Tuesday, October 28, 2014

What measurement condition is your spectro wearing?

These days, all the fashionable spectrophotometers are sporting the new measurement condition, M1. It's all the rage from Alabama to Aukland. If your spectro hasn't adopted this new look, then, sorry. It just won't get invited to the parties with all the cool spectros.

But what is this hip new fashion trend, and why should you care? This blog post goes undercover to get the inside story about the new measurement conditions in ISO 13655. This reporter investigates the four measurement conditions (M1 through M3), but more importantly, explains why you would choose one over the other.

Summary of the measurement conditions

Way back in 2009, when all of our spectros were wearing styles appropriate to that long-lost era, the ISO Technical Committee 130 came out with a new fashion edict. Spectral measurements henceforth shall be made according to one of four types of illumination, and the one chosen shall be reported along with the data.

(Note that the word "shall" is standards-speak for "you gotta do this if you want to comply with the standard. The other key word is "should", which means "as experts, we recommend doing this, but it is not a requirement for compliance".)

Why all the fuss? The big driving force behind this is the proliferation of Fluorescent Whitening Agents (FWAs) in paper. These are substances (stilbenes, for the chemists and lingo-philes in the crowd) that absorb ultraviolet light and re-emit that energy as blue light. This makes paper look whiter - which is what everyone wants and craves.

I should make a note here. The popular media (Fox News and MSNBC) call them "OBAs", meaning "Optical Whitening Agents". While this moniker is correct - stilbene does make paper brighter in an optical kind of way - I would say that so does TiO2 and bleach. The term "OBA" fails to emphasize the key operative, which is fluorescent light.

I have already written a bit about the basic problem and the reponse to it. The following exciting articles focus on the "M1" stuff.
Layman's Guide to ISO Print Standards

Here are the four exciting choices for measurement conditions:

 Most handheld spectros use an incandescent bulb to illuminate the sample. Remember those kinda lights? A little piece of wire called a "filament"? Some electricity going through it? And then the wire gets hot and glows. Guess what? This light source doesn't have all that much UV content. And guess what again? The amount of UV varies a lot from one instrument to the next.

M0 is based on a hypothetical incandescent light source. For an M0 measurement, the light hitting the sample should (note the word) conform to CIE standard illuminant A - which is to say, a light bulb. The word "should" is important and intentional. This little tiny loophole allows anyone to use the older spectros and remain compliant. This would be a totally dumb idea, but you could use a lightning bug with a hangover as the light source for your M0 illuminant and still be compliant. No one's gonna check. M0 is the nightclub that any spectro can enter.

If you want to find out if your local neighborhood color scientist is hip, just ask whether he or she is raving about M1. All the hip ones will say M1. The illumination for this measurement condition is based on a theoretical daylight called D50. This puppy packs a pretty good wallop of UV content, so this will excite those ol' OBAs, if you know what I mean.

Does your spectro want to strut its stuff at the M1 Bistro? The security guard is checking IDs at the door, and any spectro failing to provide the proper levels of OBA ain't gonna be ordering an avocado-tini at this joint.

OBAs love attention. The only reason they show up at parties is to be seen. But when the spectrophotometer breaks out it's M2 light source, the OBAs become as invisible as that woman who married Jimmy Fallon. An M2 light source will be almost kinda completely devoid of UV light.

How do you get into this club?  If your spectro comes knocking at the door, there will be a test, and I'm, not talking "written exam". The test is described in Annex H. To perform the test, you need to get pretty intimate with the instrument. It would make a TSA agent blush.

BTW, the TAGA 2015 conference in Albuquerque (March 22-25, 2015) will feature a micro-conference on OBAs. We had a whole bunch of papers on the subject, so the VP of papers decided to talk it up big. Here I am, talking it up big.

M2 might seriously have been considered a contender for preferred condition. It does level the playing field when it comes to UV excitation. All M2 instruments have no UV, so none of them should cause OBAs to glow. And they should all agree.

But, to make this whole opera work, the light booths really should have the same amount of UV content. And if there were no UV content in the viewing booth, then papers with OBAs would look dingy. And no one wants that.

