i1Profiler: Adjusting "ambient light" for paper profiles

[MichaelKoerner]

Dear Forum,

after some years of color management experience (with monitors and softproofing only) I started writing printer profiles recently with i1Profiler and i1Pro 2. So I'm a mixture of experienced and newbie, please forgive any stupid questions (and my english, too - I'm coming from Austria .

Perhaps some of you can help me understanding the way i1Profiler works:

In my current setup I use a NEC PA271Q, calibrated with SVII and Discus. Softproofing works usually very well. Thanks to many recent posts in this wonderful forum I think to understand the concept of color accomodation. Yet I was interested in different "ambient light" settings, using the same measurement data when writing a profile.

At my current state of experience results are far from usable. Example: I measure a 75W Tungsten lamp (with QT&A, CRI-Mode) as 2667K. I use this value as "ambient light" in i1Profiler. The resulting profile I use for printing the file with Photoshop (Rel. Int, BPC).

The outcoming print is harsh, with way too much contras and a yellowish cast. I don't assume that my judgings faulty due to color accomodation - this is just ugly ;-)

"Normal" prints (Same data saved with D50 as ambient light) work fine.

Here are my questions: Did I misunderstand the concept? Or does this feature simply not work in practice (too much simulation)?
And: Does someone know why "paper white" is disabled in PS Softproof when I set less than 2950K as ambient light in my profile?

Looking forward to your feedback - thanks in advance!
Michael

[digitaldog]

Unless it's a bug, the use of a custom WP for the profile, instead of D50 shouldn't produce the results you're reporting. The feature is supposed to bypass a default assumption of viewing a print under D50 and instead use the measurements of the illuminant for print viewing. When I used it in the past, the differences were subtle but visible. I was using a GTI Fluorescent booth.
Please provide a screen capture of the soft proof setup.

[MichaelKoerner]

Please provide a screen capture of the soft proof setup.

Andrew,

thanks for your quick reply! In the attached screenshot you can see the softproof settings for each of the two profiles (75 W left, D50 right).
As I mentioned in my first post, PS doesn't allow me to activate "paper white" during softproof (AFAIK because the WP of the profile is below 2950K). So softproofs looks unobtrusive here.

BUT: I also attached the softproof with a profile containing 2950K as WP. Here I can activate paper white - looks rather extreme. And as you can see from the rough scan of the actual print I attached, the output is, simply said, awful...

[digitaldog]

Whatever is going on with soft proofing, the bottom line is, that profile is awful and I'm assuming when you convert, you actually see this result (that it's not just the soft proof table)? If so, forget using that measured illuminant in the profile.

[Doug Gray]

Whatever is going on with soft proofing, the bottom line is, that profile is awful and I'm assuming when you convert, you actually see this result (that it's not just the soft proof table)? If so, forget using that measured illuminant in the profile.

Something's hosed.

I made a profile with the latest (v3.0) I1Profiler and compared it to a profile made from the exact same data earlier with the same, 4K,  HyperIcon Led spectra.

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\9800 Baryta 957 HyperIcon LED M2.icm"
Descr: "9800 Baryta 957 HyperIcon LED M2.icm" Ver: 4.2.0   Copyright: "Copyright X-Rite, Inc."
WP:  97.5  -0.3   1.4     BP_PI:   5.3  -0.5  -0.5    BP_RI:   5.3  -0.4  -0.5

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\9800 Baryta 957 HyperIcon LED NewI1P M2.icm"
Descr: "9800 Baryta 957 HyperIcon LED NewI1P M2.icm" Ver: 2.1.0   Copyright: "Copyright X-Rite, Inc."
WP:  97.7  -6.4 -16.6     BP_PI:   5.4  -1.3  -3.6    BP_RI:   5.4  -1.2  -3.6


Note the difference in the White Points. This is nuts. The first one is correctly Bradford adapting. The second (current I1Profiler) is not.

Abs. Col. actually prints with a strong reddish tint. The soft proof of the same is a strong bluish tint. It's totally messed up.

[MichaelKoerner]

Something's hosed.

Doug,

thank you for your effort recreating the effect. I just sent an email to x-rite tech support and will report on the outcome.

[MichaelKoerner]

Whatever is going on with soft proofing, the bottom line is, that profile is awful and I'm assuming when you convert, you actually see this result (that it's not just the soft proof table)? If so, forget using that measured illuminant in the profile.

@Andrew
just to be sure to get you right: When stating the profile to be "ugly", do you mean that it's technically wrong (and there was something I could do about it), or do you mean the result it shows? As I'm rather new to writing output profiles I don't own ColorThink Pro or the like (nor would I understand it's functions yet). Perhaps someone with proper knowledge could have a look at the profile (attached) for human mistakes.

To answer your question: When I convert the image from it's source color space (sRGB) to my 2950K profile in Photoshop (RC w/BPC), it looses contrast, but color hues remain stable. Softproofing this converted file (to the same 2950K profile) shows the extreme blue cast again. Printing it via ACPU shows the same yellowish, high contrast as before.

