[sane-devel] Coolscan2 IR channel problem
Wed, 22 Jan 2003 11:34:17 -0700 (MST)
OK, this is what I've done. In order to rule out scanner hardware
problems, I installed WindowsME onto a spare computer and tried out the
scanner with its windows driver and Nikon software with Digital ICE(tm). I
understand that it is the technology that makes use of the IR data. It
worked fine. The second IR scan was performed and scatch and dust flecks
were removed as advertised.
I still can't get the scan table to move on Linux, however, and the color
and IR data are still no good. Any suggestions on what to do next?
> On Fri, 17 Jan 2003, tegbert wrote:
> On Fri, 17 Jan 2003, Major A wrote:
> > > > I'm having a problem with my Nikon Coolscan 4000 ED in getting scanimage
> > > > to do its second scan of IR data. I'm using sane version 1.0.9 and with
> > > > this command line:
> > > >
> > > > scanimage -d coolscan2:scsi:/dev/sg1 --infrared=yes --batch-count=2
> > > >
> > > > The scan table doesn't move. I get two files, with the second file being
> > > > about one-third the size of the first one and containing gray-scale data,
> > > > which is OK. Unfortunately, the data in both files are just made up of
> > > > vertical lines, presumably because the scan table doesn't move.
> > >
> > > Just guessing, does the following work?
> > >
> > > scanimage -d coolscan2:scsi:/dev/sg1 >image.pnm
> > > scanimage -d coolscan2:scsi:/dev/sg1 --infrared=yes >ir.pnm
> > Ah, didn't realize the original email one went to the list...
> > No, this won't work because the coolscan2 backend stores the IR
> > information in memory and only dumps it when you scan a second
> > time. This can only work if both scans are done within the same
> > instance of the backend running.
> > The problem here must be something weird because I've done scans like
> > that and it worked (although it's on an older scanner).
> > Andras
> Sorry I can't provide additional information at this time because I'm out
> of town and away from my computer/scanner until next Tuesday. Then I'd
> like to continue this thread and find out what's wrong.
> Tim Egbert