Humanist Discussion Group, Vol. 17, No. 352.
Centre for Computing in the Humanities, King's College London
www.kcl.ac.uk/humanities/cch/humanist/
www.princeton.edu/humanist/
Submit to: humanist@princeton.edu
Date: Tue, 28 Oct 2003 07:58:27 +0000
From: "Dennis Moser" <aldus@angrek.com>
Subject: Re: 17.344 scanning
Hmmm. As an archivist heavily, yea verily, intimately involved in
digitization work and a photographer, I feel that I should comment on
some of the statements here...some of this with respectful disagreement.
> Date: Mon, 27 Oct 2003 06:48:08 +0000
> From: "Jim Marchand" <marchand@ux1.cso.uiuc.edu>
> >
> Francois' point is well taken, but things have changed. There
> was a time when ones scanner offered 150 dpi (dots per inch), but now
> we have digital cameras with many megapixels. I would not
> use a scanner nowadays, but rather a digital camera.
Uhm, for large-scale production, this isn't always the best of
decisions...
> With the
> scanner, one has always the possibility of damaging the binding, etc.,
> whereas with a camera, one has the possibility of doing
> filtration in real time, seeing the result before `printing',
> etc.
Depends on who is handling the works. I would want to be certain that
any technician manning the scanner was competent in handling
rare/fragile materials before they ever touch the materials to put them
on the scanner...only common sense, really.
You can see your results after scanning just as well as with a digital
camera, better actually if you have a high resolution monitor on the
wrokstation (and you DO have a high resolution, color corrected monitor
on your scanning work station don't you?).
Furthermore, just because you're using a digital camera doesn't mean
that you don't require a.) competency in handling rare/fragile materials
(you still have to move them around for photographing, etc.,) and b.
photographic skills.
> Francois' point on analog vs. digital is also good. A
> digital picture is like a pointilliste painting and consists of
> many separate (discrete) points, whereas one has a tendency to
> look upon film as continuous. It is not at all continuous,
> really, and zooming in or enlarging is likely to involve one with
> `grain in the negative' (breakup of the silver salts).
True, film really only gives the impression of being continuous tone. It
really is just as "pointillistic" as digital when enlarged sufficiently.
> The Nazi
> who took the photographs, if he used the best film and camera
> available (say a Leica 35 mm., with Adox KB 14 film, developed to a
> gamma of 7), would still have more of a grain problem than a
> present-day photographer with a 6 megapixel digital camera.
Uhm, no. I'd put my money on our hypothetical Nazi, especially if s/he
is using a Leica with that wonderful glass of those lenses. You might be
thinking of the stereotypical image of the spy using a Minox (which uses
16mm, considerably different in its ability to capture information,
htough I have seen exquisite 16" x 20" prints done from a Minox negative
that defy their tiny origin...). In the hands of skilled photographer,
grain can be developed in such a manner as to greatly diminish the
effects, yielding a negative capable of tremendous enlargement. The fact
is that the 35mm format, by default, yields a file anywhere from 18Mb to
80 Mb in size, dependent upon the particular scanner used (all over the
counter, readily available machines). That magic 6 megapixel camera only
yields a file of +/- 36 Mb in size. Period. More on this later...
> One
> needs to keep the analogy of the pointilliste painting in mind.
> Many photographs of yesteryear were printed with a half-tone
> screen and (like the pictures in the Sunday newspaper) are really dpi,
> as you can easily determine by magnifying them.
This is only true if the print was created for newpaper printing.
> Were I the keeper of an archive and wanted to preserve a record
> of my holdings, I would get an SLR (single lens reflex, to avoid
> parallax) 6 (or more) megapixel camera (street price ca. $1000,
> 1K) + filters (write the camera company to see what they
> recommend).
Sorry, you might want to do some more research on this. A Nikon D100 or
a Canon 10 D will cost you more than that, even with discounts
available. The bodies alone are over $1000 US, and then you need good
lenses, unless you already have either of those systems in place. And
that only works of the lenses you have are of recent enough vintage to
work with the digital bodies. If you don't have the lenses, the costs
start mounting quickly.
Further, for this kind of work I would strongly recommend looking at the
12Mb - 15 Mb cameras instead of the prosumer 6Mb cameras. The reason
being that you will have MUCH greater resolution than with the 6Mb
cameras.
