Making Profiles for Different Light Sources

[Mike Sellers]

How do you make profiles based on the light source where the print will be hung?
Mike

[howardm]

The couple of profile generation software packages I've seen give the ability to select an illuminant either from a
pulldown list of choices or from a reading of the lightsource in question.

I'm thinking of doing just that just to see the differences as my display space is lit by a bunch of 2700K incand. lights
so I'm assuming Illuminant A and then compare to the std. D50

This is from the Argyll 'colprof' documentation......

The -i parameter allows specifying a standard or custom illumination spectrum, applied to spectral .ti3 data to compute PCS (Profile Connection Space) tristimulus values. A, D50, D65, F5, F8, F10 are a selection of standard illuminant spectrums, with D50 being the default. If a filename is specified instead, it will be assumed to be an Argyll specific .sp custom spectrum file. This only works if spectral data is available. Illuminant details are:

        A   CIE tungsten filament lamp 2848K
        D50 CIE daylight 5000K
        D65 CIE daylight 6500K
        F5  CIE Fluorescent 6350K, CRI 72
        F8  CIE Fluorescent 5000K, CRI 95
        F10 CIE Fluorescent 5000K, CRI 81

Custom illuminants are most often used when a  viewing booth or other known viewing conditions is going to be used to view results. Other illuminant reference files could be created using a suitable measuring instrument such as a spectrolino, or an eyeone using spotread, although such instruments do not themselves provide the necessary response down to Ultra Violet that is needed for accurate operation of Fluorescent Whitening Agent compensation. The best way of measuring a custom illuminant is to use illumread, since it uses a special method to estimate the illuminant UV in a way that complements FWA compensation. (See the discussion above for the -f flag).

Note that if an illuminant other than D50 is chosen, the resulting ICC profile will not be standard, and may not work perfectly with other profiles that that use  the standard ICC D50 illuminant, particularly if the absolute rendering intent is used. Profiles should generally be linked with other profiles that have the same illuminant and observer.

[digitaldog]

You need to measure the illuminant with a Spectrophotometer and produce spectral data, usually in the form of a cfx file. You'll need software like ProfileMaker Pro or i1Profiler to use that new dataset instead of aiming for D50.

[Mike Sellers]

What if you are unable to physically go to the location where your print will hang to take measurements? If the buyer says they have fluorescent lighting then what do you do about a profile that will be in the "ballpark"?

[digitaldog]

What if you are unable to physically go to the location where your print will hang to take measurements? If the buyer says they have fluorescent lighting then what do you do about a profile that will be in the "ballpark"?

You could try a generic Fluorescent CfX file. You could just build a plain old ICC profile like most people and hope for the best. Fluorescent suck for some output, building an illuminant specific profile isn't going to fix the issues with that illuminant and it's possibly negative effect on the prints that have high OBA's and such.

[John Nollendorfs]

DataColor's Spyder Print has the ability to fine tune a profile based on color temp of viewing lights. Also all kinds of other fine-tune tools. You do not actually tune the profiles, but create new profiles from your existing data to reflect those changes.

[digitaldog]

DataColor's Spyder Print has the ability to fine tune a profile based on color temp of viewing lights. Also all kinds of other fine-tune tools. You do not actually tune the profiles, but create new profiles from your existing data to reflect those changes.

Editing profiles is excellent if:
You charge by the hour and, the output isn't something you pay for (OPM or Other People's Media). Good profiles don't need editing.

[John Nollendorfs]

That's true Andrew, but if you can quickly and easily create new profiles for specific needs, based on your existing good profile data. . .

[Mike Sellers]

Reading this on the Byron Jorjorian blog prompted my question:
The majority of fine art prints which I produce are produced using a mixed lighting profile. It is designed to give the most natural looking color under a varity of lighting conditions. It works well even under conditions where a combination of light sources (for example flourescent and daylight) might both be illuminating the photographic print.
If you know that the image will be viewed only under a specific light source and not moved you can let us know and we can set the color to create the art print to look best in your environment.
However, if the photograph may be moved or viewed under different light sources it is best for us to use the mixed lighting profiles that I use by default.

[digitaldog]

The human visual system is wonderful ad adapting to the white of the illuminant. It isn't so great at dealing with color casts in prints due to metameric failure or issues with OBA's in papers.

