On 02/22/2010 06:43 AM, stef wrote:
> Le lundi 22 f?vrier 2010 02:09:14 Chris Berry, vous avez ?crit :
>    
>> Hey Stef,
>>
>> I think I have isolated the gamma tables from the logs but there is a
>> significant difference; the gamma tables all seem to be sent at once to
>> the scanner. Regs 0x5b and 0x5c are both set to 0x00 and 1536 bytes are
>> sent to the scanner. I have attached the log file appearances of the
>> gamma tables to this email, there are five occurrences in my log file so
>> all sets are here. I have split them up myself into three blocks but in
>> the original snoop they all appear together.
>>
>> Will I need to modify the code base to account for all the gamma tables
>> being sent at once?
>>
>> Thanks
>>
>> Chris
>>
>>      
>       Hello,
>
>       no need to modify the backend. Both way of writing data give the same 
> result.
> Since all 3 curves are the same, you can pick one (from the '1' write, others
> are simple ramps) and turn it into a list of 256 16 bits words. You can then
> import it in a spreadsheet.
>       You may use this kind of awk script on a 1 line file with all the hex 
> values:
> {
>       for(i=1;i<NF;i+=2)
>       {
>               print strtonum($i)+256*strtonum($(i+1))
>       }
> }
>
> Regards,
>       Stef
>
>    
Hey Stef,

Thanks for the awk script, helped a lot. I put the values into a 
spreadsheet and got gamma values all around 1.0 so I didnt really need 
to alter them at all, but it gives me something to talk about in my 
report :). I have tweaked the gain and offset (offset is largely ignored 
for AD type frontend it seems so I added a similar AD_OFFSET_CALIB 
function to gl841) and now the picture seems right with the original 
shading issues.

So now I need to alter the shading as the shading files are still 
showing really odd things. How do I adjust these? I imagine I need to 
find the calibration area in some way but im not sure how I do this?

After the shading is corrected I think I should be ready to test, is 
there any formal testing procedure with SANE? There are a couple of 
people who are awaiting this scanner backend and have agreed to help 
testing so I will get it to them in some way. Im also going to sort my 
patch out and get rid of any whitespace or other rubbish ;)

Cheers

Chris

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.


Reply via email to