[argyllcms] Re: how to add patch around bw axis

>> I've note, that addition of R=G=B patches isn't meant that there  
>> will be gray patches at printer output, because of printer may be  
>> not linearized and gray-balanced. The optimum way is the selection  
>> of wanted gray points in Lab, then icclu conversion of this points  
>> to RGB data by preliminary profile. Of course, RGB data may be  
>> converted to XYZ by icclu too.
>> As I wrote yesterday:
>> 1. Computing LAB data array in table processor
>> 2. Converting LAB array to RGB: icclu -fb -ir -s100 <LAB.txt  
>> >RGB.txt PreProf.icc
>> 3. Computing XYZ data for *.ti1: icclu -ff -ia -s100 <RGB.txt >XYZ.txt
>> 4. Combining patch numbers, RGB and XYZ data in one table by table  
>> processor
>> 5. Inserting this table into *.ti1 file.
>>
> If I've understand correctly I have to start the above procedure with  
> a preliminary profile... It's ok if I make a preliminary profile with  
> only 441 patches or I have to do this with the 882 patches target?

Yes, 441 patch will be ok. I do 360-patch preliminary profile on a half of A4 
sheet and it's works well. Then I add this 360 patches measured data to the 
second (main) *.ti3 file. So I have in result more patches for colprof from 
both *.ti3 - preliminary and main.

>> You can't just add gray RGB by -g key because it changes OFP-spread  
>> in the *.ti1 file, but You can generate separate file only with  
>> grays and combine two ti1, one OFPS, one GRAYS into one big *.ti1  
>> with "standard" OFPS and additional grays.
>> I'd not recommend You to do regular gray spread across gray axis as  
>> in JPEGs You shown. Randomized distribution may be more reasonable  
>> for profiling (ask Graeme here).

> Ok I trust you!

Daniele, it's IMHO. I'm a human and I can mistake. :)

> Thanks a lot! it is vary clear! Is it ok if I use a 7dE radius or is  
> it better something else?

I recommend You to check dE in preliminary profile by profcheck and see the dE 
error of about gray patches. I think the spreading radius may be 1.3-2 size of 
max error. After 2 profiles done You understood your printer behavior. I think, 
there is no need to print hundreds of grays, I'd like to print more colors when 
grays are just ok. But if You print only B&W it may be reasonable. If You use 
not standard D50 but known illuminant for viewing prints (for example at art 
exhibition), I recommend You to save spectral data while measuring patches and 
to use -i key in colprof with appropriate illuminant while profiling for the 
best result.

> If I have understand  I have to write down all the neutral gray  
> patches and the random "semi-gray" patches in a unique LAB.txt and  
> after that I have to start the 2-5 procedure
> is it correct?

Yes, I do so. But You free do it any way. I successfully use utility features 
to input whole file after < sign and to output files after > sign. It's quite 
convenient. Note, that for any conversions by icclu and xicclu You must supply 
only color or device data without patch enumeration. The numbers must have 
decimal dot (not comma) and separated by one space each from other in line. 
Check the XYZ data scale. It is sometimes 0-1 not 0-100.

Best regards,
Nikolay.

Other related posts: