Hello,
Finally, I got some time to play with the backend and my scanner.
On Sun, 9 Jan 2000, Bernd Schroeder wrote:
>
> On Tue, Jan 04, 2000 at 09:27:47AM +0100, Levente NOVAK wrote:
> >
> > - when backend-calibration is on, there are still crashes. I discovered
> > that these always occur if there was no preview scan with backend-
> > calibration on, while (most of the time) there is no crash when I do
> > a preview scan with calib on. I did not test the previous version
> > (0.8), but I suppose it had to have the same behaviour.
>
> Sometimes the device indicates that no calibration is necessary, which is
> probably the reason why it does not always crash and is independent of
> whether it is a preview or normal scan.
I tested the new SANE snapshot of 23th of January. There are still
crashes, but these may be prevented most of the time by doing a preview
scan in noncalibrated mode before the preview in calibrated mode.
I do not know if it is because the device indicates that no calibration is
necessary after the first preview scan, as the colors are quite different
in calibrated and noncalibrated mode (approximately OK for calibrated
one).
> > - now something surprising: images scanned with calibration are
> > worse than those scanned without, as they have a strong blue tint!
> > Effectively, scanning a white sheet of paper and looking at the
> > histograms shows that while R and G histograms are very close to
> > each other, the blue histogram is markedly shifted towards brighter
> > intensities. Without backend-calibration, the left-to-right order
> > of the three histograms is different, but clearly there is no such
> > a shift in respect to blue channel. I haven't had time to retest
> > with previous backend release, but do not remember having a
> > similar result with it. Did you change something inbetween?
With the new SANE snapshot, the colors are the same for v.0.8 and
v.pre0.9. There is however a bluish tint in certain conditions with the
calibrated backends after a certain (not thoroughly investigated) sequence
of calibrated- and noncalibrated scans, following a crash. It is maybe
due to an "overcalibration" by the backend, as it does not remember having
already compensated and does a color compensation a second time (BTW the
noncalibrated scans are a little bit reddish). Anyway, it would be more
easy and reproducible to test if there was no hangup at all.
> > Finally, I am very interested to know how did you trace the commands sent
> > by the twain driver with scantest.exe? If I could do the same, maybe it
> > would be easier to find what is the trick of the driver under Windows.
> > Unfortunately, I was unable to get my scanner recognised with recent
> > releases of WINE, while the version shipped with Debian Slink (2.1) did
> > the job, but then scanimage.exe crashed right after. Maybe it is important
> > to mention that I used Windows 3.1 (and the 16 bit version of the drivers
>
> No, I cannot tell you how get it working under wine. I just tried it,
> but the 16bit ASPI layer does not work for me. Normally it is sufficient
> to just call 'wine scantest' (plus tracing enabled).
>
That was exactly my problem. With never Wine releases I couldn't get the
ASPI interface recognise my scanner. With the old version shipped with
Debian 2.1, the recognition was done but then scantest.exe hung.
Levente
-- 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 : Fri Jan 28 2000 - 01:02:49 PST