You may think that M2 is an exclusive club, but M3 is even more exclusive. First you need to get into the M2 club even to be able to bribe the bouncer to get into M3. And you gotta be wearing sunglasses. Not just any sunglasses, but sunglasses with polarizing filters. As a result, specular highlights and fluorescence are not allowed in M3 disco.

So... if you happen to be thinking about dating a spectro, how do you decide which one is right for you?

Do you have OBAs?

I know this probably sounds personal, but the first concern is whether you have OBAs in the house. If you don't have OBAs, then (theoretically), there should be no difference between M0, M1, and M2 measurements. You could use your older instrument that doesn't offer a choice, or you could go with one of the newer instruments that offer M1.

Do I smell OBAs?

That question "do you have OBAs?" is pretty much the same as "are you measuring paper?" The thing is, paper is normally brown - the color of a grocery bag. Special processing must be done to make paper white. That could involve bleaching, or it could involve OBAs. Today, almost all paper uses at least some of the latter.

So if you happen to be measuring ink on paper, then you just gotta plunk your money down for an M1 instrument, because you will find your old M0 instrument disagreeable. It will disagree with your light booth, with M1 instruments, and even with M0 instruments from other families. The thing is that pretty much all paper for commercial printing will have OBAs added, so you have a choice as to whether to deal with a petulant teenager, so buy into the M1 craze.

I should add that switching over to the M1 instrument is only part of the change that you are facing. When you make the change, you will find that you need to make sure that your lighting booth adheres to the 2009 version of ISO 3664, so it has the same amount of UV content as the spectro. And all the data that you had previously measured, like profiles of your press and target colors, will need to be updated. I know, not a simple solution. 

That covers web offset, sheetfed, and newspaper printing. The situation in packaging is a bit more complicated, so let me describe some cases. Let's say that you are measuring color on foils or poly. I might be wrong here, but I don't think you will run into any OBAs. Foils and poly and the floodcoat use something like TiO2 to add whiteness and opacity. Floodcoats wouldn't benefit by OBAs, since they are not brown. So, like I said, I could be wrong, but I think it would be silly potatoes to add OBAs to foils.

If you are measuring kraft paper, then I think you are likely free and clear as well. If the paper is brown, then clearly no one cared enough to take the time to add OBAs to make it white. Then again, if the kraft paper includes a certain amount of recycled paper, then the OBAs might be sneaking in through the back door.

How about printing on white cardboard or card stock? Now things get uncertain. You gotta ask yourself, how did the cardboard get white?  Sometimes, cardboard is made white by applying a floodcoat of white stuff. This white stuff might be something like titanium dioxide, which is naturally white. So, it is unlikely to contain OBAs. Then again, cardboard may be white because it has been laminated with paper. If that paper is white, then you can guess that it has OBAs. Another possibility is that the white card stock might be white because the paper is white. So, it could have OBAs.

When in doubt... I would recommend having a UV light source around. Anyone who survived the sixties is familiar with these. Today you can get UV flashlights made from LEDs. Cheap and convenient, and a good way to test for the presence of OBAs.

Do you want CIELAB values?

The second concern is whether you want CIELAB values. In my not-always-humble opinion, computing CIELAB values from a polarized instrument (that is, M3) is just plain silly. The whole point of CIELAB is to emulate how our eyes see color. Unless your product is destined to be viewed by people wearing polarized sunglasses under light that is polarized the other way, then CIELAB is probably the wrong choice.

ISO 13655:2009 agrees with me on this one. Here is a quote from Annex G:

"Notwithstanding the beneficial effects of crossed polarizers [M3] in the special cases mentioned above it needs to be noted that for most other instances in colorimetry the use of polarizers is counterproductive."

So, if you want to compute CIELAB values, then you must use M0, M1, or M2.

Are you doing process control on a cold set press? 

M3 does not mix with CIELAB, but it does mix with density. ISO 13655 has this to say about where the M3 condition might be applicable:

"It was discovered that the installation of crossed polarizing filters serves to extend the linear part of the density versus ink film thickness dependence towards higher values, and serves to greatly reduce density dry-back."

