[sane-devel] Backend for plustek Opticbook 3600

[Pierre Willenbrock]

stef schrieb:
> Le mardi 2 mars 2010 16:40:21 Chris Berry, vous avez écrit :
>> Hey guys, I resolved the issue I was having with the dark_shading
>> calibration. The lamp was turning off fine but the | REG03_LAMPPWR in
>> gl842_begin_scan was turning the light back on for the scan :/ . Having
>> resolved this there is still a black line down the center of the page.
>> The black_shading.pnm file is now completely black and the
>> white_shading.pnm file is mostly white with some dark edges.
>>
>> Any thoughts on how I sort this out now? Is it related to the number of
>> sensor pixels or the shading area or something?
>>
>> Chris
>>
> 	Hello,
> 
> 	the black patterns are often used to detect the exact geometry of the 
> scanner. What is reliable is these black/white areas, while the casing may 
> vary. For instance the HP2400/HP2300 and HP3670 have a black corner which is 
> aligned on the exact vertical margin of the visible scan area.	So the dark 
> edges may be harmless if they are outside the effective scan area used for 
> final scan.
> 	For the big black vertical line (which is a third of the width of the 
> picture), it seems that shading coefficients sent aren't send in the correct 
> format. It's hard to tell without the code, but the results looks like 
> 'planar' format (for CIS) was used when 'chunky' format (for CCD) was needed. 
> Surely an issue in the genesys_send_shading_coefficient() where 
> compute_coefficient() is for chunky, compute_planar_coefficient() is for 
> planar and the CCD_CANONLIDE35 case looks for the planar case to me. Pierre 
> may confirm this last point.

yes, CCD_CANONLIDE35 should have been named CIS_CANONLIDE35, and that
section is for planar shading. That whole case will go away sometime in
the future and be replaced by a call to compute_planar_coefficient.

Regards,
  Pierre