Printer Profile Validation: M1 vs. M2 and ΔE2000 Evaluation

know · 2026-10-23T12:45:06+0100

jctay said:
After resolving the measurement issue with Strip 36 and rescanning several other strips with misread patches, selected excerpts from the measurement records are presented below.

Very nice

Thanks for the reporting, it helps a lot to find and fix bugs...

Things are improving, but I think there is two more bugs hidden in the resulting reports you show us...
Thus this is input for @itsab1989 and claude.

1. Grey balance is failing:

The 572 patch chart you are using has a single line of neutral grays, but after you fill the chart with "within-gamut-colour" using the "From Profile's Gamut" feature, that feature does not place enough grey patches along this line of neutral greys. I think this was possible a result of the work we did to distribute patches more evenly, but should be checked that the same amount of greys that are present in the chart from before is placed when using the "From Profile's Gamut" feature. This should be analysed.
The 572 patch chart is found here if analysis of the patch set is required:
ArgyllCMS 576-patch target + 64 greys

2. Evenness check requirements not fulfilled:

The report note says that the chart used only covers 55.7% of the page (and 60% is the required limit), which looks strange when looking at one page from the 572 patch chart:

If one adds all the margin areas, it might be true that this becomes mathematically 44.3% of the page... But that calculation does not consider that the patch area is spread out to cover most of the page and the margins are around the patch area. The minimum limit of 60% of the page area calculation, and the method used, needs to be re-evaluated so that the calculation takes into account that the patch area is spread evenly out on most of the page, and not purely looking at a mathematical area calculation, since the margins summed up give a false impression of the whole page coverage of the patch area. Dropping the limit to 50% is of course also an alternative, but this method, as I mention, is a bit misleading to represent coverage of the patch area across the page.

The chart also fulfils the other requirement for the evenness checks to be calculated, which a minimum of 9 by 9 columns and rows. This metric should work on this kind of chart.

Cheers,
Knut

itsab1989 · 2026-10-23T13:01:04+0100

I have added this to the to do list. Computer is running 9 hours straight till this point so I hope with this added 3 hours are enough for it to release the update…