> With a proper stand and lighting, I would make a
> digital photograph of each page, using the proper filter. Look
> at the photograph before you preserve it, with the filter over
> the lens; see what looks best. Record filter, time, etc. I
> would keep the picture in TIFF format.
The DSLRs (digital single lens reflex) cameras of which we are speaking
do not offer a TIFF output option. This is done in the processing
software. Yes, for storage purposes, TIFF is the most desirable for
these purposes.
{SNIP}
> What the
> megapixel SLR camera bring us is the ability of a total amateur
> (with care) to take satisfactory archival pictures. Do not
> fiddle with the result, put it in storage. If you want to use a
> program to fiddle with the result, fine, but keep your first
> picture in storage. Above all, do not treat your archival
> pictures with non-algorithmic methods.
Er, uhm. I would hesitate to say that any total amateur is going to
consistently be able to provide satisfactory archival pictures,
regardles of the digital camera used. And I'm not sure what you mean by
"fiddling" with the result, as well as "treating" the pictures with
non-algorithmic methods.
> When you take a digital photograph, an LUT (Look-up table) is
> generated, where each pixel has at least the form of f (x,y),
> where x and y are the familiar geometric location (10 over, 9
> down) of the pixel and f represents the gray-level, color,
> whatever, the radiometric information. In general, any fiddling with
> the geometric information could lead to forgery, radiometric is ok,
> ceteris paribus. Every archive in the world ought to
> generate a digital record of its holdings, making it available to
> scholars. Let the scholars do what they will, so long as the
> original is kept. Every record ought to be in TIFF, 6 megapixel or
> more.
Mumbo-jumbo...there are other aspects to dealing with questions of
forgery, chain of custody being one of them. I'll repeat myslef here,
digital cameras do not of themselves generate TIFF files. And a 6Mb TIFF
file, as a result of one of the DSLRs above, will not be of adequate
resolution to use for "archival" purposes.
I think there might be a misunderstanding about what that magic 6
megapixels means. In the current crop of cameras, the limits are
determined by the size of the sensors used. The Canon 1D and Kodak
(sorry the precise model # eludes me here) have sensors that are the
same size as a 35mm film format, i.e., 24mm x 36mm. This results in a
camera that produces a 12 Mb file. Let me say that again: it produces a
12Mb file. Notice I didn't say WHAT file format that is.
All of the high-end Nikons and the Canon 10D, 30D, and 60D cameras
produce smaller sized files (these are the "6 megapixel" cameras).
Again, no comments as to WHAT KIND of file is produced.
What ALL of these cameras produce is a file format referred to as RAW,
as in "raw, unprocessed, uncooked." It means that you need to use
software before doing anything with the files. Software such as Photoshop is
presently unable to read these RAW files because they are of a
PROPRIETARY nature and require PROPRIETARY software to be interpreted.
Each camera manufacturer has their own software for converting these RAW
files into TIFFs or JPEGs, which can then be read by Photoshop or other
software tools.
Of course, the situation is more complicated than that. This proprietary
software applies various corrective (color balance, color temperature,
sharpness, etc.,) algorithms to the base RAW files.
Once this has been done, the result is a 36Mb TIFF file (or in the case of
the two 12+Mb cameras, a 72Mb TIFF file) AND the original RAW file. You
can see that the 72Mb TIFF file approaches the file size of the high-end
scan from a film scanner (Nikon 8000 produces an 80 Mb file from 35mm film
at 4000dpi), so THOSE cameras truly do rival 35mm film. Much as I love
using my Canon 10D, it's still a little shy of the results of a superior
35mm slide, well-scanned (but that's only due to personal financial
impediments! I simply didn't have a spare $8,000 for the 1D body!).
The trick is to keep both the original, unprocessed RAW file, keep the
software for processing them current and treat the file as if it were an
original film negative or slide, i.e., guard it jealously and store them
in conditions to guarantee their long-term viability. The TIFF file than
becomse the work copy from which all else devolves. You might want to
consider keeping a copy of that TIFF file in the same manner as the RAW
file from which it was derived, for much the same reasons.
I guess that's enough for this morning...
Dennis Moser, MILS
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
mailto:aldus@angrek.com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"That so few now dare to be eccentric, marks the chief
danger of the time"
--John Stuart Mill (1806-73)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
This archive was generated by hypermail 2b30 : Tue Oct 28 2003 - 03:21:15 EST