Scanning (was Re: recent acquisitions for the House of VAX

From: Carlini, Antonio <>
Date: Tue Sep 11 05:52:10 2001

Jim Battle wrote:
> At 02:16 PM 9/10/01 -0700, you wrote:
> excess. The KA655 TM
> >that is on the DFWCUG site appears from the PDF to be 100
> DPI/8 bit (note
> >that's 800 bits per inch net and the result is harder to

I don't know where 100dpi/8-bit came from -
its 600dpi/1-bit (apart from maybe one or two pages
which are 600dpi/8-bit). I don't know how
to persuade Acrobat to tell me what the
underlying resolution is, but I expect it's

> I agree that 300 dpi _at_ 1bpp is superior for text and line art
> than 100 dpi
> _at_ 8bpp, but I think your math is askew. You need to compare
> the # of bits
> per sq in, not inch. So 100 dpi _at_ 8bpp is 10KB/in^2, while
> 300 dpi _at_ 1bpp
> is 11.25 KB/in^2. For text and line art, though, the 1bpp image will
> compress a lot better.

I make 600dpi/1-bit to be a smidgen under 44KB/sq-in.
This is an A4 scan (of a photocopy of an original
which is less than A4 in all dimensions IIRC) so
thats 8.5x11.5in i.e. anything less than ~4MB/page
is compression kicking in! I think the pages
average something like 300KB/page.

I get PDF direct from the scanner but I don't think
it does G4 compression. If anyone has a tool which will
take in a PDF and spit out a G4-compressed PDF,
I'd be very interested. I think I've seen people
claim that 300dpi/1-bit scans run at about 60KB/page
so I assume that G4 compression would bring these
documents down to about 250KB/page (15-20% smaller).

Of course, "perfect" OCR would knock it down
a good deal more than that!

> And if, for some reason, you want to scan line art and text
> in gray scale,
> 4bpp is plenty enough. Use 8bpp only for continuous-tone images.

Possibly true (although almost every page
that I use 8-bit for does have a photo
of some gubbin or other). But the scanner
spits out 1-bit or 8-bit and there are not
that many knobs available for tweaking.

Received on Tue Sep 11 2001 - 05:52:10 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:34:24 BST