Wednesday, September 24, 2014

A Guide to reflectance measurement devices, part 2

This blog post is a second or third or maybe fourth in a series of blog posts attempting to unbewilder the bewildering array of geometries for spectrophotometers. In the zero-eth post, I introduced the idea that reflected light comes in two forms: specular and bulk. That was background for the first post on spectrophotometers, where I discussed the 0/45 and 45/0 spectros. Then in a wild frenzy of blogomania, I followed up with a special article on measuring metallic inks.

Blogging my way to insanity

Today I look at one special case of 0/45 spectros, the polarized instruments. Stay tuned for the last section of the post, which is a public service announcement about the difference between process control and customer specs.

A problem unique to newsprint

Suppose your business is printing newspapers. You print on a rough, uncoated stock. You can't get a terribly high density - everyone knows that, and accepts that a black ink with a density of 1.2D on newsprint is pretty dark. Everyone assumes it's because the ink soaks into the paper. Actually, the larger effect is because you are seeing more of the specular component, but that's not the point of this blog post.

Here's a funny thing that everyone in the business of measuring the color of newspaper printing knows about: dryback. If you measure the density of ink hot off the press, and then again minutes or hours later, you will see a drop in the richness of the color. The density will drop by around 0.10D. They call this dryback.

Do you realize just how much dryback we have today!!!??

And here's a funny thing that few people in the business of measuring the color of newspaper printing know. If you were able to measure the density of ink on the press as it is running - not just "hot off the press", but "hot a few milliseconds after the ink hits the paper", you would be astounded at the amount of dryback there is. (I was astonished.)

A rich black ink might have a density of 1.10D when it just comes off the press. That same ink will dryback to maybe 1.03D. When it has just been put on the paper, the ink has a density approaching 2.00D. To put that in perspective, that number is higher than almost all printing of black ink on fancy-shmancy ultra-high quality paper.

Wow. Really?

Yes. Really. I was involved in the development of an newspaper color control system, and I have seen it myself with my own two sensors. When ink is first applied to the paper, the surface of the ink is very smooth. As a result, all of the specular light heads off at one angle, and a 0/45 spectrophotometer won't even notice it. As the ink dries, it conforms to the rough surface of the paper, and the spectrophotometer will start seeing the specular reflection.

What are we gonna do?!?!?

Dryback is troublesome for process control. But three guys who were working for Gretag came up with a solution. As the story goes, Tino Celio, Hans Ott, and Mast (I don't recall this last guy's first name) were sunning themselves at Malibu Beach. These guys were sipping umbrella drinks and talking about the sad state of affairs when it comes to measuring the color of ink on newsprint.

Celio (I think it was him) pointed out some attractive lady, commenting that it would take an instrument with a pretty small aperture to measure the color of that bikini. I am sure most everyone reading this column can relate to the situation... pointing out a hot babe or stud to a buddy. (I assume that most everyone in divorce court can relate to pointing out a hot babe or stud to their spouse.)

Ott said "What are you pointing at? All I see is glare!". Mast. always the clever one, pointed out that Celio was wearing polarized sunglasses, and Ott was not. When the two switched glasses, Ott said "Ahhh... I see her now." A light bulb suddenly appeared above all three heads.

Celio, Mast and Ott, celebrating their brilliant discovery

I may not have gotten the story quite right. I wasn't there at the time. Maybe it was the Riviera, since these guys were Europeans. I think they were from Switzerland or Uruguay or somewhere? Maybe they were drinking a good Bordeaux? Who knows?  All I am sure of is that one of these guys got this idea.

A pair of polarizing filters can be used to eliminate practically all the specular reflection that a 0/45 spectro sees. Bear in mind that bulk reflection is randomly polarized, but specular reflections remembers the polarization of the incident light. The following diagrams explain how we can take advantage of this fact to separate the bulk from the specular.

All tiny facet of the sample surface that are tilted at 22.5o
will direct specular light to the detector

A polarizer is added just after the light --
note that the specular reflection has the same polarization

Adding an s polarizer at the detector will eliminate this specular reflection

With these filters in place in your densitomoter, you can measure the sheets right as they come off the press, an hour later, or a week later. The polarized density won't change. A polarized densitometer is a great tool because it is immune to changes in gloss.

