Date   : Fri, 20 Jun 2008 08:26:58 -0500
From   : jules.richardson99@... (Jules Richardson)
Subject: The Micro User

Brian Foley wrote:
> On Fri, Jun 20, 2008 at 12:20:40PM +0100, Dave Moore wrote:
>> I'll take them, as I know someone who is willing to scan in old TMUs (he
>> actually has a great track record in scanning old retro mags!).
> 
> On a related note, does anyone have any advice on nondestructively
> scanning old mags like these?

If they're stapled, I normally take the staples out, scan the pages, then 
re-staple. Anything that's glued / stitched is a bit more tricky - scanners do 
exist to handle this, but they're incredibly expensive.

> How do people handle the curvature introduced by perfect binding

There are actually tools around to compensate for this - I've not tried any of 
them yet, but when I searched around for such things a few months ago I found 
ones for both Linux and Windows, free and otherwise. I've seen a few results 
of such tools though and they weren't always very good.

> and what file formats/resolutions are regarded as adequate?

Me, I tend to do around 300dpi greyscale - I don't like doing bi-level because 
if there's page contamination (dirt, creases, stains etc.) then there's a risk 
that data will be lost. With a greyscale image there's at least the option of 
doing some post-processing to remove unwanted artifacts before a subsequent 
OCR step (I don't think that OCR technology's 'quite there' yet, so I leave 
everything as scans for the moment)

For colour pages I'll either used 8 bit or 24 bit colour as appropriate 
depending on the content.

Capturing as much as possible seems important, because lost info can't be 
magicked out of thin air :-)

> I know Al Kossow (at <http://bitsavers.org>) prefers 400 dpi scans
> saved nonlosslessly because OCR packages tend not to like the
> artifacts introduced by JPEG compression. This would get a bit
> unweildy for full colour images though. Is JPEG2000 better in this
> regard?

I believe JPEG2000 has a true 'non-lossy' mode - but of course the format's 
not as widely supported as TIFF, so there's not really any benefit in using 
it. Just use TIFF; pretty much anything can handle that and the compression's 
non-lossy (make sure you actually have compression enabled - lots of tools 
default to saving an uncompressed image)

Yes, the storage requirements get big pretty quick. Make sure you have 
backups, too :-)

cheers

Jules