YAAW - Yet Another Argyll Wrapper

[riclin]

Did some more testing. It's only the -S switch which gives the differences in chartread output/behavior. I measured the printed target in YAAW and in ChromIQ and both results are nearly identical. The helpful meant warnings from YAAW brought me on the wrong track. Sorry about that. Now I can concentrate on the quality of the profile.

As I mentioned, the -S switch just reduces warning and error messages - it shouldn't actually impact the quality of the read per se. You might want to have a play with the most current version to which I have some default settings for i1 and 3p instruments - they generate useful looking targets although the proportions look a little unusual - note that they have not been tested IRL. Also, based on your earlier observations, the -H chartread switch now defaults to off for anything other than the CM device - let me know whether it being off changes the read experience:

https://drive.google.com/file/d/1GgbzhyTyIdKH28yU0LIH8G5pB_Jv1JIs/view?usp=sharing

Pick the device at the top right hand corner then choose the patch density at "Patches" - if selecting from the dropdown options at the right edge of the Patches field, then Paper size is autofilled.

r.

 

[rjanowsky]

Did some testing with the INSTRUMENT_LAYOUTS. Checked out a set with 484 patches on one A4 page. Only change under "Additional printtarg Args" was: -a 0.95 to scale the size of patch fields and spacers a little bit down. The resulting print could be easily measured with the i1pro device. With this setting as base one can do a two (968) or three (1452 )page layout.

 

[rjanowsky]

As I mentioned, the -S switch just reduces warning and error messages - it shouldn't actually impact the quality of the read per se. You might want to have a play with the most current version to which I have some default settings for i1 and 3p instruments - they generate useful looking targets although the proportions look a little unusual - note that they have not been tested IRL. Also, based on your earlier observations, the -H chartread switch now defaults to off for anything other than the CM device - let me know whether it being off changes the read experience:

Finally I found the error:

[2026-07-11 16:36:01] Running: targen -v -d2 -f 484 -G -e 4 -B 4 -g 28 -n 0 -N 0.5 -V 1.0 -l '' -c none EPS_Test_OEM-Argyll_i1_484
Running targen with the -c none switch, results in a complete other whitepoint in the .TI12 files:

APPROX_WHITE_POINT "128.335403 100.000000 34.800696"

[2026-07-11 17:33:09] Running: targen -v -d2 -f 484 -G -e 4 -B 4 -g 28 -n 0 -N 0.5 -V 1.0 -l '' EPS_Test_OEM-Argyll_i1_484
Running targen with an empty -c switch results in:

APPROX_WHITE_POINT "95.106486 100.000000 108.844025"

In the first case all values are massively off (about 100∆E and more). Lot's of warnings in YAAW. In the second case everything went smooth. The resulting profile is excellent.

 

[riclin]

Finally I found the error:

[2026-07-11 16:36:01] Running: targen -v -d2 -f 484 -G -e 4 -B 4 -g 28 -n 0 -N 0.5 -V 1.0 -l '' -c none EPS_Test_OEM-Argyll_i1_484
Running targen with the -c none switch, results in a complete other whitepoint in the .TI12 files:

APPROX_WHITE_POINT "128.335403 100.000000 34.800696"

[2026-07-11 17:33:09] Running: targen -v -d2 -f 484 -G -e 4 -B 4 -g 28 -n 0 -N 0.5 -V 1.0 -l '' EPS_Test_OEM-Argyll_i1_484
Running targen with an empty -c switch results in:

APPROX_WHITE_POINT "95.106486 100.000000 108.844025"

In the first case all values are massively off (about 100∆E and more). Lot's of warnings in YAAW. In the second case everything went smooth. The resulting profile is excellent.

Very interesting! Good find Ron. On reflection, I must admit that I found issues that I thought may have been related to the "-c none" switch too and in earlier versions of yaaw inserted a test that masked out the -c switch if no preconditioning profile had been selected. In a more recent code cleanup, I removed the test and simply used the "-c none" argument which is what the targen documentation states as correct. From your experience, it appears that that mask did in fact help. I'll dig out the old code and reinsert it - the additional code was nominal but I was trying to clean out any redundant code that had crept in during testing.

Thanks for that. It's interesting that the "-c none" code worked OK for me but not for you. Possibly a subtle difference between how Argyll treats the CM vs the i1? Or perhaps a difference between ArgyllCMS versions? - what version are you running? I am using Graeme Gills most current distribution - v3.5.0. Please let me know what version you have - and if not the latest and you feel inclined, see whether the problem remains with ArgyllCMS 3.5.0. (https://www.argyllcms.com/)

r/

 

[riclin]

OK - the error was mine - I had misunderstood the difference between how targen processes no "-c" field vs "-c none". Having gone back to RTFM:

• with no -c option, targen uses its default device-space model;
• -c <profile> uses an ICC/MPP profile for preconditioning;
• -c none explicitly turns off the default preconditioning model.

So by defaulting to "-c none", YAAW was completely disabling targen's internal preconditioning behaviour. targen must write approximate CIE values into the .ti1 file. Without either a real profile or the normal default model, those estimates can be extremely crude. In Ron’s case, the calculated white point (128.335403 100.000000 34.800696) was wildly unlike the real paper white. Hence chartread compared valid measurements against nonsensical expected values and emitted enormous “unexpected value” warnings.