There are two points here. The first is that M3 "extends the linear relationship between density and ink film thickness." This is believed by many, but it is unfortunately incorrect. In my blog post on polarized densitometers I presented a plot showing that there is a linear relationship between polarized and unpolarized density. If polarized density is linear with ink film thickness, then unpolarized must be as well.

The second point from the ISO 13655 quote is that M3 reduces dry-back. Measurements made directly after printing on a cold set press will not change as the ink dries. This is true, and that was really the whole point of the blog post on polarized densitometers. So... I won't belabor the point here. But I will encourage you to go read the post.

And I will also reiterate a point that I have iterated a couple three times in previous blog posts. There is a difference between "process control" and "satisfying-your-customer control". The first is about making sure your process is appropriate and repeatable. The second is about making sure that the payments from your customer are appropriate and repeatable. Density and M3 are process control parameters. CIELAB is a product expectation control parameter.

So, M3 can be useful when you are measuring ink that is not quite dry, but its use should be limited to within a given plant. No interchanging of data, you hear?

Are you doing process control on materials with OBAs?

Just in case you have been just too enthralled with this page-turner of a blog post to have been keeping track...

M0 is ok if you have a legacy instrument, and you aren't really that into OBAs.
M1 is preferred, especially if you might think you have a little issue with OBAs.
M3 is acceptable, but only for process control - no CIELAB allowed.

What about M2?

OBAs are tricky little devils. They make the paper look whiter. But when you apply ink to them, a funny thing happens. The ink blocks the UV, so that the paper under the ink does not get artificially brightened. This can make things a little weird if you are a process control freak. Especially when you come upon an ink that doesn't happen to block the OBAs. The relationship between the density of the paper and the density of the solid ink gets befuddled.

That last paragraph was written yesterday. This morning, I looked at some data that I got from my good buddy, Gerry Gerlach. His data refutes the stupid statement that I made "The ink blocks the UV." Good God! What was I thinking. In his data, cyan, magenta, and black all do a pretty good job of blocking the UV. But for yellow ink (and aqueous coating on bare paper) there is a large difference in b* between the M1 and M2 measurements. In other words, the ink and coating are letting the UV light through, i.e. are transparent in the UV. I suspect that not all yellow inks do that, and certainly clear coatings may differ. 

So, M2 is a process control thing, maybe better than M1. If you are trying to establish that you are putting a consistent amount of pigment on the paper from day to day, this might be a good thing to try. But as with M3, I caution that this is not the same as making the correct color.

M2 has found another purpose in life just recently with the invention of the OBA index of a paper. It has been noted that OBAs will tend to decrease the b* of a paper. A substrate might measure a little yellowish under M2, maybe b* is +2. If you measure that same substrate under M1, the b* might go negative, maybe -3. The OBA is the difference between the M2 and M1 measurements. In this example, the OBA index would be 5.

Are you sufficiently confused?

I hope this has blog post on illumination conditions been enlightening, no pun intended. Actually, the pun was intended. But the enlightenment was also intended.


  1. Hi John,

    Another good post.

    I have to differ on the use of M3 though. In building profiles for inkjet printers on water color papers, canvas (especially glossy canvas), and other specialty papers, M3 makes a HUGE difference.

    If the various surface reflections aren't knocked down then the profiles tend to lose a lot of detail in the shadows. Slap a polarizer on and suddenly you're getting "pro photographer" results instead of just good "color matches". It reminds me of the difference a good drum scanner could get out of a full-range transparency.

    And while soft-proofing using the profile is a bit optimistic in the density and saturation of dark colors, it's not too bad either. It still seems somewhat counter-intuitive to me but we've been building profiles for people around the world using this technique and the years of happy picky people have proven the point to me.


    Steve Upton

  2. Hello Steve,

    I'm glad that the blog is appreciated. I also appreciate comments, especially from sharp people like yourself.

    Let me offer a possible explanation for what you have seen... feel free to tell me I am full of beans.

    On a glossy canvas, you can get very high densities, first, since it is glossy and there is little contamination with surface reflectance, and second, because it has surface texture which can trap some of the light. We also know that people tend to prefer images that are richer than real life... more saturated colors and higher densities. And of course, polarized spectros naturally give you higher densities and richer colors than non-polarized.

