[Milsurplus] scanning old photos/books

[WA5CAB at cs.com]

Barry,

I started doing TM scanning in 1999, which seems almost yesterday but is now 
8 years 2 months back.  Since the beginning, I've used Xerox equipment but 
have acquired some scanned manuals done with other equipment.

Parameters to consider (ignoring color) are scanner resolution capability (in 
both directions), storage file format, scanner response (output or contrast 
slope from B&W to 256 gray scale), and file compression methods (somewhat 
contrained by chosen storage file format).

My early work was solely aimed at being able to print hard copy to 8-1/2 x 11 
or larger pages.  The original equipment that I had would print at 600 dpi 
but only scan up to 400.  Initially I found that TM photos scanned at 300 and 
printed looked actually better than those scanned at 400.  Completely forgetting 
earlier work with digitizers and aliasing problems, for several months I did 
all my scanning at 300 dpi.  One day while reading the printer/scanner manual 
again I discovered that the printer would also print at 400 dpi, a mental 
lightbulb flashed on and I ran some test scans at 400 dpi of earlier work, printed 
them at 400 dpi and compared them to the 300 dpi scans printed at 300 and 
600.  What I found was that the 300 printed at 300 and 600 couldn't be told 
apart.  The 400 printed at 400 looked much better.

The next Xerox machine I had and the present one both scan at up to 600 dpi.  
Every once in a while, for one reason or another I have scanned pages at 400 
dpi and sometimes compared them to the same pages scanned at 600 (and if 
printed, printed at the scan resolution).  Mostly, there's little to choose between 
the appearance of the two but as storage space is no longer any issue (I 
currently have about 1.5 TB online not counting removable media) I do almost all 
scanning at 600 dpi.

In the other direction, I will just state categorically that less than 300 
dpi scanning of vintage B&W photos will result in obvious degradation of 
reproduced image quality and not go into details of comparisons I've run.

Now, ideally, one would match up the scan pitch to the printed dot pitch. for 
the best results, but that would take considerable extra work and for optimum 
results (best results with minimum file size) require not only matching the 
pitch but syncing the scanner with the printed dots.  Given differential 
shrinkage of typical vintage pages (the dot rows and columns aren't straight nor 
exactly at right angles to each other), it would take gigabuck equipment.

So that's it for scanning resolution for photos.  Except to say that less 
than 300 dpi scanning will also degrade legibility of printed text, especially on 
schematics and wiring diagrams which often have small print (test also run on 
this).

The preferred working file format for B&W work (either text or B&W photos) is 
TIFF.  People with XP and later MS OS's will have trouble viewing this as for 
some unknown reason, MS no longer ships a working .TIF viewer, although they 
apparently still own the rights to one.  However, there are a number of third 
party viewers that work OK, although most lack editing capability.  JPEG is 
not suitable.  Besides having lossy compression, the file sizes at a given dpi 
are too large.  TIFF files can be converted to PDF with no loss of quality.  
However, all cleaning and straightening must be done before conversion as none 
of the PDF editors I'm familiar with have that capability.

Apparently most of the consumer grade scanners (unfortunately I've not used 
any of them, just seen some of the results) do color and 256 gray scale.  From 
Barry's comments I gather his also does black and white (or something other 
than 256 gray scale).  The difference between 256 gray scale and black and white 
is the contrast slope.  The Xerox scanners do 256 gray scale (called Photo), 
black and white (called Text) and one or sometimes two intermediate slopes.  
The DC430 I currently have does one intermediate called Auto which I think the 
generic name for would be Half Tone.  Auto works best on photos in manuals.  
The only use I ever found for Photo was on actual original photographs (I have 
a set of 8 x 10's from the company that built the CDA-T for the AN/ART-13 and 
it worked fine on them).  Text works best for text as it will ignore a lot of 
background yellowing of the pages.  My SOP is to scan an entire manual with 
the Text setting and then go back and scan all pages with photos with the Auto 
setting.  I then cut and past the photos into the text page.  I often have to 
play with the brightness (light to dark) control depending upon the manual.  
But this produces text generally free of background clutter (defects like punch 
or staple holes, coffee or beer stains, dirt, etc. will have to be manually 
deleted).

The best compressions algorithms for text with or without half tone photos 
appear to be CCITT Group 3 or CCITT Group 4.  You may have to look around in 
your editing software in order to find them.  They are apparently only applicable 
to black and white.  For other slopes, you will find LZW and JPEG.  Look 
around in your editor.  It appears, from source documents I've received from other 
sources, that many scanners (except to JPEG) save the file with no 
compression.  Applying LZW compression to a 256 or 16 gray scale file may drop the 
physical file size by 70-80 %.  I've also found that you can play a lot of tricks 
with an editor that handles everything from True Color down to Black and White 
and also supports copy and paste.  

Finally, what the scanned files look like when you view them depend greatly 
upon what you view them with.  As I indicated earlier, a photo scanned at 400 
dpi and printed at 600 dpi looks like hell.  The same can be true of what you 
see on a monitor, depending upon the scanned dpi versus the monitor maximum 
resolution (actual physical dot pitch) and current resolution setting.  And the 
image appearance can vary greatly at different zooms.  And may not mean 
anything.  Look at the image as you expect it to mostly be viewed.  

In a message dated 7/16/2007 10:19:40 PM Central Daylight Time, 
btuttleman at worldnet.att.net writes: 
> i have a small collection of older military unit history books; i 
> started w/ this one entitled "Historical &Pictorial Review, 40th
> Infantry Division, Army of the United States, San Luis Obispo, 1941"
> and attempted to scan some of the radio/communications/weapons/vehicles
> photos - here's an example from this work:
> 
> http://home.att.net/~btuttleman1/radio1.jpg
> 
> my eyes are constantly degrading, and this scan appears pretty "grainy"
> and not very clear to me, but when comparing it to the original it kind
> of looks pretty similar; anyone (robert?) have any ideas on scanner
> settings for older photos - this one was done in .jpg format @ 200 dpi;
> i tried it at other resolutions, but it only appeared to make the size
> (memory-wise) much larger, not clearer - so i'm thinking that this is 
> what i'm going to get out of these scans.  greyscale sent the size of
> the photo thru my roof.
> 
> some of these photos might prove very interesting as far as what was in
> use at the time.
> 

Robert & Susan Downs - Houston
<http://www.wa5cab.com> (Web Store)
MVPA 9480
<wa5cab at cs.com> (Primary email)
<wa5cab at houston.rr.com> (Backup email)
   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.qth.net/pipermail/milsurplus/attachments/20070717/537b63ac/attachment.htm