is the background. My trouble is that what I want requires more control
than TWAIN seems willing to afford. Or rather I should say, more information
than it will surrender. I frankly do not find TWAIN very forthcoming about
what it's actually doing in there. EG, I can alter the scan's gray scale,
but I cannot figure out what the gray scale actually is by "default" before
I alter it. Does anyone know what TWAIN's "auto" gray scale values actually
are? How about its "reset" scale? Do I need to hunker down and coax this
information indirectly from histograms by experimenting?

Another problem seems to be inconsistencies in these outcomes. Last week I
scanned a neg with "auto". It seems to open up shadow detail. (This was on
negs scanned into positives, though, so my observation is backwards from
what it is when I scan them as neg-to-neg.) Also, my use of "auto" does at
least establish a benchmark. From there, I am systematically adjusting the
pre-scan gray scale and rescanning. This is giving me a number of carefully
planned outcomes, with histograms that I am able to evaluate. It is a
wonderful exercise in how to get intuitive with histograms -- kind of like
learning how to instinctively gauge the need of a subtle change on the
enlarger.

Today, though, when I repeated the same scan, I actually got a different
histogram result! It's a better one, with values more evenly distributed;
but I didn't want a better one, I wanted the same one. I wonder why I'm
unable to achieve the identical result. Incidentally, this is not because I
was playing with bit depth or size. And so I keep circling around the
meaning of "auto" and "reset" in this thing. These variables seem to elude
precise defintion. As a photographer I am accustomed to making careful
shifts away from an established value, AFTER establishing that value, not
simply as arbitrary shifts in relation to some undefined "default" I can't
measure.

Other questions about TWAIN while I'm on the subject:

(1) Is it me (or the 2450), or does TWAIN not permit cropping with the
overhead transparency reader? I preview for a print, and I am able to crop.
I preview for a transparency, and I can't. Probably there is some logical
design reason for this, but it sure seems to be one of the stupidest
shortcomings I've encountered in all my years w/computers.

(2) How come I am unable actually to handle a scanned image unless I first
close TWAIN? The images are there on my PhotoShop workspace, but TWAIN won't
surrender to Photoshop. This means, EG, that I can't evaluate an image's
histogram on the fly. I have to close TWAIN, play with the image, reopen
TWAIN, and kick in yet another preview. This is time-consuming, and seems
ludicrous. I don't seem to recall experiencing this problem last summer
while using TWAIN through PS 6 on a new Microtek (in Mac...) Is this a
limitation in my own system? Even a single scan of 500k can't be accessed
w/o closing TWAIN, so I seriously doubt it's my RAM.

(3) What is the difference between OCR, Epson Printer (Fine), and Epson
Printer (Photo)? I understand that OCR is for line drawings and especially
text. What I can't seem to pin down in my readings, though, is the
implication in a scan of a continuous tone slide or neg. And what
implication is in the Printer choices? Are these just more consumer-pleasing
bells and whistles, or do they alter outcomes? If so, in what measurable
ways?

(4) What about other scan drivers? I have tried SilverFast and absolutely
despise it. If TWAIN throws us impenetrable "auto" and "reset", SF needs to
dump two dozen slabs of mystery meat labeled "Agfa" and "Ilford Delta 100",
whatever the hell THESE built-in gray scales even might do. To someone
trying to actually measure and gauge, this seems like 50 steps back, not two
steps forward.  Mike Demyan has suggested VueScan. I've reviewed their
website, and it looks promising -- does it address some of these issues more
transparently? Are there other pros and cons?

Thoughts and feedback appreciated.

Mike Healy


-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.