[sane-devel] Genesys backend update to Dec 06, cvs

[Pierre Willenbrock]

Gerald Murray schrieb:
> Hi all,
> 
> # These tests results are using sane-backends-20051206-cvs:
> 
> GL841 - Canoscan lide35:
> 
> This has changed for the worse since Nov 24.
> I have previously made scans at 150, 300, 600, in color.
> Now the 'image' is no longer recognizable.
> 

Not recognizable means color noise?
Last time i tested my Canoscan lide35 worked for me.

Please try scanimage, and if the problems persist, send a log with
SANE_DEBUG_GENESYS=255 and SANE_DEBUG_GENESYS_GL841=255. When using
these environment variables, you should get a file named
black_white_shading.pnm in your working directory. As both the log and
the image are pretty large, please compress them, and send them directly
to me, not to the list. Better put them on some webspace, where i can
download them. I can deal with e-mail larger than 10mb, but i prefer
downloading it somewhere.

> color, 75dpi, (scan fails, obtaining 0 feed steps, then stops)
> 
> color, 150dpi (unrecognizable image created)
> warmup(?) is very odd.
> move head, wink lite, park head <-repeated several times.
> Usually a warmup would keep the light on, move to strip, scan, rescan if
> necessary.  I am not sure where this is coming from.
> flag GENESYS_FLAG_SKIP_WARMUP is still there.
> 

The LiDE scanners don't need a warmup. What you are seeing is the
calibration process. It should look like this:

 - move home(currently missing, but not critical)
   (head is now on black strip)
 - do multiple single line scans to calibrate analog offset
 - move to white strip
 - do multiple single line scans to calibrate analog gain
 - move home

 - move to white strip
 - do multiple single line scans to calibrate exposure time per color

 - move home
 - do multiple single line scans to calibrate analog offset
 - move to white strip
 - do multiple single line scans to calibrate analog gain
 - move home

 - scan full calibration area for shading calibration
 - move home

Calibration should theoretically only be done once, but it is far easier
to do it before every scan.

> color, 300dpi -  SAME as 150dpi
> 
> color 1200dpi - Very slow, backtracks 195 times on a 107mm scan length.
> image size was 100 megabytes, for about 1/3 of the available scan area.
> (Note: using usb 1.1 fullspeed, not hi-speed)
> 

What do you expect? usb 1.1 fullspeed means 11mbit/s or about 1mb/s. Not
counting the overhead from checking the scanners state. In 1200dpi mode
the scanner potentially delivers about 5 million pixel components per
second. I am measuring about 2.5 million pixel components per second.
The scanner buffers data, but needs to do backtracking if the buffer
runs full.

The image size is normal. Your image contains about 50 million pixels,
making about 150 megabytes.

Regards,
  Pierre