    Putting those two together, we have a substrate that is able to make a print that is preferable to the "correct" color, coupled with a measurement technology that "artificially" gives the profile more punch.

    Is this a plausible explanation?


  3. "Is this a plausible explanation?"

    Could be... but not in this situation.

    What I'm referring to is a (sometimes) startling difference between plugged shadows and shadows will lots of detail and differentiation. So the higher density is nice, but that would be available in either scenario. With M3 (pol) we get the higher density and the smoother tone range as well.

    Also, this is not just limited to glossy canvas - though that might be the most noticeable case. We've seen great results on simple watercolor papers such as Arches.

    My initial concern was not that the images would be richer than accurate, but in fact the opposite. I was thinking that the system would think for instance "wow, I can get really saturated browns. I'll need less ink to get the requested color" and it would appear unsaturated to the unpolarized viewer. In practice, however, it doesn't behave that way. It's something I don't have my head fully around yet. But it works.



  4. John - I do not think that you make the point strongly enough that M0 and M2 are not equivalent. Thus M0 and M1 are not so incompatible as you state. Many converters and print buyers have been converting and buying print using M0 for a long time with good success.

    So if you happen to be measuring ink on paper, then you just gotta plunk your money down for an M1 instrument, because you will find your old M0 instrument disagreeable. It will disagree with your light booth, with M1 instruments, and even with M0 instruments from other families. The thing is that pretty much all paper for commercial printing will have OBAs added, so you have a choice as to whether to deal with a petulant teenager, so buy into the M1 craze

    The key to success in characterizing objects with fluorescent colorants is consistency. M0 is very consistent, if it is truly based on an incandescent lamp of known color temperature. The amount of UV to VIS in its spectrum can be predicted. Not so lucky with M2 or even M3 where the amount of radiance removed from the shorter wavelengths is a function of the optical system and materials. But, if you have viewing cabinets equipped with a lamps that produce a UV to VIS ratio that is in closer conformance to the D50 spectral power distirbution then M0 will consistently under excite the OBA and the radiance factor used to produce CIELAB values will have b* values that do not correlate well with the visual examination. The advantage of M1 is that it wil produce radiance factor curves and by tristimulus integration CIELAB coordinates that are more consistent with visual assessments according to ISO 3664.

    Case in point - how do you describe an instrument system with a white LED with a color temperature of 2850 K? Is this M0 or not?

  5. Anonymous - You have pointed out the need for another use case: If viewing booths are not part of the workflow, and if the same make and model of spectro is used throughout the entire supply chain from designer to prepress to press to print buyer, then M0 will work fine, regardless of OBA content. Coca Cola has mandated just this.

    You said that M0 is very consistent... As I think we agree, there is an issue with the fact that viewing booths will excite the OBAs differently than an M0 instrument. You alluded to the fact that not all M0 instruments agree, but I am not sure that this is widely understood. So let me pontificate on that...

    ISO 13655 defines "M0", and recommends that the illumination look like an incandescent light bulb with color temperature of 2856K. But his is only a recommendation. The standard deliberately uses the word "should", rather than "shall".

    I recall this choice of words being deliberated at some length in standards meetings. In the end, M0 became the catch-all that allows continued use of any of the spectros currently being used. But, there are no mandatory requirements to qualify for M0.

    Is a white LED instrument compliant with M0? Yes. How about an instrument with a xenon strobe, which may put a lot of UV? Yes. An instrument that uses a bottle of fireflies for illumination? Yes. I have just filed a patent for the firefly design, so it must be a great idea.

    Next question... why weren't some mandatory requirements for M0 put into 13655? If virtually every handheld device (at the time) used a light bulb, then why couldn't the standard just require adherence to 2856K to within some tolerance?

    The answer is simple and practical. Different existing instruments that use an incandescent bulb had vastly different amounts of UV light.

    Why is there a variation in the amount of UV? There are a number of reasons why engineers may have made choices that effect the amount of UV.

