What Illumination choice??

[Grandad35]

I was unaware of this "feature" in argyll. It reminds me of the many situations at work when competitors routinely announced new concepts that drove our sales department into a frenzy. My response was usually "Just because you can do something doesn't mean that you should" as most of these concepts were proven to have major flaws which were readily apparent if you just took some time to think about them. However, since "innovations" get lots of free press (even if they are bad ideas), everyone does it.

Take a look at the illumination spectra for various types of lighting in this post. The top two plots show that the standard 5000K illumination is very close for direct low morning sun (it changes during the day as the sun rises and the distance that the light travels through the atmosphere changes), but who displays their photos under direct sunlight? However, when using reflected morning light, the light is now 3200-3300K. Next, compare the illumination values - 20000 LUX for direct sunlight vs. 40-70 LUX for reflected light. Would that make a difference? Two different types of fluorescent bulbs generated light at 2700K and 3900K, and they had very different spectra. How many people look at the color temperature when they buy light bulbs? Since CFL bulbs lose intensity over time, the temperature of the light may also change over time.

Next, consider how well our brains naturally adjust to various lighting conditions. When you move into an area with a different type of lighting do the colors look really strange, or does your brain automatically take into account the change in ambient lighting and adjust your perception accordingly? This is fortunate for those of us who hang our prints in rooms that are lit with reflected sunlight for part of the day and with artificial light at other times.

Finally, you can achieve the same type of effect by introducing a "photo filter" adjustment layer in Photoshop to put the desired color cast into the image (which is what the modified profiles apparently do). This has the added benefit that you can customize the color cast to match your particular lighting and not be limited to just a few fixed choices.

All of this is not to say that using standardized viewing booths to make critical color comparisons isn't valid, only that there are very few people who own a museum and display their prints in a closed room with calibrated lighting. Even then, why not just buy 5000K lights to illuminate the prints? Imagine telling Rembrandt to change his colors to match our lighting.

Sorry for the rant, but "just because you can doesn't mean that you should".

 

[George in Georgia]

Think I'll buy the 5000 K lamps for the general lighting here in Command Central where I edit photos, print, surf the net, etc. The overhead fixture has sockets for three bulbs. Then I'll use the D50 ICC option.

Thanks to all for the help!

 

[RogerB]

Finally, you can achieve the same type of effect by introducing a "photo filter" adjustment layer in Photoshop to put the desired color cast into the image (which is what the modified profiles apparently do). This has the added benefit that you can customize the color cast to match your particular lighting and not be limited to just a few fixed choices.

All of this is not to say that using standardized viewing booths to make critical color comparisons isn't valid, only that there are very few people who own a museum and display their prints in a closed room with calibrated lighting. Even then, why not just buy 5000K lights to illuminate the prints? Imagine telling Rembrandt to change his colors to match our lighting.

Sorry for the rant, but "just because you can doesn't mean that you should".

I'm not sure it's quite as simple as that. The reason for having profiles for different viewing illuminants is that the colour balance of some ink sets will vary with different illuminants but not all colours are affected to the same degree. There are quite a lot of posts in the forums where people ask about red casts in their B&W prints. These are usually people using dye inks and viewing their B&W prints in tungsten lighting. The fact is that dye inks (CcMmYK) generally have poor colour constancy which is most obvious in B&W prints or in neutral areas in colour prints. If they look neutral in daylight they will generally look too red in tungsten and too cyan in "north" light.

Now, it may be possible to get around this by introducing an adjustment layer in PS, but only if the effect is linear and uniform. It's the same with any colour cast - you either "fiddle with the sliders" or you make a custom profile. In this case I believe that the poor colour constancy is caused by the cyan ink; most cyan dyes have a secondary transmission in the deep red. Because tungsten light is very rich in red the secondary transmission becomes very noticeable. Of course, there's a whole range of colours that use little or no cyan ink. These colours, which include Caucasian skin tones, have pretty good colour constancy, so applying an overall colour adjustment will probably make them worse.

Since Argyll gives the option to generate a profile for a specific illuminant it seems the easiest way to avoid colour casts if the illumination of the print is known in advance. Unlike an adjustment layer it will only correct the colours that need correcting. Furthermore, it doesn't require different charts to be printed and doesn't even require more measurements - one set of measurements can be used for profiles for any of the standard illuminants.

Having said all of that I have to confess that I have no direct experience of illuminant-specific profiles since my (pigment) printer has very good colour constancy. However, I commend Precision Colors for offering different profiles and would love to hear if anyone has used them to good effect (or otherwise).

 

[Grandad35]

I'm not sure it's quite as simple as that. The reason for having profiles for different viewing illuminants is that the colour balance of some ink sets will vary with different illuminants but not all colours are affected to the same degree. There are quite a lot of posts in the forums where people ask about red casts in their B&W prints. These are usually people using dye inks and viewing their B&W prints in tungsten lighting. The fact is that dye inks (CcMmYK) generally have poor colour constancy which is most obvious in B&W prints or in neutral areas in colour prints. If they look neutral in daylight they will generally look too red in tungsten and too cyan in "north" light.