[Robert Ardill]

The couple of profile generation software packages I've seen give the ability to select an illuminant either from a
pulldown list of choices or from a reading of the lightsource in question.

I'm thinking of doing just that just to see the differences as my display space is lit by a bunch of 2700K incand. lights
so I'm assuming Illuminant A and then compare to the std. D50

This is from the Argyll 'colprof' documentation......

The -i parameter allows specifying a standard or custom illumination spectrum, applied to spectral .ti3 data to compute PCS (Profile Connection Space) tristimulus values. A, D50, D65, F5, F8, F10 are a selection of standard illuminant spectrums, with D50 being the default. If a filename is specified instead, it will be assumed to be an Argyll specific .sp custom spectrum file. This only works if spectral data is available. Illuminant details are:

        A   CIE tungsten filament lamp 2848K
        D50 CIE daylight 5000K
        D65 CIE daylight 6500K
        F5  CIE Fluorescent 6350K, CRI 72
        F8  CIE Fluorescent 5000K, CRI 95
        F10 CIE Fluorescent 5000K, CRI 81

Custom illuminants are most often used when a  viewing booth or other known viewing conditions is going to be used to view results. Other illuminant reference files could be created using a suitable measuring instrument such as a spectrolino, or an eyeone using spotread, although such instruments do not themselves provide the necessary response down to Ultra Violet that is needed for accurate operation of Fluorescent Whitening Agent compensation. The best way of measuring a custom illuminant is to use illumread, since it uses a special method to estimate the illuminant UV in a way that complements FWA compensation. (See the discussion above for the -f flag).

Note that if an illuminant other than D50 is chosen, the resulting ICC profile will not be standard, and may not work perfectly with other profiles that that use  the standard ICC D50 illuminant, particularly if the absolute rendering intent is used. Profiles should generally be linked with other profiles that have the same illuminant and observer.

Hi,

I also use ArgyllCMS but I'm confused about this whole business of illuminants and FWA compensation.  It's clear that if you produce a profile with a different illuminant (say A) that it is different to the standard D50 illuminant (comparing a copy of the image converted to the two profiles).  However, what I'm confused about is the objective.  Is it to compensate for the slight color casts introduced by the different illuminants? (seems like it when comparing the converted images) ... so that if the print is viewed under illuminant A and then under 'D50' using the appropriate profiles, that the relative color differences will be the same?

Clearly the profile cannot make the illuminant A print have the same white point as the D50 print (it would have to make the whole image way cooler, and it cannot change the paper white).

And what about FWA compensation?  Does the profile attempt to remove the effect of UV so that a print made with FWA compensation, viewed using the actual illuminant which does have UV, will appear as it would with UV filtered out from the illuminant?  In other words that it would push down the blue-whiteness caused by the optical brightener?  Again ... the profile can't actually change the white of the paper (unless it puts down ink over paper white, I guess).

Robert

[digitaldog]

  Is it to compensate for the slight color casts introduced by the different illuminants? (seems like it when comparing the converted images) ... so that if the print is viewed under illuminant A and then under 'D50' using the appropriate profiles, that the relative color differences will be the same?

Yes. FWA compensation (at least the X-rite method) and uses visual targets you view then update profile so different.

[Robert Ardill]

Yes. FWA compensation (at least the X-rite method) and uses visual targets you view then update profile so different.

Hi Andrew ... thanks for your reply ... but I don't quite understand what you are saying.

Robert

[digitaldog]

Hi Andrew ... thanks for your reply ... but I don't quite understand what you are saying.

Two different options in the software for two different functions. And FWA compensation isn't available for all data (you need an iSis and I think a newer i1Pro Spectrophotometer).
http://www.xrite.com/documents/literature/en/OBC_User_Guide_en.pdf

[Robert Ardill]

OK, yes, I know that FWA compensation and illumination compensation are different corrections.  What I'm not clear about is what the correction attempts to achieve.

For example, with illumination compensation, just taking the illuminant white point and not more complicated things like spikes in the spectrum:

• Without compensation I assume that the tri-stimulus values are normalized to D50. So when you print you are effectively printing to D50.
• With compensation, say the actual white point is 4000K. The tri-stimulus values will be normalized to that temperature and not to D50 (I guess??)

