Ink savings in offset printing: some historical context

“Is this you?” That was the question I got via e-mail recently, with a link to a 2011 article in the Polish print magazine Świat DRUKU. It was a message from Jinkai Qian from Techkon, one of my LinkedIn contacts, who wanted some more information on ICC profiles with a much lower TAC (total area coverage) than usual. And someone pointed him to that article, dating back to when I was the general manager of the innovation center for the printing industry. But this trip down memory lane didn’t just cover those lower TAC profiles, it also touched on some basics of innovation. There are some very important lessons you can – and should – draw from this story…

CONTENTS: The food packaging printer | Why? Because! | What are we looking at? | Looking at the numbers | Look beyond the numbers, the curves, the 3D graphs! | The real why… | Why is this important?
Originally published: 30/05/2026 – Last update: 04/06/2026

This story starts with a test one of my colleagues did, I guess in 2006, or 2007. He printed all kinds of CMYK color combinations on a web offset press, at the printing company he used to work for. After that test, he mentioned that the only way to get the deepest black when printing with four colors (CMYK) is to use 100% black (K) as part of the combination. If the K is below 100%, it will immediately become lighter. At that moment, all common ICC profiles topped the K between 95 or 98%, if I’m not mistaken, and a TAC between 320 and 360%. But as far as I can remember, he only used this new knowledge to create a test patch for print trials.

The food packaging printer
Fast-forward several years, to 2011. After a meeting on a different subject, a packaging printer who produced cardboard packages for the food industry (maybe even exclusively for frozen foods) asked me if I knew of any ‘fast-drying’ offset inks. He explained that until recently, he was printing with UV offset inks, but due to a contamination issue in a food product (in Italy, I think), the commonly used photoinitiators in UV inks were banned from food packaging… And with regular offset inks, it could take him up to a week before print jobs with high ink coverage were completely dry. And most of his customers had designs with high ink coverage…

I didn’t have an immediate answer for him. But when he mentioned those high-ink-consumption designs, I asked for more information. And it seemed that he was always printing in CMYK, no spot colors. And the profile they used had a 340 or 350% TAC, which was common in those days.

Why? Because!
So, I started asking around, with seasoned printers: why do we need such a high TAC? And the answer: because that’s what we need to get a deep black! But they couldn’t really dig deeper into the why, it’s just the way it is…

When checking the then-popular profiles in Chromix Color Think, I noticed that none reached 100% K, which reminded me of the test my colleague ran. And I wondered if making ICC profiles that do go to 100% K could make a difference. So I asked my colleagues to create a series of profiles, with a 20% decline in TAC. It started at 320% and went down to 180% TAC. Rediculously low according to common sense at that moment. But my take was: why not? If we are going to test this, why not figure out how low you can go before you will see a clear difference? And please note: before you will visually see a clear difference, not just a difference in measurements…

BTW: I also asked multiple people why the common profiles didn’t go to 100% K. Most didn’t have a clue, not even seasoned printers with a huge track record. Only one person (the technical director of a packaging printing company) said it might have to do with ‘picking‘ if you use a lower-quality board. But even he wasn’t sure about that, it was just a guess…

It took a few iterations to get the profiles right: several levels of GCR (grey component replacement) were used, some were too aggressive and showed artifacts when converting challenging images. I used my own set of test images, more on that later.

What are we looking at?
And then we got images printed… Several different images were printed, every one converted with different TACs, always including that 180%. And an important one: all images had a white border (1-2 cm wide).

The printing was done at a local industry training center, it was the lead instructor who did the printing. Upfront, I didn’t tell him what he was going to print, I just said that it needed to conform to the ISO 12647-2 specifications. When printing was finished, and I revealed what he had just done, he couldn’t believe it. From his point of view, those images all looked the same, they were all great. Was there really a 180% TAC in one of those images? Remember: this was the lead instructor of the training center, someone with expert knowledge.

A bit later, I showed those print samples to the board of directors: a commercial printer with an excellent track record of printing high-quality work, a large folding-carton packaging printer, and a printer of all kinds of specialty printing. Their first question: “What are we looking at?” They didn’t see a difference between the images and genuinely wondered what I was showing them. And that’s also key in this kind of study/test: I did not tell them upfront what they were going to see. If I had told them what it was, I could have influenced the result, they would have looked at it differently (this phenomenon is called ‘priming’ in behavioral economics). It was only when I told them these images had a different TAC, with one as low as 180%, that they were really surprised, they couldn’t believe they hadn’t immediately noticed a quality difference… And they started looking much closer, to figure out which was which.

