Frederik Ramm wrote:
>
> Hi,
>
> at first I thought it was a problem with the driver I'm writing but
> the Windows driver didn't deliver any better results. When scanning in
> color at the highest resolution (300x600dpi), black letters on a white
> background exhibit a green "shadow" to one side and a red "shadow" to
> the other. The shadow is about 3-4 pixels high. I've put an example
> here:
>
> http://www.remote.org/frederik/tmp/gap8.ppm
>
> You'll have to magnify the window to see it clearly.
This looks like a common colour merging problem. In most scanners there
is a physical separation between the red green and blue sensors. Most
scanners just feed data to the host with this separation, and the driver
needs to shuffle the scan lines to accurately merge the colours. You
will find code to perform this function in several sane backends. A
typical offset between each sensor row is 1/75 inch, but varies between
models.
In your example scan, the physical shift looks quite small. It looks
like the software is attempting to merge the colours, but doesn't use
the correct physical offset.
> I would be very happy if someone who is a bit more experienced with
> scanners than I am could look at this and tell me if this would be
> considered "normal" in a medium-priced scanner (Fujitsu M3091, ca.
> USD 800) or if I that should be considered a fault in this specific
> model and I should send it back to the shop for an exchange?
Today an $800 scanner is hardly mid range - its a rather expensive
scanner. Even a $50 scanner usually merges the colours well enough that
colour fringing is very hard to detect.
Regards
Steve
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com
This archive was generated by hypermail 2b29 : Thu Mar 01 2001 - 18:54:04 PST