But what does that mean? Does it mean:

• a. that the profile will attempt to make the print, viewed at 4000K, look like a non-compensated print viewed at D50? In other words that it will print 'cooler'?  (would probably be impossible to do, actually!)
• b. that the profile will correct the colors so that the relative difference between them will be retained? So there's no attempt being made to alter the print temperature
• c. or something else? ... for example that the only purpose is to adjust for small differences based on very similar illuminants and papers for proofing purposes?
• ...

BTW ... you can do FWA compensation with ArgyllCMS with any supported spectrometer (it uses a special algorithm to estimate the UV effect).  With i1Profiler you need to have the i1Pro2.

If anyone is interested in using Argyll, I have some useful batch files (Windows) that make some of the tasks somewhat automated (Argyll can be a bit daunting as it's all command-based, although there is a GUI available for monitor calibration - DispCalGUI).

Robert

[Tim Lookingbill]

Robert, if you do a search on LuLa using the term "ArgyllCMS" you'll find just how familiar most of us are with that software. You seem to be far more knowledgeable for someone to be asking basic questions on the variances among print viewing light behavior on prints.

Have you directed your questions to the authors of that software? You might get more informed answers.

I used an ArgyllCMS made ICC profile for a Fuji Frontier drylab and it was quite useful, but what it couldn't predict or mitigate against (and no other software could as well) is the fluorescing under all forms of light regardless of white point illuminant caused by the cleverly engineered polycoated embedded ink that was prevented from being absorbed into the paper. It was a very interesting effect as if the print was backlit or something similar to viewing Ektachrome slides on a light table. The farther away from the light the less vibrant it became.

The ArgyllCMS profile did not show this effect during Soft Proofing.

[Robert Ardill]

But what does that mean? Does it mean:

• a. that the profile will attempt to make the print, viewed at 4000K, look like a non-compensated print viewed at D50? In other words that it will print 'cooler'?  (would probably be impossible to do, actually!)
• b. that the profile will correct the colors so that the relative difference between them will be retained? So there's no attempt being made to alter the print temperature
• c. or something else? ... for example that the only purpose is to adjust for small differences based on very similar illuminants and papers for proofing purposes?
• ...

I know I'm quoting myself here ( , but I did a quick test this morning that's probably answered my question.

I printed a color chart using illuminants A and D65 (with Argyll-generated profiles). What seems to happen is that in relative terms, the "A" illuminant profile bumps up the blues and greens and reduces the reds.

So, looking at the relative SPDs of illuminant A (red line below) and D65:


it appears that the profile is attempting to adjust the image to compensate for the weakness of the A illuminant at the blue-green end and it's dominance at the red end.

So the compensation does what it says: it attempts to normalize the image viewed under the specific illuminant to (I assume) a D50 color balance.  Compared to D50 I expect that the D65 profile would reduce the blues and increase the reds.

The viewed image (allowing for chromatic adaptation and with illumination compensation) will appear nearer to what it would be if viewed under daylight conditions and with no illumination compensation.

I assume that FWA compensation would have the same effect: that is, that if the illuminant has less UV than D50, the blues will be bumped up, if it has more, the blues will be reduced.  There should be no effect (or very little) at the green-red side of the spectrum.

Does that make sense?  It sort of does to me, but then I'm a newbie at color management.

Robert

[/list]

[Robert Ardill]

Robert, if you do a search on LuLa using the term "ArgyllCMS" you'll find just how familiar most of us are with that software. You seem to be far more knowledgeable for someone to be asking basic questions on the variances among print viewing light behavior on prints.

Have you directed your questions to the authors of that software? You might get more informed answers.

I used an ArgyllCMS made ICC profile for a Fuji Frontier drylab and it was quite useful, but what it couldn't predict or mitigate against (and no other software could as well) is the fluorescing under all forms of light regardless of white point illuminant caused by the cleverly engineered polycoated embedded ink that was prevented from being absorbed into the paper. It was a very interesting effect as if the print was backlit or something similar to viewing Ektachrome slides on a light table. The farther away from the light the less vibrant it became.

The ArgyllCMS profile did not show this effect during Soft Proofing.

