A basic guide (see post #1) to setting up ARGYLL CMS profiling on your computer

Emulator said:

I'm not entirely sure what you mean. There was no colour management in the generation of the profiles.

Click to expand...

If you select "None" as the rendering intent in Gamutvision it will show you the intrinsic response of the printer, or at least it will show a closer representation of the measured data, unmodified by the profile.

RogerB said:

The i1Pro wouldn't read it with the spacers so I tried it with no spacers. I managed to read it, albeit with lots of re-reads, and generated a profile.

Click to expand...

Just for the record, I printed my standard i1 target after scaling the patch sizes to be the same as those in Emulator's 750 patch Argyll target. Reading this target with the i1Pro using the x-Rite driver is easy - no errors whatsoever. So, maybe Argyll doesn't do such a good job with the i1Pro as the native x-Rite driver.

RogerB said:

If you select "None" as the rendering intent in Gamutvision it will show you the intrinsic response of the printer, or at least it will show a closer representation of the measured data, unmodified by the profile.

Click to expand...

Well here are the "none" rendering intent images in GamutVision. I am not sure how you interpret them.

Emulator said:

Well here are the "none" rendering intent images in GamutVision. I am not sure how you interpret them.

Click to expand...

Unfortunately(?) these look very well-behaved - no nasty kinks or dramatic changes in gradient so I'm afraid they don't tell us much in this instance. Not sure what you can do now, except generate another 750 chart using bigger patches to see if the dog-leg is still there.

I have a number of existing profiles both CM and Argyll, here are a different Argyll 750 patch and a ColorMunki 2nd optimisation 12 step B&W. It doesn't look as though it is the Xrite CM that is responsible.

Looking at some of the commercial profiles, they also show the same effect, to a greater or lesser degree.

I have edited post#1 to remove some potential ambiguities.

If anyone has created alternative .bat files for a different number of patches and is satisfied that the file works OK and wishes to share the potential profile with other users, feel free to add it to this thread.

Emulator said:

I have edited post#1 to remove some potential ambiguities.

If anyone has created alternative .bat files for a different number of patches and is satisfied that the file works OK and wishes to share the potential profile with other users, feel free to add it to this thread.

Click to expand...

First I'd like to thank pharmacist and Emulator for their work in this area that has encouraged me to take a closer look at Argyll CMS. After some experimentation I have decided that a chart on 2 A4 sheets give the best results with my i1Pro and I am now using a chart with 1058 patches. I can read this very reliably using the standard ruler for the i1. The command line I used to generate this is a straightforward edit of Emulator's CAP750.bat

targen -v -d2 -G -e8 -B8 -g128 -f1058 %pcn1%

To generate the TIFF files I again used a modification of Emulator's file. I did however specify an 8-bit file (Can't print 16-bits) at 180 ppi and forced B&W spacers. The dimensions and scaling factors give a chart that fits perfectly (for the i1) on 2 x A4 sheets.

printtarg -v -ii1 -b -a0.98 -A0.6 -t180 -m2 -p220x310 %pcn2%

For the overall batch file I took a slightly different approach in order to achieve something closer to a turnkey application and one that allows me to easily try different settings for the profile generation. The batch file lets me select the chart to use, and saves the .ti3 file so that I can re-use it without re-reading the chart. To make this work you need a folder that contains the relevant .ti2 files, another for the .ti3 files and a third for the generated profiles.

I am attaching a copy of the batch file, but make no claims for the quality of the writing! Any comments or questions welcomed. (Should say that you need to change the file extension from .txt to .bat to run it.)

ArgProf.txt

As part of my experimentation I wanted to see how good the Argyll profiles were, using an objective method. To do this I modified a 288-patch profiling target so that all the colours were within the gamut of my printer, with a good proportion of them on or near the gamut boundary. I then printed this, using Absolute Colorimetric rendering, measured the print and compared it with the original file. I used MeasureTool from ProfileMaker to do the comparison. The result for a profile generated with High quality looks like this:



I think this is pretty impressive. If you look closely you will see that each patch is split diagonally. The top left section shows the target colour and the bottom right is the actual printed colour. The comparison shows that the overall average error is only 1.23. Bearing in mind that the average person cannot see a difference of 1.0, even in near-neutral colours, this suggests that the accuracy is more than good enough.

Even more interesting, I repeated the exercise with a profile generated in Medium quality, using the same .ti3 file. Here's the result:



The average error is virtually the same, and the maximum is only very slightly worse. As before, this is more than good enough.

Unfortunately I have no way to do this exercise with CM profiles, but if anyone would like to try it I'm only too happy to assist.

I should say that as a result of what I have done so far, Argyll is now my preferred profiling software.

I looked closely, but I can't say I identified any diagonally split patches, is it possible that the uploading process has a limited colour step range? Or is it me!

Interesting experiment, the results are better than the standard in general use.

Incidentally the CAP stands for Create Argyll Profile, so I could add it to post #1 as CAP1058. Is it in its finished form or are you still working on it?