-------- Original Message --------
Subject:        Re: Aerial Photos Scanning Parameters Question
Date:   Tue, 2 Sep 2008 20:40:33 -0500
From:   Dennis McClendon <[log in to unmask]>
To:     [log in to unmask]
CC:     [log in to unmask]




> These Mylar sheets are custom enlargements produced for CMAP’s
> predecessor, the Northeastern Illinois Planning Commission.

From my background in print production, perhaps I can shed a little
light on the technical issues, to help you make an informed
decision.  In some ways, you're trying to reconstruct an apple from
the applesauce.

There are two basic categories of images: panchromatic or continuous-
tone photographs and orthochromatic "linework."

Continuous-tone images are captured on films or photographic papers
that can record a variety of values for each grain of emulsion, or
their modern equivalent, digital sensors that record various values
for each pixel.

"Linework," by contrast, has only black or white--ink or not ink--in
various places on the paper.  This is obviously the case for words
printed on paper, whether produced by traditional printing with ink,
electrostatic/xerographic processes such as laser printers, or light-
exposure/chemical processes such as blueprinting and diazo.

To allow reproduction of continuous-tone images as ink on paper, the
halftone process was developed.  A continuous-tone image is masked by
a screen and optically broken up into "linework" dots of various
sizes which the eye perceives as tones of different values.  In
offset printing, this was done as a publication was being produced,
and photos and type together were processed as "linework" as they
went from "pasteup" to printing negatives to printing plates to ink
on paper.  These days, conversion of continuous-tone images into
halftone dots is done at the last minute by the "RIP" calculations of
the laser printer, imagesetter, or printing press, just before ink or
toner is put on paper.  The RIP expects to be dealing with continuous-
tone or linework images only.

Your Mylar masters already have the halftone dots in them, and there
is no process that can convert those back to continuous tones except
by reducing the resolution by a factor of roughly eight.
Essentially, you're telling the scanner and computer to look at a
group of four or nine halftone dots, decide on an average value, and
represent that in one pixel of the new continuous-tone image.  It is
the same process as scanning a coarse halftone from an old newspaper
for reproduction in a modern book.  Twenty years ago, we were able to
keep that as "linework" from beginning to end, a technique known as
"dot-for-dot" reproduction.  We maintained the old newspaper's
halftone dots as linework dots all the way through the production
process.  In today's Photoshop/digital world, this is difficult
except for extremely coarse halftone images.  Among other problems,
when a RIP introduces halftone dots to an image that already contains
halftone dots, moiré patterns result from the interference patterns
of the two screens.  The only solution is to blur the halftone image
so that the RIP can introduce new dots into what appears to be a
contone image.

This is the basis for my opinion that digitizing the Mylar halftone
images will result in a product that is only useful at a quarter of
the existing scale or less.  Houses and streets will continue to be
visible, making a digital version useful for regional land use
history and change detection, but not for more detailed site
analysis.  Worse, the halftone dots will introduce artifacts into the
imagery that may be interpreted as features on the landscape rather
than on the Mylar.  If the continuous-tone photos are available, they
would be a much more valuable resource.  As for the current Mylars,
perhaps a search for a new diazo service would be easier than
creating and storing a 75-megabyte image for more than 900 sheets.