Colormunki/ArgyllCMS profile lacking dark tones (compared to ccStudio)

[lmylm]

please be aware that ArgyllCMS is raising the blacklevel for a profile with rel. col. intent and BPC on, exactly tlhis was already discussed in the thread a year ago, and as well with the author of ArgyllCMS, and there is a related problem with the use of the perceptual intent; these are deviations of the ArgyllCMS behaviour vs. other profiling packages

If I understand this correct, this would make ArgyllCMS unusable for photographers that are interested in actually having darker black tones in the printout.

particular combinations of BPC and coloring intents should not be used.

I only know of four rendering intents (Perceptual, Relative Colorimetric, Absolute Colorimetric and Saturation). I read that for photography, which I profile for, only Perceptual and Relative Colorimetric are useful. Also, from what I read and understand for the Perceptual rendering intent it doesn’t matter if BPC is activated or not. That leaves me with these three combinations:

- Perceptual
- Relative Colorimetric without BPC
- Relative Colorimetric with BPC

I really hope that not all of them are combinations that should not be used. No matter wich combination I tried I get the same result:

Colormunki/ArgyllCMS profile: overall look of printout lacks contrast because of too bright black tones
Colormunki/ccStudio profile: prints look much better (yet less accurate colors because of 50+50 patch target sheet)

As so many people (photographers?) seem to work successfully with ArgyllCMS I really wonder which colprof parameters use all the (480) Colormunki readings to create an ICC profile which is accurate both in colors and black levels.

I must admit I don't like an icc-profile which changes the black level that much; I don't think it's the job of an icc-profile at all to manipulate the blacklevel.

I think I agree. From what I learned about color management so far I would prefer options for the kind and strength of black point correction in the printing software. But there isn’t. With that in mind all I care for is a color/contrast accurate ICC profile that I can use for my prints. ArgyllCMS profiles deliver color accuracy, ccStudio profiles deliver contrast accuracy. Maybe I find a way to combine two profiles

I will get deeper into analysing those profile and understanding ICC profiles in general. Need to setup some Windows environment first though, so I can run Gamutvision myself.

@pharmacist

In the meantime I started profiling a few papers with your 480 patch target sheet. Every profile I created so far looks really messy:

The messyness differs from profile to profile but overall is very similar. I don’t get this from targets I created myself (or from ccStudio). Is there something wrong with the ti files you supplied? Is my Colormunki broken?

 

[pharmacist]

The 3D-model looks like wrong reading. Please reread the target at a much slower pace, about 3 seconds over 1 row. The reason is the Colormunki is 4 times slower in taking samples from the patches (50 samples/second instead of 200 samples/second for the i1Pro2). It is full of errors. I always do a softproofing (using a with a generated profile to evaluate the quality: if you see strange colour transitions in your profile, then you know the profile is not well done.

Use this image and look carefully the colour transitions in the coloured bars in the left lower area. If you see strange and sudden colour changes the profile is wrong.

 

[lmylm]

@pharmacist Thank you for your reply. I was afraid of exactly this. I already assumed that because of the small patches I need to be very precise and slow in reading the target sheet.

That’s why I took my time and scanned at about 6 to 8 seconds per row. The very same speed I used for the target sheets I generated myself (17 vs 20 patches per row). My own target sheets come out smooth though.

Actually at about 3 seconds per row (probably a bit less) ArgyllCMS does not accept the scanned row and asks to rescan it (“Strip read failed due to misread (Not enough samples per patch - Slow Down!)”). So, I was hoping that as long as I do not get the message about ArgyllCMS not having enough samples I actually do get enough samples.

Maybe there is some problem other than the reading speed?

Anyways! I will reread the sheet and make sure to read slower than 3 seconds per row, just to make it’s not that.

 

[Ink stained Fingers]

this would make ArgyllCMS unusable for photographers

lots of these photographers will use glossy papers which have a much lower black level - probably in the range of L=+-3 , and even if ArgyllCMs would raise the level by 50% to 5 about nobody would recognize it unless you remeasure the blacks in a profiled print. And other photographers may use OEM papers or papers to which the paper manufacturer supplies profiles.

 

[nertog]

Argyll is, in fact, doing the correct thing which is minimizing colorimetric errors. This leads in some scenarios to raised blacks. You can use the perceptual rendering intent and ask Argyll to do minimal gamut compression when building the ICC-profile. This should give you lower black levels.