@Doug
Just to be sure to get you right: You assume this to be a bug in the current i1Profiler version?

In the meantime: Can you confirm (or correct) my understanding, that the concept of setting an ambient light in i1Profiler is
a) to adapt a print to "look the same" in a non D50 viewing light
or b) Just to simulate how it would look in this non D50 unchanged, neglecting the effects of color accomodation (is this the correct english expression?)).

If the discussed profile's behaviour was not due to a bug or a mistake of mine, I would assume the latter.

[Doug Gray]

@Doug
Just to be sure to get you right: You assume this to be a bug in the current i1Profiler version?

In the meantime: Can you confirm (or correct) my understanding, that the concept of setting an ambient light in i1Profiler is
a) to adapt a print to "look the same" in a non D50 viewing light
or b) Just to simulate how it would look in this non D50 unchanged, neglecting the effects of color accomodation (is this the correct english expression?)).

If the discussed profile's behaviour was not due to a bug or a mistake of mine, I would assume the latter.

First, Profiles created using anything other than D50 as the illuminant produces non-conforming profiles. ICC profiles are always based on D50 and media whitepoints are adaptated to D50. This is why an image in Adobe RGB, which has a media white point of D65 will "soft proof" the same in Abs. Col. and Rel. Col. when using a monitor that is set to D50. ICC's specification of Abs. Col. requires this and deviates from CIE's definition of Abs. Col. This was an area of confusion in the early days and still shows up in Microsoft's color engine. Adobe's ACE color engine conforms to this ICC requirement.

Now to printer specific profiles.

These are always D50 based. There is no ICC process for making ICC profiles for prints that are to be viewed under other illuminants. At least for versions V2 and V4 profiles.

I've avoided using soft proofing for printer profiles created using other illuminants than D50 and always soft proof using the standard, D50 profiles. I have occasionally used profiles made with different illuminants to get a better viewing experience with fluorescents and LEDs but only with CCTs of 4000K or more. I only use these profiles for printing. I soft proof using D50 profiles even though this is less than accurate near the printer's gamut edges.

Since ICC profiles don't support different illuminants for printer profiles, one can't say what XRite does is right or wrong. The problem is that it has changed and that change is not documented. The current approach completely hoses over soft proofing using show paper color.

What I don't know is whether this change has affected how prints are made using Perc. or Rel. Col. But it has radically changed prints made using Abs. Col. Basically, Abs. Col. prints are now made in such a way that, when viewed under D50, will appear as if they were viewed under tungsten. That is they will print quite reddish. When viewed under tungsten they will appear grossly reddish and totally wrong. This is just weird and not what people expect.

In any case it's rarely appropriate to make profiles using tungsten, or even LED, illuminants that are < 4000K CCT.

Since I haven't used the current I1Profiler for anything other than D50, I don't know whether it has changed Rel. or Perc. printing but it has radically changed soft proofing and Abs. Col. printing.

[digitaldog]

Now to printer specific profiles.
These are always D50 based. There is no ICC process for making ICC profiles for prints that are to be viewed under other illuminants. At least for versions V2 and V4 profiles.

[Doug Gray]

XRite, and Argyll, provide processes for using different illuminants to create printer profiles but the ICC does not specify any constraints or definitions for these processes. For instance, D50 profiles are defined for Rel. Col. including AtoB1 and BtoA1 transforms.

[digitaldog]

XRite, and Argyll, provide processes for using different illuminants to create printer profiles but the ICC does not specify any constraints or definitions for these processes.

There is no ICC process for making ICC profiles for prints that are to be viewed under other illuminants.

I'm simply pointing out the possible confusion for your readers.

For instance, D50 profiles are defined for Rel. Col. including AtoB1 and BtoA1 transforms.

Which is why we have other rendering intent options for such profiles. 

[Doug Gray]

I'm simply pointing out the possible confusion for your readers. Which is why we have other rendering intent options for such profiles. 

Yep. Anyone that sells profile software wants to be able to distinguish themselves from competitors, Perceptual and Saturation intents provide a mechanism for so doing while maintaining ICC compliance and the printed results can be reversed using AtoB1 tables enabling good softproofing.

And then there's the problem  of some non-compliant (for the latest V2 and V4 specs), but still D50, profiles such as those that came with my 9800 and had baked in BPC in the AtoB1 tables which messed up softproofing in dark shadows.

[Doug Gray]

V3.0 of I1Profiler is completely screwed up with not only wrong results for A, and F2 but skewed in the opposite direction of what they should be doing. V1.7 performs as expected.

Using both V3.0 and 1.7 of I1Profiler, I just made 3 profiles using illuminants D50, F2, and A.

F2=industrial, crappy fluorescent useful because it has a much reduced long wavelength spectra and makes reds look very subdued. This requires pushing reds when printing to compensate. So it's a good test of I1Profiler. OTOH, A is the standard tungsten black body illuminant with intense longer wavelengths.

I'm going to start a new thread on this very major bug and detail the results.