Printer Profile Validation: M1 vs. M2 and ΔE2000 Evaluation

[know]

The reason is that i1Profiler can correct exactly the influence of standard UV-light on OBA when scanning (i1Pro1: 2-pass method, and i1Pro3 in 1-pass), and the corresponding readings are written as separate data with the extension MO, M1 or M2 to the file. This way you can generate several profiles under certain desired environments.

I did not know that.

 

[pharmacist]

OK, I just have succesfully scanned a ChromIQ target with i1Profiler using the dual scan method and it works like a charm.

I have moved the letters of the row in such a way it is not scanned by the spectro (using the standard ChromIQ setting a big black letter is on the headway of the scanning eye). I can now safely start from the scanning jig white over the white paper edge and end on the other side of the scanning jig white without any errors:

The 3 page target (13x18 cm) with 702 patches is considered a one long page in i1Profiler.

Now I am saving the scanned data in mxf format for ChromIQ to generate the target:

I have generated a target too with i1profiler and with ChromIQ using the same data from the mxf-file and it is interested there is a slight difference (but negligible) in the white point, which is interesting, but to be expected because of the slight difference how those 2 engines handle perceptual intent.

Here a softproof (relative colorimetric with BP=on), first picture with the i1Profiler generated target:

The softproof using the ChromIQ generated target (using the standard settings):

 

[itsab1989]

Sounds great! Just out of curiosity: would you consider either Argyll or i1Profiler strictly better for building the profile out of the measurements or are both good - just in their own ways?

 

[jctay]

RGB Values in Profiling Charts

i1Profiler uses 8-bit integer RGB values for profiling. When a patch set is loaded into i1Profiler, any fractional parts of the RGB values are truncated, rather than rounded, to the nearest integer.

Examples:
42.5001 → 42
84.9999 → 84


CP1500-600p-i1profiler.txt

CGATS.5
ORIGINATOR "ChromIQ"
DESCRIPTOR "CP1500-600p"
CREATED "October 20, 2026"
INSTRUMENTATION "Not specified"
MEASUREMENT_SOURCE "Not specified"
PRINT_CONDITIONS "Not specified"
KEYWORD "SampleID"
NUMBER_OF_FIELDS 4
BEGIN_DATA_FORMAT
SampleID RGB_R RGB_G RGB_B
END_DATA_FORMAT

NUMBER_OF_SETS 600
BEGIN_DATA
1 0.0000 0.0000 0.0000
2 0.0000 0.0000 42.5001
3 0.0000 0.0000 84.9999
4 0.0000 0.0000 127.5000
5 0.0000 0.0000 170.0001
6 0.0000 0.0000 212.4999
7 0.0000 0.0000 255.0000
8 0.0000 42.5001 0.0000
9 0.0000 42.5001 42.5001
.....

 

[jctay]

OK, I just have succesfully scanned a ChromIQ target with i1Profiler using the dual scan method and it works like charm.

View attachment 19700

View attachment 19701

I have moved the letters of the row in such a way it is not scanned by the spectro (using the standard ChromIQ setting a big black letter is on the headway of the scanning eye). I can now safely start from the scanning jig white over the white paper edge and end on the other side of the scanning jig white without any errors:

View attachment 19702

The 3 page target (13x18 cm) with 702 patches is considered a one long page in i1Profiler.

View attachment 19703

Now I am saving the scanned data in mxf format for ChromIQ to generate the target:

View attachment 19704
I have generated a target too with i1profiler and with ChromIQ using the same data from the mxf-file and it is interested there is a slight difference (but negligible) in the white point, which is interesting, but to be expected because of the slight difference how those 2 engines handle perceptual intent.

View attachment 19705


Here a softproof (relative colorimetric with BP=on), first picture with the i1Profiler generated target:

View attachment 19706

The softproof using the ChromIQ generated target (using the standard settings):

View attachment 19707

Take a breather!

Potential error: Digitization inaccuracies resulting from the alignment of the two targets
Data points: 100x60 = 6000
Color Space: sRGB

 

[jctay]

i1Profiler Patch Reading & Loading Sequences for Chrom2i1 Converter

Kindly note that the details below are observational and may be subject to minor inaccuracies.

1. Patch Reading Sequence (ArgyllCMS Source Patch Set)

i1Profiler reads patches from the source patch set in row-major order:

Begin with Row 1 and read patches from left to right.
Proceed to Row 2 and repeat the same left-to-right reading sequence.
Continue row by row until all patches in the source patch set have been read.

2. Patch Loading Sequence (Target Layout)

Patches are loaded into the target layout according to the calculated matrix configuration and the resulting page arrangement. The loading sequence follows a column-first, page-by-page order:

Begin with Column 1 on Page 1.
Continue with Column 1 on each subsequent page until all pages in that column have been processed.
Proceed to Column 2 on Page 1 and repeat the same page-by-page sequence.
Continue column by column until all columns and patches have been loaded.