Hi Tim,

First of all, I'm not at all knowledgeable about color management ... I'm totally green at it and I'm only just beginning to understand the basics.  I'm equally ignorant of ArgyllCMS, how to use it, how it does things like illuminant and FWA compensation.  And I don't understand basics in Photoshop like how soft-proofing works. When I look at the Lab value of a color and the RGB value on my monitor in sRGB and Adobe RGB and see that the RGB values are different for the same Lab value, I'm only guessing that the reason for this is that the sRGB color space is smaller than the Adobe RGB color space, so the red value in the sRGB color space will be higher than in the Adobe RGB color space if the color is towards the red (because the RGB color is relative to the color space) ... and so on.

The reason I'm asking these questions here is that many members of this forum are very knowledgeable and have gone through this process of understanding what's what already.  If I keep bugging people like Graeme Gill ... well I'm going to get very unpopular and I'll distract him from improving his already excellent software. Also, he's only one person, whereas there are many on this forum.  If I asked the question of Adobe ... would I get an answer?  Eventually, maybe, but I don't know anyone on the development team (again, unlike some members of this forum), so it would be difficult for me.  As for X-Rite ... forget it, in my experience. That company is the closest thing I've come to a black hole.

Regarding Argyll and specially formulated inks ... well, what do you expect?  It's hard enough compensating for optical brighteners in the paper, but when you get the inks doing all kinds of weird and wonderful things ... of course, you have to print and you can't expect soft-proofing to do it for you.  You know that yourself.

Robert

[digitaldog]

OK, yes, I know that FWA compensation and illumination compensation are different corrections.  What I'm not clear about is what the correction attempts to achieve.

Primarily one adjusts the 'effect of FWA" by visually viewing output of grays from the software, then imputing which patch looks best after which the software tries to compensate (probably by altering neutrals and near neutrals, that's a guess).

[Robert Ardill]

Primarily one adjusts the 'effect of FWA" by visually viewing output of grays from the software, then imputing which patch looks best after which the software tries to compensate (probably by altering neutrals and near neutrals, that's a guess).

Hi Andrew,

I don't really understand what you are saying ... I take it there's some software that can be used with manual selection of the best gray to estimate the effect of the optical brightener?  i1Profiler? Something else?  ArgyllCMS has an algorithm that's described here: file:///H:/bin/Argyll_V1.6.3/doc/FWA.html and in more detail here: http://www.imaging.org/IST/store/epub.cfm?abstrid=22190.  Essentially, the software flattens out the characteristic lump in the spectrum which occurs around 450nm.  By looking at the paper white measured by the instrument (which has a known UV content), taking a reading of the reflected white from the illuminant, and measuring the illuminant directly, it can estimate the UV content in the illuminant and with this estimate the FWA correction required for the paper under the particular illuminant.  I don't know how well this works.

I do have an i1Pro2 which has two light sources, one with no UV, so I should be able to use it to compensate for FWA ... but I made the mistake of buying the i1Basic package so I'm now waiting on delivery of the i1 Publish upgrade (which should hopefully allow me to check this out).  However ... mostly this isn't a problem for me because I don't use papers with optical brighteners much.

If you look at my 11:15 post (one before last), I say "I assume that FWA compensation would have the same effect: that is, that if the illuminant has less UV than D50, the blues will be bumped up, if it has more, the blues will be reduced.  There should be no effect (or very little) at the green-red side of the spectrum.". 

When it comes to soft-proofing, my assumption is that one should have two profiles, one with Illumination and FWA compensation, the other without; the choice the is:

• Print using the profile 'with compensation' and soft-proof with the profile 'without compensation' (as there is no compensation required to the monitor image.
• Print using the profile 'without compensation' and soft-proof using the profile 'with compensation'.

The second option would seem wrong, but if we use the first option then we are attempting to remove or reduce the effect of FWA and the illumination difference to D50 ... which I don't think one should do unless we know precisely what illuminant will be used to view the print (which mostly I do not, except for exhibitions).  So the second option leaves things as they are, but should help to match the soft-proof to the print more accurately.  Assuming, that is, that I don't have this head-to-tail or worse!

I'll check this out with Graeme Gill, once I'm reasonably confident that I understand what I'm talking about.  As usual, any help would be most welcome!!

Robert

[/list]