[sane-devel] Canon LiDE 90
Pierre Willenbrock
pierre at pirsoft.dnsalias.org
Tue Feb 19 21:09:38 UTC 2008
Guillaume Gastebois schrieb:
> Hello,
>
> I modified lines 4596 and 4712 and reenable SCAN_FLAG_DISABLE_LAMP flag.
> Result can be found on : http://ggastebois.free.fr/lide90_snoop/19_test1.tar
Okay, results look good so far:
[genesys_gl841] gl841_offset_calibration: first set: 191/683,191/482,191/76
but there must be a little bug in the code:
[genesys_gl841] gl841_offset_calibration: second set:
0/-1080773208,8/-1212144018,-1080773236/134721688
this very much looks like the variables for the second set are getting
overwritten/not initialized. Please try to find the problem(misplaced
brackets perhaps? copy+pasto when calculating the second set?), or send
the source.
Regards,
Pierre
> Regards
> Guillaume
>
> Pierre Willenbrock a écrit :
>> Guillaume Gastebois schrieb:
>>> Hello,
>>>
>>> OK, I'll try this tonight. What is the best : WITH or WITHOUT
>>> SCAN_FLAG_DISABLE_LAMP ?
>> Not using SCAN_FLAG_DISABLE_LAMP is a bit counter productive when trying
>> to get black levels on a white-only calibration area.
>>
>> Regards,
>> Pierre
>>
>>> Regards
>>> Guillaume
>>>
>>> Selon Pierre Willenbrock <pierre at pirsoft.dnsalias.org>:
>>>
>>>> Guillaume Gastebois schrieb:
>>>>> Hello,
>>>>>
>>>>> I made two tests today :
>>>>>
>>>>> test 1 : too bright/too dard = 10/65525 WITH flag :
>>>>> SCAN_FLAG_DISABLE_LAMP. Result can bee found on :
>>>>> http://ggastebois.free.fr/lide90_snoop/18_test1.tar
>>>>>
>>>>> test 2 : too bright/too dard = 10/65525 WITHOUT flag :
>>>>> SCAN_FLAG_DISABLE_LAMP. Result can bee found on :
>>>>> http://ggastebois.free.fr/lide90_snoop/18_test2.tar
>>>>>
>>>> Not what i expected, although the debug images are looking good.
>>>>
>>>> Please try to change the first pixel used for minimum calculation to 200
>>>> at about lines 4596 and 4712:
>>>> - for (i = 0; i < num_pixels; i++)
>>>> + for (i = 150; i < num_pixels; i++)
>>>> {
>>>> if (dev->model->is_cis)
>>>> val =
>>>>
>>>>
>>>> Regards,
>>>> Pierre
>>>>
>>>
>>
>>
>
More information about the sane-devel
mailing list