2015-02-01 by richard@...
My response from a few days ago didn't show up, and am just now getting back to this.
You need to relinearize every time you change the settings the ink limits, gray values, or other gray shadows/highlights/gamma options. You need to do these things in the following steps:
Change some setting(s)
Delete previous linearization settings from the profile you are editing . You can not linearize on top of a previously linearized profile because there can only be one 'Linearize="97.4 ... 16.35" ' line (well, it turn out that you really can do it, but it takes a lot of work... I will explain later).
Save the edited profile
Print 21 step target with the new profile
Measure the target (lightest to darkest-should be left to right), and drop the measurement file onto the LinearizeData script
If you did not get the "values not in order" error Plug the resulting Linearize="97.4 ... 16.35" back into the profile. (or paste into the Windows gui) If you still get the error, then see below.
Resave and reinstall the profile and make a print to check that the linearization it worked.
One reason you are getting the "Lab values not constantly increasing" or "not in order" is might be because the ink limit settings were changed, or the cross over points are not accurate. If you do keep getting those errors I would bet there is something wrong with your cross over points. On the other hand, it could be problems with errors in the measurements themselves. These errors can be averaged out by using the 21x4 target. I built a custom reference file that allows me to average 6 measurements from a single 21 step target to save and paper/ink rather than printing multiples of the same patch values. The built in 21x4-random target might be just as good. . .
Accurate Cross Over Points:
The cross overs determine where the next ink starts and trails off. If they are set to the wrong value then one ink will be starting and dropping off before and after it really should. That is usually where you will get spikes and dips in the tonal scale and in your measurements, which will cause problems linearizing the profile. This is the one reason why I am constantly saying "you don't mess with the ink limits after you create a working profile or after you have set your cross over points". When you are creating profiles with 4 or more overlapping inks the cross over point becomes more critical because they are jumping in and out faster than with fewer inks.
I actually don't know if what follows below is possible on a PC, but is one huge benefit of being able to see what is happening in the Terminal on the Mac. To do this you will need to have "GRAPH_CURVE=YES" in the top of the profile/descriptor file.
If you don't have reversals, but the Lab_L values are too close together and you get the "not constantly increasing" error when you try to install the profile you can rough in your gray curve to get close to linear by using the input/output correction curve the QTR linearization script creates for linearizing the .quad values. This correction curve is created before you get the "not constantly increasing error"so you can use it whenever you have a problem with installing/linearizing a profile.
Go the terminal window after running the install script and look for the profile that is causing the problem. You can find this by looking for "Creating Curve curvename" ( It should be in alphabetical order) Find the lines for "LINEARIZE_CURVE = 0,0 3.43,5 7.4,10 ... or whatever your actual values are ". Copy and paste those to numbers to the GRAY_CURVE= line in your profile/descriptor file — You will need to put double quotes before and after the whole line and change the commas to semicolons between the input/output numbers (I just use find/replace in the text editor). You can even delete every other input/output pair to give you 11 10% steps that gets the profile close and then fine tune that with the 5% 21-step target.
I don't use this method too much now, but it works well to get a profile close to linear before actual final linearization. Since you are going to relinearize the roughed in curve/profile you need to delete the original linearization values, save and reinstall the profile, and reprint the 21step target with the new profile. Then measure it again and plug the Lab_L measurements back into the linearize="..." line. If that doesn't work then I would start from scratch measuring the ink limit page and resetting your cross overs...
Richard Boutwell
---In QuadtoneRIP@yahoogroups.com, <jacques.caron@...> wrote :
Me again
Do I have to read the patches from black-to-white or the other way around?
Printed the 21 step scale, read it with Spyderprint and exported it to a text file. Made 8 sets of measurements from light to dark.
Any text file dropped on "QTR-Linearize data" give me the same answer : 'The Lab values are not in order.
Cannot be linearized."
Don,t know if I like the way it goes…
Hi again
small question: do I need to relinearize if I'm just changing the tray shadow value?
I did increase it and I dropped the whole profile on to the QTR_Linearize date, then I changed the values in the linearize line. After that I made the quad file.
Is it the way to do it?
Thanks again
You don't want to mess with the ink limits, because that will throw off the gray values (cross over points) settings for each of the lighter dilutions and can cause banding in smooth gradients or even bad tonal reversals.
To give the shadows more separation you can try increasing the GRAY_SHADOW=" " default is something like 4-8 but I've been using a shadow setting of 15-18 for this inkset)
You can also try increasing the GRAY_GAMMA=" " start with an increase to 1.2 to see if that helps,
You will need to then relinearize the profile.
Richard Boutwell