    There are advantages and disadvantages to running a bulb above or below 2856K. Running it higher ("over voltage") gives you more light, especially at the blue end, and hence better noise characteristics, but it drastically shortens bulb life. So, different engineers made different tradeoffs. The higher the bulb temperature, the higher the ratio of UV light.

    Glass and some clear plastics absorb UV - it is hard to get a sunburn through a window. The number of pieces of glass/plastic and the thickness of them will alter the amount of UV.

    Incandescent light is weak at the blue end. Silicon detectors (which are in most if not all handheld devices) are also very insensitive at the blue end. Some spectros add a light cyan filter to dampen the red and somewhat dampen the green.

    And of course... when the latest revisions of 13655 were on the table, LEDs were just starting to be considered as a light source in spectros. White LEDS (as well as any other LEDs that provide visible light) put out virtually no UV.

    You said that M2 and M3 are not necessarily consistent between models. I am going to disagree and agree with that statement. M2 was envisioned to be the UV cut version - it requires virtually no OBA excitation. M3 requires M2 compliance as well as the polarizing filters. So I disagree.

    On the other hand, while there is a "shall" in the definition, it is a fuzzy "shall". It says: "[M2] shall only contain substantial radiation power in the wavelength range above 400 nm." There is no definition of what the word substantial means, and not explanation of how to test for this.

    Also, the standard indicates that measurement at 400 nm and 410 nm are desirable, so there must be light available at these wavelengths. This light will also excite the OBAs. So, I agree with your statement.

  6. I've been shopping around for a spectro for dye sublimated apparel. I've been going between the Barbieri LFP (No M1) and Spectropad (M1). The latter has the newer tech but the former does a lot more (multiple aperture sizes, multiple patch reads) that works for meshes and different textures in fabric. I toss in some charts printed on poly/spandex fabric in our Spectralight III and the white glows under pure UV. In fluorescent illumination and Illuminant A, I toggle the UV and definitely notice the lighter/brighter colors reacting to the UV in those conditions.

    What's curious is that we have D65 glass in front of some halogen bulbs for 'daylight' illumination, and the UV toggle really doesn't have a effect when I toggle it under that condition. Is it because the D65 with Halogen already has UV in it?

    Are there FWA/OBA's in the fabric? I think so if its reacting to the UV in fluorescent light and Illuminant A. Which leads me to think I would be missing out without an M1 spectro.

    Any thoughts comments? I could use some insight since not much technical nitty gritty goes on regarding dye sub apparel.


  7. Alex, Thanks for the questions... you are learning me about textiles!

    I am not sure why the UV toggle should have no effect under D65, but I have a guess. The UV cut filter will remove most all of the UV. If it is having no effect, then it must be that the D65 glass has already taken out all the UV. I am not sure why they would do that, or even if they did, but prior to 2009 with the definition of M1, there were no official requirements for "D65", so it would have been acceptable.

    Do fabrics have OBAs? Other than the fact that I wear clothes most every day, and the fact that I lived through the 60's and the black light phase with paintings of Elvis on black velvet, I don't have a lot of experience with OBAs and textiles. But I have heard from various reliable sources (or at least, people who have told me that they are reliable) that fabric often does have OBAs. I know for sure that most laundry detergents have "bluing agents", which are OBAs.

    In general, you have two choices: Either you a) go to M1 so that you have a chance for various make and model of spectros throughout your supply chain to work and for measurements to agree with visual assessment in a light booth, or b) make sure that all the instruments in your supply chain are the same make and model.
    For more information, I direct you to my blog on measuring fluorescent materials:

    As for the measurement issues raised by the fact that some textiles have texture, there are two solutions. One is to use a big aperture and/or to average multiple measurements, such as you have done. This will reduce the variability. Another option is to use a spherical instrument (also called d/8 or 8/d). Since this collects light that bounces off in all directions, it is less susceptible to this variation.

    On matte fabrics, I have seen that the two instruments will measure reasonably close. On fabrics that have some sheen to them, there is a difference even when you do averaging with the 45/0 instruments. The spherical instruments will read lighter in color, and will probably not correlated so well with what we see when we are measuring dark of richly color fabrics that have some sheen to them.