Other ICC packages might trade-in color accuracy for black level near absolute black. Maybe a manual control option would be nice to have.

 

[nertog]

I must admit I don't like an icc-profile which changes the black level that much; I don't think it's the job of an icc-profile at all to manipulate the blacklevel. But if that is not known to the user the softproof function of some programs deliver you a distorted image and may lead the user to some suboptimal changes to his image.

That's exactly what Argyll promotos: strict adherence to the ICC-standard. If X-rite or others deviate from that and cheat a bit to get deeper, but less neutral black...then that is a deliberate choice.

The ICC-profile does not change the black level, but the way it is interpreted and used by your printing software might. Also, as discussed before, BPC is not part of the ICC standard and might be implemented differently by different developers.

Having said that, I find contrast more important than color accuracy as well...so X-rite's approach might not be that bad

 

[Ink stained Fingers]

Argyll promotos: strict adherence to the ICC-standard. If X-rite or others deviate from that

Do you have any documented evidence for that ?

The icc-profile is a set of data tables + some misc parameters but is not any active program code which does anything on its own ; I don't see that this was understood differently.

The actual approach by ArgyllCMS is causing issues for the user - specifically with some paper types and usage parameters - rendering and BPC, and these cannot be tested away with some more patches or this or that other option. And 'Argyll promotos: strict adherence to the ICC-standard ' is not of help for the user in these cases.

 

[Ink stained Fingers]

Perceptual
- Relative Colorimetric without BPC
- Relative Colorimetric with BPC

I really hope that not all of them are combinations that should not be used.

That's a kind of similar situation which got discussed here longer time ago with Dataco...profiles in combination with Photoshop Elements causing as well elevated blacks in softproof mode and in print.

It took a while to find out that BPC is always on with the perceptual rendering intent in PS Elements/Lightroom , and it took a while longer to find a statement in the Dataco...documention hidden on page 4 of a leaflet - which about nobody reads - that the perc. rendering intent should not be used at all and the rel. col. intent only with BPC off.

 

[lmylm]

@pharmacist @nertog @Ink stained Fingers

Thank you for your latest comments on this topic. In the meantime I reread the 480 patch target sheet on all papers. This time actually counting seconds for each strip. Turns out I indeed was too sloppy, at least with some of the rows. Thank you for the hint!

With the newly created profile(s) I printed more test prints and scanned them. This obviously is a highly subjective test. It gives a rough estimate though. I used all combinations for Relative Colorimetric and Perceptual rendering intents with and without BPC.
For me the ArgyllCMS Perceptual (no BPC) profile (left) and the ccStudio Perceptual (with BPC) profile (right) get the closest to the original (center):

ArgyllCMS is still far off in color and contrast. The only advantage I see is higher detail in the shadows. This may count in some cases, especially if there is some scientific background to the images where shadow detail may be of importance. For general reproduction of artistic images though I like to think one would rather sacrifice a bit of shadow detail and instead have the overall look and feel of the original.

With this in mind I still don’t get the fuzz about ArgyllCMS. Either it is useless for artistic prints or I am using it wrong. In general I tend to believe it’s me who is wrong—not the entire world. So I would love to find out how to use ArgyllCMS the correct way (so I don’t have to waste two sheets for each profile and am allowed to share my profiles).

 

[Ink stained Fingers]

This is a look to some of the profile data, the icc-profile is created along the standardized icc-rules, a file header contains various parameters like the black point - or the white point, the black point is displayed in various colorimetric formats and is the value you gathered from your scanning of the target sheet - in this case a L=18.7 as the luminance value. There are various A to B and B to A tables - these contain the color correction values for the colorimetric transformation from and to the printer and the PCS - profile connection space for the various rendering intents. The BPC does not apply at this level. You could - if you have enough time - do a readout of these table values with a binary file editor. It is relevant to understand that the program Gamutvision does not do a readout of the raw black level but of the value stored in the profile sections - regrettably displayed in a logarithmic format D=density, but it shows at least that the blacklevel D varies with the setting of the rendering intent+BPC. .

In general I tend to believe it’s me who is wrong—not the entire world.

Not that many users go that deep into the details as you do and are able to identify inconsistencies as you do.