[sane-devel] pixma MP810 working with sane

Nicolas nicolas.martin at freesurf.fr
Thu Sep 4 06:29:08 UTC 2008

Le jeudi 04 septembre 2008 à 03:17 +0200, Arne a écrit :
> > Concerning the vertical lines at 4800/2400 dpi for MP810, I'm wondering
> > whether:
> > 
> > - this is a normal behavior of the scanner (which would mean that it has
> > to be compensated by the backend), 
> > - or if it's an alignment problem of the sensor, the lines corresponding
> > each one to a particular sensor cell, but only at high dpi, the sensor
> > uses all its cells to scan.
> > 
> > I think it may be interesting that other MP810 users could give feedack
> > on the current CVS backend, if possible, that's why I copy also the
> > sane-devel list
> Yes I hope we get some feedback, as there was at least one other MP810
> user asking for the drivers.
> I think it's normal behaviour, but if not, I need to know before the 
> warranty runs out!
> > Do you know if there's an alignment procedure for MP810 ? This is
> > usually the case when you get a new scanning device.
> There is a printer-head alignment procedure but nothing for the scanner.
> > Have you the possibility to test the Canon Windows driver, and see if it
> > behaves the same ? 
> I borrowed a WinXP laptop and installed the latest drivers downloaded from 
> Canon Europe and found that they only support up tp 1200dpi.
> If I was a Windows user I would feel very cheated by Canon about this as 
> they advertise 4800x4800dpi optical on the tech specs!
> In fact I might send a message to Canon support about this and see what 
> they say.
Same situation for more recent models like MP610, MP970, ...
Canon drivers offer only up to 600 dpi or 1200 dpi, far less than the
4800 dpi advertised ! Never tried to contact Canon about that, but
should be worth doing it. 
AFAIK, only Sane pixma backend can currently drive those PIXMAs at such
high dpi, not Canon's drivers.

> > Last point concerning the white lines at the bottom: this is simply due
> > to the color planes shifting, as the backend algorithm to shift the
> > color planes uses only that portion of the image that will contain data
> > for the 3 colors. The bottom of the scanned image contains 2 and 1 color
> > zones only, due to the color shift. 
> > So this is a little limitation for pixma CCD sensors, you must select a
> > scan zone a little bit bigger at the bottom, then remove the image
> > bottom white zone, that is padded on purpose by the backend to keep the
> > image size the frontend expects.
> ok I understand, it's not great but I guess there is nothing that can be 
> done about it.
Following Allan's suggest, should be easy to implement, I'll propose you
soon a newer pixma_mp150.c file which should solve this issue, I'll keep
you posted.


More information about the sane-devel mailing list