We have ourselves a process control tool! This tool has gained acceptance among densitometerophiles in Europe. But as good of an idea as this might be, it has never really caught on in the US. I suspect that the invention of the Atlantic Ocean has to do with this.

The definition of what goes into a polarized densitometer has been enshrined in ISO 13655, and polarized densitometers are referred to by the euphonious name "the M3 condition".


Naturally, you're gonna ask how polarized and non-polarized densities compare to one another. Surely there is a simple conversion, right?  As is often the case when I am asked a question, I have two answers: yes and no.

The plot shows measurements of a total of forty solid black patches, all of which were fully dried. Each patch was measured by a densitometer with and without polarization. The patches range in density from very light to very heavy, according to the pressman's subjective view. Ten of the patches were printed on a matte stock, ten on low gloss stock, ten on a medium gloss, and ten on a high gloss stock. The gloss of each was eyeballed by my very carefully calibrated eyeball.

The x axis of the plot is the density as measured without a polarizer.  The y axis is the amount that the density of the patch increases when measured with a polarized densitometer.

The red arrows illustrate the conversion from non-polarized to polarized on a matte stock. For that particular stock a density of 1.10D shows a difference of about 0.28D. That is, a 1.10D non-polarized density would be read as 1.38D polarized. Similarly, the blue arrows show how the density changes for a matte stock. A nonpolarized density of 2.00D is increased by only 0.06D when the polarizer is kicked in.

There are a couple of interesting things to note from this graph. First, for any particular stock and ink, there is a very simple transform between non-polarized and polarized. The difference between one and the other falls along a nice straight line. That's good news. There is a simple transform!

Second, the actual line for correction is highly dependent on the paper stock. For a very glossy stock, the correction is minimal; for a matte stock it is greater. That's bad news. The simple transform is not universal. To make matters worse, note that the medium gloss patches show the largest change. The patches on the low gloss stock have a conversion much more similar to the patches on the glossy stock.

Third, I will put to rest an old wive's tale. Density is kinda sorta linear with ink film thickness. In truth, the relationship flattens out as you go higher in density / ink film thickness. The tale that the old wife told me is that polarized density is much more better - polarized density maintains this linearity over a wider range of density.

The plot above shows this wives tale is just an old wive's tale. For any particular stock, there is a linear relationship between polarized and non-polarized density of dry ink, so any comments about linearity with ink film thickness that you can make about one holds equally true for the other.

Process control, or meeting customer requirements?

A great man once said that a polarized densitometer is a great tool because it is immune to changes in gloss. It gives you an indirect indication of the ink film thickness without confounding it with the gloss.

On the other hand, that same wise man is about to say that a polarized densitometer is a lousy tool because it only gives you an indication of what the sample looks like when it is viewed under polarized light while wearing polarized sunglasses with your head tilted properly. I haven't done any in-depth surveys, but I think that most print buyers and readers of newspapers don't look at the newspaper under this condition.

Using a polarized densitometer to monitor color
is like using this guy as an accountant

The annoying thing about customers is that they have this nasty habit of getting annoyed when the product is not what they wanted. Go figger. If we could only get rid of our customers, life would be so much easier. The print buyer (and end user) really and truly doesn't care if the "correct" amount of ink has been applied to the paper. If the color on the paper is not the color that they were expecting, then it's time for them to ask for rebates or go looking for another printer.

Therein lies a dilemma which is inherent to deciding on a color measurement instrument. Are you trying to do process control, or are you trying to measure the color?  Do you want your printing press to run predictably, or would you prefer to get the color that your customer wants?  Process control or customer satisfaction?

Process control often leads to meeting customer requirements. If the whole process is under control, then this is indeed the case. But, for anyone who has either dealt with customer complaints from the field, or who is married, it will be obvious that the best laid plans of mice and men oft go awry.   

ISO 12647, parts 2 and 3

And now for the public service announcement...

