lostinspace wrote:
> Jon Noring wrote:
> IMO, your expecting far too much to get a 2400ppi plate quality scan
> from an older book? The paper, printers and ink methods used in
> older publications is NOT comparable to today's methods and quality.
This book dates from 1940 (and going through the process of securing
permission -- it does look like the copyright was renewed), and it is
of *excellent* typeset and printing quality.
In addition, this book is a classic about typography, with the book
itself typeset specific to the book's recommendations. If any 20th
century book is best reprinted in facsimile form, this is certainly
one of the few.
Regarding 2400 dpi, I did some trial scans at 600 dpi optical, and
there was a clear degradation in quality of reproduction of the text.
This apparently occurs since a pixel has finite width and must either
be black or white (assuming 2-bit b/w), even if the center of the
pixel lies on the boundary between the black letter and the white of
the paper -- this is what I saw at 600 dpi -- jagged edges which will
affect the reproduced letter quality -- the effect on the serifs was
even more pronounced. One must go to a significantly higher rez.
(The plate processes are at 2400 or 3600 dpi or so, so they have high
resolution capability in making plates.)
> As far as retaining the condition of the book your scanning from?
> IMO, a better solution is to search the used online book stores for
> a 2nd copy in which the bindings and cover are already in bad
> condition or perhaps even coming apart. In most instances these type
> of items may be purchased very cheaply as compared to the cost of a
> book in which you desire to preserve the overall condition.
This I thought of, but it is a hard-to-find book (Bill Hill at
Microsoft, who highly recommended this book, spent a few years looking
for a copy). The copy I do have on loan from a university library does
have the binding coming apart, but it is not my book to rip apart.
> Another alternative is to see if some text that you desire has
> already been preserved?
I have close ties to Project Gutenberg. <smile/> I know what's out
there.
> Cornell has quite a bit online in a Making of America section:
> http://cdl.library.cornell.edu/moa/
I'm quite familiar with MOA, having written about it for the last few
years on the Ebook Community forum. A few years ago they were cutting
edge and pointed the way towards how to properly digitize texts.
> In summary, there is no general rule in quality of scans. I've seen
> older paper and printing methods vary greatly from week-to-week or
> month-to-month in publications making some scanning and OCR a
> nightmare.
As I've said, these particular books are special cases:
1) They should be reprinted exactly as they are now (facsimile).
2) They are of very high print quality (helping facsimile printing).
3) They also need to be digitized for online viewing, and also
converted to electronic text, probably TEI-Lite marked up XML (as
Distributed Proofreaders is moving towards.)
This points to no ordinary scanning for these (the scans must also be
used for facsimile printing.) It's been mentioned that the best
resolution one can *effectively* achieve with the common flatbed
scanners is about 1000 ppi (even if they advertise a much higher
optical resolution), which does not appear to be quite good enough for
high-quality facsimile prints. It was mentioned as an alternative to
use a drum scanner (which I know nothing about) to get true 2400 ppi
but this requires I rip the book apart, which I may inquire on doing
(so long as I rebind it at my expense -- the binding is very plain.)
So, anyone here willing to volunteer their drum scanner for these
two books?
Thanks!
Jon Noring