[sane-devel] Canon LiDE 90

[Pierre Willenbrock]

Hi Guillaume,

Guillaume Gastebois schrieb:
> Hello,
> 
> Why calibration is so long (~50/60s) ?

It is probably failing. Should take about 3-5 seconds. Look at the logs,
the calculated averages and calibration are dumped there.

> What are /* Start of white strip in mm (y) */ and /* Start of black mark 
> in mm (x) */ in genesys_devices.c ?

Those are configuration values for calibration steps. I don't know if
any of these are currently used or if the values are hardcoded.

I think the start-of-black-mark is used to detect the beginning of the
document area for some gl646 scanners. The start-of-white-strip was once
used in shading calibration. Currently, the shading calibration is setup
for a calibration area looking like this:

home position
+--------------------------------
!   black area
+--------------------------------
!   white area
+--------------------------------

The border between black area and white area is autodetected per pixel,
as the border is usually not straight.

You scanner seems to offer only a white area, so we will need to do
shading calibration differently. My current idea is this:
* always gather data on a white area
* for black data, reduce the led exposure time to the minimum(0x101,
those registers cannot be set to 0. per byte.).
* for white data, use the normal exposure times
I tried something like this for offset calibration, to see if there is
any difference between white area+0x101 exposure time and black
area+normal exposure time. There was no difference in the final images,
and i think the resulting calibration was the same as well.

> Regarding the log file you said :
> W  ! 0x23 ! 0x050 ! dac value rgb(offset value)
> W  ! 0x2b ! 0x028 ! pga gain rgb
> But on debug, I see that these two registers are never written.

0x23 and 0x2b are merely convenience registers. Writing to 0x23 and 0x2b
is equivalent to a write to each of 0x20-0x22 and 0x28-0x2a. For
cis-sensors, there is only one channel used, so we could get away with
only two registers writes(for the correct channel or 0x23/0x2b), but
this won't work for ccd-sensors.

> Another thing : when scaning in color the leds are blue ????

I'd expect a shade of white, perhaps blueish. my scanner does a
magentaish white. You may also see the single colors when quickly moving
your eyes relatively to the scanner.

Regards,
  Pierre