Of course it not simple - if it were, color management would be easy to learn and we wouldn't have so many books devoted to the subject.

To start, let's look at profiling a camera for a specific illuminant. You take a photo of a known test chart under the lighting that you want to profile, and use the data in that image to create the profile. Note that the actual illuminant for the desired profile is used to illuminate the test chart while you take the photo, not a white light. My spectro (and I suspect all of the others) uses its calibrated internal white light source as the illuminant when scanning the targets. To get the same effect as the example shown in the video, it would be necessary to somehow change the spectro's internal light source to the custom illuminant that is to be profiled, and I know of no way to do that. I have never used Argyll, but you state "Furthermore, it doesn't require different charts to be printed and doesn't even require more measurements - one set of measurements can be used for profiles for any of the standard illuminants". Therefore, the only way for them to generate the non-D50 profiles is to apply the equivalent of a "Photoshop photo filter adjustment layer" to the scanned values, with all of the same limitations and pitfalls that you properly enumerated. In addition, we have seen that there is no typical "tungsten" or "fluorescent" light source, so we can only hope that they used your lighting to develop their filter.

Modifying the measured color values in this way is the specific step that I was objecting to in my post, as it doesn't follow accepted practice. I can see a math/programming wizard thinking that this would be a great feature, without even knowing that it isn't based on proper science. Do you remember a few years ago when scanner based profiling was all the rage? That approach also made some assumptions that didn't adhere to the rules, and seems to have fallen out of favor. Is there is a connection between following the rules and long term success?

The second point that I obviously didn't make clear was that there are very few people (outside of museum curators) who display their prints under a constant light source to justify a custom profile based on the lighting. If your lighting varies during the day what profile do you use? How would you display a painting with the same lighting? Is it time to replace photos with active displays that change photos every few seconds (wait - don't we already have them?).

It is very difficult for any printer using only CMYK inks to avoid color casts in the gray tones as the CMY inks are blended for the lighter tones and everything must be perfect for them to get true neutral colors. Isn't that why the current generation of high end printers typically include two gray inks so that the CMY inks aren't needed to generate the full range of gray tones? My 8 color (refilled dye) i9900 may not be perfect, but the grays are close enough to neutral that no one mentions seeing a color cast since I started profiling.

 

[RogerB]

Of course it not simple - if it were, color management would be easy to learn and we wouldn't have so many books devoted to the subject.

To start, let's look at profiling a camera for a specific illuminant. You take a photo of a known test chart under the lighting that you want to profile, and use the data in that image to create the profile. Note that the actual illuminant for the desired profile is used to illuminate the test chart while you take the photo, not a white light. My spectro (and I suspect all of the others) uses its calibrated internal white light source as the illuminant when scanning the targets. To get the same effect as the example shown in the video, it would be necessary to somehow change the spectro's internal light source to the custom illuminant that is to be profiled, and I know of no way to do that. I have never used Argyll, but you state "Furthermore, it doesn't require different charts to be printed and doesn't even require more measurements - one set of measurements can be used for profiles for any of the standard illuminants". Therefore, the only way for them to generate the non-D50 profiles is to apply the equivalent of a "Photoshop photo filter adjustment layer" to the scanned values, with all of the same limitations and pitfalls that you properly enumerated. In addition, we have seen that there is no typical "tungsten" or "fluorescent" light source, so we can only hope that they used your lighting to develop their filter.

Modifying the measured color values in this way is the specific step that I was objecting to in my post, as it doesn't follow accepted practice. I can see a math/programming wizard thinking that this would be a great feature, without even knowing that it isn't based on proper science. Do you remember a few years ago when scanner based profiling was all the rage? That approach also made some assumptions that didn't adhere to the rules, and seems to have fallen out of favor. Is there is a connection between following the rules and long term success?

The second point that I obviously didn't make clear was that there are very few people (outside of museum curators) who display their prints under a constant light source to justify a custom profile based on the lighting. If your lighting varies during the day what profile do you use? How would you display a painting with the same lighting? Is it time to replace photos with active displays that change photos every few seconds (wait - don't we already have them?).

It is very difficult for any printer using only CMYK inks to avoid color casts in the gray tones as the CMY inks are blended for the lighter tones and everything must be perfect for them to get true neutral colors. Isn't that why the current generation of high end printers typically include two gray inks so that the CMY inks aren't needed to generate the full range of gray tones? My 8 color (refilled dye) i9900 may not be perfect, but the grays are close enough to neutral that no one mentions seeing a color cast since I started profiling.