The updated code below now leaves the default preconditioning field empty, and "-c" is now treated as it should be. In the unlikely event that it is necessary to explicitly turn off the preconditioning defaults, the "-c none" argument is now accepted instead of being masked out as it was in early versions of YAAW.

https://drive.google.com/file/d/1SyKR2AX6DPTGv_qsgF-AsTYqI-HSJqK5/view?usp=drive_link

Thanks for the very valuable report Ron. And sorry about that folks - mea culpa.

 

[rjanowsky]

Thanks for the very valuable report Ron. And sorry about that folks - mea culpa.

Searching for the error was a very interesting experience. It needed some time to get a clue what could be the reason for that unexpected behavior. Glad I could help.

Regarding the PLACEHOLDER tables: Using 484, 968 or 1452 instead of 440, 880 or 1320 give roundabout 10 percent more measurements (two rows per page) and helps getting a better profile. One must have only in mind that an extra printtarg argument of -a 0.95 is needed.

 

[rjanowsky]

The updated code below now leaves the default preconditioning field empty, and "-c" is now treated as it should be. In the unlikely event that it is necessary to explicitly turn off the preconditioning defaults, the "-c none" argument is now accepted instead of being masked out as it was in early versions of YAAW.

Did just a run with the latest version. Everything is fine now.

Starting profile:
[2026-07-11 18:19:35] Profile check complete, peak err = 1.511695, avg err = 0.425881, RMS = 0.497089
[2026-07-11 18:19:35]
[2026-07-11 18:19:35] Validating profile with profcheck...
[2026-07-11 18:19:35] Running: profcheck -k EPS_Test_OEM-Argyll_i1_484.ti3 EPS_Test_OEM-Argyll_i1_484.icc
[2026-07-11 18:19:35] Profile check complete, errors(CIEDE2000): max. = 0.900768, avg. = 0.234340, RMS = 0.274559
[2026-07-11 18:19:35]

Preconditioned profile:
[2026-07-12 13:31:12] Profile check complete, peak err = 1.467402, avg err = 0.397021, RMS = 0.465743
[2026-07-12 13:31:12]
[2026-07-12 13:31:12] Validating profile with profcheck...
[2026-07-12 13:31:12] Running: profcheck -k EPS_ILFGSP_OEM-Argyll_i1_484.ti3 EPS_ILFGSP_OEM-Argyll_i1_484.icc
[2026-07-12 13:31:12] Profile check complete, errors(CIEDE2000): max. = 0.701889, avg. = 0.215310, RMS = 0.251094
[2026-07-12 13:31:12]

The second preconditioned profile is just a small bit better in the ∆E values. Overall I would say the profiles are really excellent and with the help of YAAW it was no big effort to produce them.

Nice work!

 

[riclin]

Searching for the error was a very interesting experience. It needed some time to get a clue what could be the reason for that unexpected behavior. Glad I could help.

Regarding the PLACEHOLDER tables: Using 484, 968 or 1452 instead of 440, 880 or 1320 give roundabout 10 percent more measurements (two rows per page) and helps getting a better profile. One must have only in mind that an extra printtarg argument of -a 0.95 is needed.

Thanks for the suggestions regarding the higher density targets - very useful information. Don't be offended if I don't incorporate them per se into the current release - an unstated part of the "ethos" behind yaaw is that it provide a standard working framework (recent glitches notwithstanding) whilst also being sufficiently transparent and flexible that it might also facilitate further experimentation and fine-tuning by anyone wishing to do so. Your example is a perfect example of that - but since it also requires scaling changes, it may open a can of worms to the unwary. My inclination is to not introduce non-standard scaling options into the current mix, and to instead mention that potential in the accompanying yaaw.pdf to encourage experimentation by anyone wishing to investigate it. That said, that's just my inclination and I'm willing to be converted if you think otherwise.

Thanks again Ron for your interest and valuable feedback.

r/

 

[riclin]

Did just a run with the latest version. Everything is fine now.

Starting profile:
[2026-07-11 18:19:35] Profile check complete, peak err = 1.511695, avg err = 0.425881, RMS = 0.497089
[2026-07-11 18:19:35]
[2026-07-11 18:19:35] Validating profile with profcheck...
[2026-07-11 18:19:35] Running: profcheck -k EPS_Test_OEM-Argyll_i1_484.ti3 EPS_Test_OEM-Argyll_i1_484.icc
[2026-07-11 18:19:35] Profile check complete, errors(CIEDE2000): max. = 0.900768, avg. = 0.234340, RMS = 0.274559
[2026-07-11 18:19:35]

Preconditioned profile:
[2026-07-12 13:31:12] Profile check complete, peak err = 1.467402, avg err = 0.397021, RMS = 0.465743
[2026-07-12 13:31:12]
[2026-07-12 13:31:12] Validating profile with profcheck...
[2026-07-12 13:31:12] Running: profcheck -k EPS_ILFGSP_OEM-Argyll_i1_484.ti3 EPS_ILFGSP_OEM-Argyll_i1_484.icc
[2026-07-12 13:31:12] Profile check complete, errors(CIEDE2000): max. = 0.701889, avg. = 0.215310, RMS = 0.251094
[2026-07-12 13:31:12]

The second preconditioned profile is just a small bit better in the ∆E values. Overall I would say the profiles are really excellent and with the help of YAAW it was no big effort to produce them.

Nice work!

Thanks Ron. Pleased to read your comments. The profcheck error improvements with the preconditioned profile are actually quite sigfnificant. Yaaw might not be beautiful, but it's simple to use and seems to work quite well.

r/