That ‘priming’ can seriously influence tests and create bias in the results, undermining the validity. Let me show you two examples. In this  test, participants were asked if they noticed a color difference between a reference and 8 different print samples. But we had also included an identical copy as the reference in the test. And guess what: almost 1 out of 3 print professionals who participated in the test claimed to see a color difference between those identical samples. Why? One: they were asked whether they noticed a color difference, which prompted them to search for and see a color difference, even when none existed. Two: they were print professionals, working with color every day, so they HAD to see color differences, because they were professionals! Among the non-print professionals who participated in that test, the number of people reporting a color difference between the identical samples was lower: about 1 in 5.

Another example: in this survey on brand loyalty, I explicitly avoided revealing the main purpose of the survey. So I did not ask whether a color difference would lead to buying a different brand, but I asked whether ‘looking a bit different’ would influence buying decisions. ‘Looking a bit different’ is not only more generic, but it’s also more encompassing than ‘seeing a color difference’. And guess what: almost nobody said that ‘looking a bit different’ would influence their buying behavior. And when I repeated this question in a FOGRA Colour Management Café a few years ago. Nobody said that this would influence their buying behavior, and please note that a few people from X-Rite / Pantone were attending that Colour Management Café… 

What was very important in this visual evaluation was the white border around the images: if you would cut that border away and placed them on top of each other (as is often the case with press checks), you could see a difference, if there is enough light (like P1: 2000 lux). And you could measure a slight difference in the darkest black. But in normal viewing conditions, you wouldn’t notice, you wouldn’t see it. In the picture below, the prints are order from the highest TAC in the back (MC-C1) to the lowest TAC (MC-C3) on top.

And this is what is wrong in the printing industry, what I have been exposing for some time: we are holding the color of print jobs to an extremely high standard, which has no relation anymore to how print is looked at in real life. Visual print quality assessment is done on a well-lit (ISO 3664:2025 – P1: 2000 lux), evenly lit flat table with gray surroundings, while the real world, outside the print bubble, is 3-dimensional, no standard or even lighting, plus all kinds of colored objects nearby. And some color differences that can be measured and therefore expressed in numbers are not visible to the average viewer, certainly not under average viewing conditions. Beware of the professionals bias: it’s not because you can see a color difference with your trained eye and under that perfect lighting during an extensive press check, that an average consumer will notice it while shopping, in a cluthered, badly lit environment, in a fraction of a second.

But back to the TAC.

Eventually, we selected four of them for public release: 320%, 300%, 260%, and 220%. Two of them can still be found in the ICC Profile Registry. We also got an evaluation set printed, so people could judge the quality with their own eyes: three designs and four different TACs, printed on three different paper types. The prints shown above is one of these images.

On one occasion, during a small local print show, we showed this test set to people and asked them if they noticed a difference. One participant in that test, the technical director of a mid-sized printing company specialized in books with lots of photos, said that he would be very happy if all his reprints were this close… And added that when you would ask different printers to print the same job, the differences would be bigger. Nice!

The first one to receive the new profiles was the food packaging printer. He immediately started using the 260% profile as the default, and his drying issues were resolved… In some situations, with extremely heavy artwork, he even switched to the 220% profile, just to be safe. And the most important part: his customers never noticed… Plus, his press operators were happy: the stability on the press increased with these profiles, making their lives easier.

But, in those days, people didn’t believe you could go as low as 260% or even 220% and still get a decent print quality. That’s also the reason why we didn’t submit the 220% profile to the ICC profile registry. Eventually, not many people adopted the profiles at that time. People seemed addicted to a high TAC…

It was only about a decade later that software vendors started to embrace GCR and reduce TAC and ink consumption, eventually proving that my quest for lower TAC conversions was a valid idea.

Looking at numbers
I guess you might be anxious to look at the numbers, not just the pictures. And I will get to those in a minute, after this: too many people are looking at numbers without considering what these really mean, what these mean in real life, not just on a computer screen. On LinkedIn, I’ve heard brand owners demanding a 1 dE00. Why? Because that’s the lowest number before you get into decimals maybe? And what does that 1 dE00 mean in real life? And does that make any difference compared to a 2 dE00, or even a 4 dE00 difference in a real-life setting like a shop, which is, as I already mentioned, a 3-dimensional environment, not a flat table with perfect lighting?

