[sane-devel] Backend for plustek Opticbook 3600

stef stef.dev at free.fr
Sat Mar 20 06:44:11 UTC 2010


Le jeudi 18 mars 2010 22:14:26 Chris Berry, vous avez écrit :
> > 	Hello,
> >
> > 	in order to properly review the result of your hard work, we need the
> > changes you done as patches. Either against the version you started to
> > modify, or as a git patchset against latest version, which is the best
> > solution. On your site, I only found the full sources, may you have a
> > look into it ?
> >
> > Regards,
> > 	Stef
> 
> Ok Stef I will hopefully have a patch to you by the end of the weekend.
> Looking through the git changes and such I really feel the best thing
> for me to do is get the latest version, start over adding things back in
> from my current version and then I can have one commit of just my
> changes and not any other rubbish (white space etc).
> 
> As for the quality of the scans is there anything you can advise on
> maybe what I need to do to improve it? I have been a ill this week and
> had other things to work on but after tomorrow i'm back on this project
> full time (there is about a week left until I hand in my written report
> so hopefully I can get some testing done next week to include in the
> report).
> 
> Thanks
> 
> Chris
> 
	Hello,

	for the image quality, I think you could concentrate on on full sensor 
resolution and 'half ccd' resolution. In these modes AVEENB is of no influence 
since no averaging or deletion occurs. Then you can adjust the offset to be 
used for shading data. To find the correct offset some trial and error is 
needed. You'll have to change it's value until bright vertical lines match the 
closest dark vertical line in scans (you could scan only the white cover of 
the scanner, lines are more visible). You may try first the offset values from 
other GL841 scanners. 
	Once done, you may choose to keep averaging, but in this case you'll have to 
tune compute_coefficients to do the same averaging than the planar method used 
for the CCD_CANONLIDE35.

Regards,
	Stef



More information about the sane-devel mailing list