Measuring a printer test chart is very different from profiling a camera. Clearly a camera profile needs to use the actual light source, but when we measure the chart with a spectrophotometer we are measuring the reflectance of each patch at specific wavelengths. True the light source needs to be calibrated, but to generate a D50 profile it doesn't need to be even close to D50. The tungsten source in the i1Pro is closer to Illuminant A (2856K), and doesn't the ColorMunki use a white LED as its source? Once you have this detailed spectral data it's possible to calculate colour of each patch for any light source whose spectrum is accurately defined. The mathematical transforms for doing this are well established and are not just adding an adjustment layer.

If you use X-Rite's Measure Tool to measure a test chart using your i1Pro (with spectral measurements) you can actually choose the illuminant for which the results will be calculated. It's not just some software engineer's whim - it's the basis of a large part of colour science, and Graham Gill, the author of ArgyllCMS, knows how to use it .

As far as neutral greys go, I can't say anything about your printer. However I can say something about other printers that I've dealt with. For example, I made a (D50) profile for someone's Canon printer using OEM inks and he later told me that his B&W prints had a red cast. I asked him to print a 21-step grey wedge and it looked good to me - in daylight. I measured it (MeasureTool) using the D50 setting and sure enough it was very acceptable. However, when I used the Illuminant A setting it did confirm that there was a strong red cast. Here's how the two measurements came out.


And yes, the red cast was very visible under tungsten lighting, especially when placed next to a genuinely neutral grey card.

I'm not suggesting that illuminant-specific profiles are a must for everyone. At the same time I can say that I've seen prints with obvious red casts in camera club competitions where tungsten lighting was used. Maybe those photographers did need them.

 

[The Hat]

Ah guys doesn’t this say it all..

Of course it not simple - if it were, color management would be easy to learn and we wouldn't have so many books devoted to the subject.

You are both right and yet still cannot agree entirely on anything, just like pissing up a rope, where does it get you.
I suppose you could always write yet another book on the subject, but do you really think it would do any good.
It’s a colourful world out there so why not just enjoy it while we still can..

 

[mikling]

I decided to add a little sizzle like Colorbyte has done and made this advanced option available to the refiller and look what I caused!

For the user of my profiles, they are not likely to own stuff to sample the light source. As a result, if the D50 doesn't cut it, then you've got an alternative in two other directions that might be better. Neither of the three would be correct but one is likely to be closer to the desired result. Most often it is the D50; the standard but I've left a tool to use in the left and right backpockets......just in case.

For the readers not understanding this, even a "custom" profle will not help because it is the light source that is unique. So sometimes if a cast is observed, it is not necessarily the profile but it could be a result of the lighting that is used. You would not know until you brought it into a calibrated lightbox.

Indeed it tells how far this refill forum has progressed to be discussing such esoteric points of color.

We should be happy that we have had many knowledgeable posters who can debate this aspect to advance the knowledge of users.

 

[3dogs]

Thing I like about these "Esoteric" discussions is that mixed up in each and every one is a wealth of experience, so if a person who has an interest reads carefully there are almost always scraps of information that answer or just fill gaps in that persons overall understanding of a subject of great depth and span.


I just about peed my pants reading that Mike had the marketing "nouse" to offer alternate lightsource profiles. A punter might experiment and find one version gives better results.........and BINGO a hook is set home.......11/10 for smarts in marketing....love your work!

See my printing pattern has changed. Was a time I only printed a certain time of night under lights......fluro!
I, I ,) read in one of these discussions a vague reference to the lighting values when creating a profile, so busted out the Munki went to the advanced option, did an ambient light check and also took a lightmeter reading at the table where I have my Eizo set up. Then I did the same thing next morning and afternoon, a clear day then did the same thing on a dull overcast day all were measured as different (obviously) but remenber Columbus and the egg!!!

Result I have custom profiles that are selectable to suit. Thankfully no more colour cast problems... Very very SMART Mike!!
PS

I am over 50 so its OK ( the pants thing)

 

[stratman]

I just about peed my pants reading that Mike had the marketing "nouse" to offer alternate lightsource profiles. A punter might experiment and find one version gives better results.........and BINGO a hook is set home.......11/10 for smarts in marketing....love your work!

I hope I do not misunderstand your point. I think it is more than cheeky Munki marketing on Mikling's part.

He has a passion for printing that his customers may benefit from at no extra cost. Seeing how he is a prime Image Specialist ink e-tailer in this part of the world, why would he need to work so hard with something that he is not reimbursed, and, without those profiles he would still have his market share?

I have not had a vibe that Mikling was either slick Mad Men or cheesy Tin Men. Rather, Mikling is a genuine good guy who loves printers and printing and values his customers.

 

[Emulator]

Incidentally, Argyll wise, if anyone wishes to see what all this is about, Google "colprof" and read all the discussion in the "options", in particular under the -f flag on page 6.