    For a bit more information her, I point you to another blog post:

  8. John,

    On the spectralight, the UV bulbs are not behind the D65 glass.

    I went back under the hood and noticed that there is a pair of long UV bulbs that activate only for the non-daylight settings, and then a separate skinny 8" bulb that comes on for the 'daylight' (D65). After looking up tungsten-halogen lamps (http://en.wikipedia.org/wiki/Halogen_lamp) it looks like these bulbs emit some UV anyway, so they might be adding the smaller UV lamp to supplement the halogen lamp's UV. Following that notion, I would conclude that the 'daylight' setting on the SPLIII is already exciting the optical brighteners which is why there is no significant toggle effect with the smaller UV bulb.

    The fabric definitely fluoresces under black light, but I've got my rep contacting the mill to confirm for OBAs.

    Aside from supply chain, my main focus is on understanding the color gamut and color accuracy of a D50 M1 spectro over an Illuminant A spectro. My 'tentative understanding' is that greater excitation of those optical brighteners with the D50 will feed more color data into my profile over a standard Illum.A spectro. Does getting a columbia blue to 'pop' more by having UV when profiling (M1) make for a better/wider/truer profile? Does that columbia blue sublimated football pant look better in a bowl game when profiled with M1 D50 spectro vs M0 Tungsten?

    Thanks for the links. I'm glad to have found this blog.

  9. Alex,

    Yes, a halogen bulb will out out some UV... hence the need for a UV cut filter on the pre-2009 instruments. Not a lot, and not as much as D65.

    Will your profiles be more accurate with M1? I can say that an M1 profile will do better at predicting the appearance of the pants in the stadium, if daylight is the main illuminant on the field. If the game is nighttime and being played under mercury vapor lights, then I don't know!

    Feel free to contact me at john@johnthemathguy.com if have more specific questions.

  10. It’s a common practice in folding carton packaging printing on SBS paperboard to have florescent dyes added to aqueous or UV coating so knockouts for glue flaps can be seen better under a black light for QC process control.

    Steve Suffoletto

  11. Thanks for the comment, Steve. An important question... do the fluorescent dyes have the same behavior (excitation and emission bands) as the stilbene used as an OBA in paper?

  12. Thank you so much for the information generously provided. I'm also "knee-deep" into M1 measurements. As I was reading your comments above about Coca-Cola, I could not help conclude that the problem with which numbers everyone in the supply chain gets (M0 type of numbers since that's still the majority of instruments anyone in the chain is still having access to) but this does nothing for managing the color appearance. In effect, as I read your note, Coca-Cola is effectively washing their hands of it. Meaning that, the visual results does not matter all that much to them, it is more important that everybody involved in the production of their prints get the "number" comparable -- never mind what the thing looks like.

    The other comment I want to make is the bogus claim that light from booths "shall" or "should" have a higher UV-content than before and, therefore, when M1 measurements (whatever that means) are used, visual appearance will correlate better with M1 measurements?

    It is true that, slowly but surely, as printers replace their aging pre-2009 fluorescent bulbs in their overhead luminaires, the new ones they get from GTI ("E" series) and JNL are supposed to have a higher UV content. There is the revised UV-Metamerism index that has to be calculated to prove conformance but who has instruments that are sensitive enough below 400nm to reliably measure the amount of UV? The EyeOnePro which most folks have at their disposal A) are insensitive in that part of the spectrum, and B) offer too wide of a bandwidth to be useful for ISO-3664:2009 measurements.

    So, who is actually checking the actual content of their overhead luminaire at press consoles? The FD-7 has an irradiance mode that seem to measure down to 360nm… I'm still unsure whether that's enough to do the test…

    So, where does that take us?

    Minolta FD-xx seems to be doing the right thing with its bi-spectral reflectance measurements but does that actually result in "true" D50-like illumination? I'm still not 100% clear on their technology but I suspect they're on to something that X-Rite is not…

    / Roger Breton


  13. nice article, using great equipment to measure the thickness of your paint will produce you with good information about your stuffs.


    Paint Spray Gun