[ ref1 ref2 ]

 

[pharmacist]

Sounds great! Just out of curiosity: would you consider either Argyll or i1Profiler strictly better for building the profile out of the measurements or are both good - just in their own ways?

Actually from my experiences is that ArgyllCMS generated profiles are a bit more neutral in the greys than i1Profiler, despite using than neutralize grays at 50 % in the i1Profiler setting, probably of using a more sophisticated engine in ArgyllCMS. In the old versions of ArgyllCMS there was strange bug that when using relative colorimetric with BPC=on the darkest black is ignored causing pure black nog being printed (but a dark grey, as the engine want to adapt the slope from pure white to pure black according to the measured values and often pure black is not measured as such, thus causing this strange clipping). One can check this with the verification process in ChromIQ and use different intents with/without BPC=on to see if this affects the pure black patch value.

If you see carefully in the softproofing is the slight yellowish hue in the ArgyllCMS profile to compensate for the OBA of this particular glossy TPI Paper (probably built-in the colprof standard parameter setting in ChromIQ as it uses a similulation not a real measurement).

 

[df11]

На самом деле, судя по моему опыту, профили, сгенерированные ArgyllCMS, немного нейтральнее в оттенках серого, чем профили i1Profiler, несмотря на использование нейтрализации серых оттенков на уровне 50 % в настройках i1Profiler, вероятно, из-за использования более совершенного движка в ArgyllCMS. В старых версиях ArgyllCMS существовала странная ошибка: при использовании относительной колориметрии с BPC=on игнорировался самый тёмный чёрный цвет, в результате чего печатался не чистый чёрный (а тёмно-серый, поскольку движок пытается адаптировать наклон от чистого белого к чистому чёрному в соответствии с измеренными значениями, а чистый чёрный часто не измеряется как таковой, что и вызывает это странное обрезку). Это можно проверить с помощью процесса верификации в ChromIQ, используя разные интенты с/без BPC=on, чтобы увидеть, влияет ли это на значение чистого чёрного пятна.

Если внимательно присмотреться к цветовой индикации, то можно заметить легкий желтоватый оттенок в профиле ArgyllCMS, компенсирующий оптическое отбеливающее средство этой конкретной глянцевой бумаги с плотностью печатных букв (вероятно, это встроенная настройка параметров colprof в ChromIQ, поскольку используется имитация, а не реальное измерение).

That is precisely why I stopped using RGB profiles a long time ago and now work exclusively in CMYK, so that I can better control the black channel and the grey balance.

 

[PK Project Manager]

Questions on Custom Printer Profile Validation and Measurement Conditions

I would appreciate some advice on the best approach for validating a custom ICC printer profile, particularly regarding analytical verification and measurement conditions.

1. Analytical Validation vs. Visual Evaluation

When validating a custom printer profile, do you normally perform an analytical evaluation by comparing the measured ΔE2000 values against the ColorChecker 24 (CC24) reference values, or do you rely primarily on visual evaluation of the printed target?

In particular, is there a generally accepted ΔE2000-based methodology or tolerance for determining whether a custom printer profile is performing satisfactorily?

2. Appropriate Measurement Condition for Profile Validation

I created my printer profile using M1 measurements. However, when subsequently measuring the printed CC24 target for validation, I have noticed that M2 (UV-cut) consistently produces lower ΔE2000 values than M0 or M1.

This raises a question about which measurement condition should be used for profile verification.

Should validation be performed using the same M1 measurement condition used to create the profile, so that the verification is consistent with the profiling data?

Or is it technically valid to use M2 (UV-cut) for validation if it produces a better numerical agreement with the CC24 reference values, even though the ICC profile itself was built from M1 measurements?

I am particularly interested in understanding whether the measurement condition should be considered part of the profiling/validation chain, or whether the validation condition can be selected independently based on which condition provides the most meaningful assessment of the printed colors.

Any practical experience or recommendations would be greatly appreciated.

Hi @jctay. This is an interesting deep dive into printer profiling and color accuracy.

We've featured your thread on the PrinterKnowledge homepage so more members can check out your findings and join the discussion. It's great to see members sharing detailed tests and evaluations like these.

 

[pharmacist]

Just an update with the latest version of ChromIQ with the bug fix for the truncated values of i1Profiler.

I have scanned the same target again twice: the first with i1Profiler and the target generated with the i1Profiler engine (neutralize greys at 50 procent) and again with ChromIQ (with the ArgyllCMS i1Pro driver loaded) and the target generated by ChromIQ with standard settings:

This time the white point of both profiles are exactly the same. A slight difference in gamut volume (most probably by slight scanning deviations within the margin of error)

Softproof with the ChromIQ version 4.3.20 scanned and ChromIQ generated target:

Target with the new exported txf file from ChromIQ 4.2.20 for i1Profiler scanning and profile generating by i1Profiler:

The difference is now negligible and no slight yellowish hue in the ChromIQ profile softproofing like previously.