ISO 12647-2 is the standard when it comes to defining print. The purpose of this standard is to serve as a set of acceptance criteria for print. It is often cited as part of a contract for a printing job. Part 2 is about commercial web offset printing and part 3 is about cold set web offset (AKA newspaper) printing.

Since the inception of part 2 in 1996, this standard has been clear that quality monitoring is the thing, and not process control. (Well, at least for the solids.) Density (be it polarized or non-polarized) is not a reliable indicator of the color that you see, and should not be used as an acceptance criteria. All the colors of the solids and the solid overprints in the standard are specified in CIELAB, since CIELAB is the closest thing we have to our perception of color.

The standards are clear that densitometers are a useful tool for process control, internal to the printing plant. Generally, the printer establishes the density that will get to the proper CIELAB value with any particular substrate and ink combination, and will run to that. But ISO 12647-2 and -3 make it clear that the printer and print buyer should not converse in density when it comes to setting targets and tolerances. And as I have noted here, the printer and print buyer darn well better not even think about talking polarized density. There are just some things that are better left behind closed doors.


  1. Nice article, very informative. And it all makes (physical) sense- pretty good for a rogue mathematician.

  2. So John, What about the use of polarizing filters with spectrophotometers? You have a measure of color without the specular light?

    Max Derhak

  3. Yes Max... The same principle apply for densitometers and spectros. A polarized spectro would give you the spectrum of an object without the specular component, pretty much just the bulk reflection.

    As I said, this can be useful for process control, but I don;t see much use for it when it comes to identifying the "color" of an object. Do you see an application?

  4. Hi again,John
    You don't mention printing with colored inks on newsprint. That has got to be a total nightmare to 'color match' or even get close. Everything comes out pastel, I would think- of course, 'round these parts, the only color printing on newsprint is the Sunday comics, so unless Gary Trudeau or Scot Adams b*tches, who's to know?


  5. John's explanation of the truth (and light) is correct as usual. The biggest part of the problem was created by those European densitometer salesmen that assumed that their American audience was similar to their European counterparts; carefully trained in a rigorous apprenticeship, fully appreciative of the technicalities of their chosen trade, dedicated to always increasing their knowledge.

    The American audience went with what they THOUGHT they heard; "This here polar-icing desmotometer tells me what my print is gonna measure when it dries." And therefor they can compare old data (taken with NON-POLARIZING densitometers) to readings of wet samples. A PREDICTIVE device, no need to wait for dryback or estimate it with a rule of thumb.


    Polarization minimizes the effect of reduction of gloss due to scattering and veiling glare on DRY samples, making them measure AS IF THEY WERE WET. A reverse prediction if such a thing exists. A Time Machine?

    But this is the Graphic Arts and not the Graphic Sciences, and the densitometer salesmen went home happy.

    Larry Goldberg

  6. What you write at the end is very true. Our printers loved to use their Techkon densitometer but different circumstances led to identical density values having different colours as a result and that was not intended of course so we invested in a SpectroDrive combined with the software PressSIGN which does both at the same time: keeps colours in check using ISO 12647 as base and also using density to keep the printed sheets even across the board. It took a LOT of convincing to get the printers to give up their old densitometers (I actually had to hide them in a cupboard) but now, two years later, they'd never go back! (not to mention that density says nothing about dot gain)

  7. Given some substrate (gloss coated, for argument sake), is it then possible to linearly-relate M3 and M0 spectral reflectances? And vice-versa?

    1. Sort of. :)

      There was a method described by Kenji Imura of Konica Minolta. You purchase an FWA standard plaque from Avian, and measure it with the instrument you wish to use as a golden standard. This might be any M1 instrument, or it might be the bispectral instrument at the Canadian Standards lab. (I believe they are the only standards lab with one!)

      You then measure your standard FWA tile with your instrument using M1 and M2 (UV rich and UV excluded) illuminants. You then determine the linear combination of those two spectra that gives you the best match to the standard spectrum. That ratio will be used when you make subsequent measurements with your instrument, and the two will magically always agree.

      Or at least sort of. I am guessing that putting ink on the paper will slightly mess things up, at least for cyan and magenta inks.