[Northkeep] Hey, Old Folks/Nordic Saga

Fitzmorgan at cs.com Fitzmorgan at cs.com
Tue Feb 26 16:45:08 PST 2002


--
[ Picked text/plain from multipart/alternative ]
In a message dated 2/26/02 2:52:23 PM Pacific Standard Time,
istvan at micahtek.com writes:
>
> I agree with Zahava. To satisfy both views, I propose the following
> (which is done with period MS quite often).
>
> Create exact images of the documents, smudges, faint words, and all.
> Leave one copy of the images alone. On a second copy, make the corrections
> and clean them up so they are readable. You may want to go so far as
> to OCR them into real text and convert them into (HTML?) documents,
> with embedded graphics. Link this copy to the old copy. (If you just
> clean up the graphic, and don't create text, you may want to create an
> "index" graphic with the original and the "clean" one side by side.)
>
> If you like this proposal, go for it. If not, don't. ;) It will take more
> time, but the final result will be much nicer in my opinion.
>
> Istvan


       My experience with OCR has not been very satisifying.  It's does Ok if
you have a high quality original in a large type size but as the print gets
smaller or the quality drops you spend nearly as much time correcting OCR
errors as you would retyping it.  An unless you are a better proof reader
than I am you get nearly as many errors in the final product.
       B&W or Grayscale images scanned at somewere between 100 to 300 dpi
then converted to a jpeg should be very readable and small enough to make the
archive a manageable size.

Robert



More information about the Northkeep mailing list