And this is what, in my opinion, also happened with the new Pantone formulations: adding that 0,06% (yes, this is correct…) of a certain base ink might theoretically (i.e., in a 3D graph or a spectral curve) get the resulting color a tiny bit closer to the master data, but what’s the consequence in real life? Can you reliably mix a 0,06%, which equals 0,6 gram in 1 kilogram, e.g., with viscous offset inks? A friend of mine asked ink suppliers shortly after Pantone launched the new formulations, their answer was clear: you can’t. 

But let’s check the numbers. Fortunately, I did still have a copy of that set. Remember: it was printed in the fall of 2011, but it was kept in the dark all the time. The test set consisted of 3 images specifically selected for this test (e.g., chocolates: brown can be challenging to print in CMYK), with 4 different TAC settings, printed on 3 different paper types: gloss coated, machine coated, and uncoated.

I measured the darkest point in one of them (the picture I already shared above) with my Nix Spectro 2, always averaging 3 measurements: D50, M1, 2 °. 

These are the readings of the L-axis (a and b were always close to 0), and in the graph below, you can see the difference between the the 320% TAC and the lower percentages.

On the gloss coated version, the difference between the 320% TAC and the 220% TAC is only 1,1 dE00. That’s extremely low. And yes, it is something you can ‘see’ when you measure it, but visually you can only see it under the right conditions: enough light, and the colors need to touch or overlap. With a 1 cm white border around it, you won’t see it.

With the two other paper types, it’s around 2 dE00. But again, you will see that 2 dE00 when you measure it, but you will only notice a difference under very specific conditions: enough light (like viewing condition P1: 2000 lux), and even then, probably only when they are close enough. That 1 cm border works like magic!

Look beyond the numbers, the curves, the 3D graphs!
And even a smaller border will already make comparisons more difficult: when X-Rite launched the online version of the Farnsworth-Munsell 100 hue test, the colors touched, and the test was a piece of cake! They changed that quite fast and added a small black border, probably one or a few pixels wide, which already makes it much more difficult. Try it yourself! The physical version (which we had at the office) has a much larger black border around the colors, which inspired me to include a white border around the images with the different TACs.

BTW, when you look closely at packaging in a shop, you will notice there will always be some border: the curvature of the folded carton, a shadow effect of the curvature of a can of Coke, and the black of the space between the packages. Under no circumstances will they touch or be on top of each other, as in a press check. Nor will the lighting be perfect. Remember that. All of this has a significant impact on being able to detect color differences: it makes it much harder. The world outside the print bubble is quite different from the perfect and sterile environment of press checks. Even more: there is a disconnect between viewing conditions in print production and those in print consumption in the real world. Just check the image below, a random picture of a random shop, filled with that iconic Coca-Cola red.

There is also something else I need to mention here: the test images. You might know the Roman 16 test images. It’s a good set, but it’s not all-inclusive: some situations might not be revealed by those images. That’s why I tested the different iterations of the profiles with my own set of test images. Having moved to digital photography around 2005, I already had several thousand digital images by 2011. And I had already checked the rendering of some challenging ones with common ICC profiles in the past, which showed unusual renderings and artifacts.

With the packaging printer’s core business in mind, I also created several photos featuring chocolates: white, milk, and dark. This kind of brown can be very challenging when printing in CMYK (often, either the K is switched to brown or brown is added as a fifth color). OK, I admit, I also needed an excuse to go to the best chocolatier in Antwerp – DelRey – to buy a kilogram of chocolates. 😉

So, an important tip if you are a printer or repro house: keep challenging customer work at hand to test new software and hardware. This is the kind of work that you will reproduce in the future. Standard test sets are very helpful as a first check, but the real proof is the challenging work from your own customers that shows whether investments will fit your needs.

And a more recent experience highlights this: the publication of my ‘anniversary book’ for my 60th birthday. It’s a booklet with photos I made during the past decade, my 50s. Multiple pictures were taken during the last moments of the blue hour, when the sky is deep blue. I love that! But it’s challenging to get that printed: regular profiles turn that nice deep blue into purple… However, the XCMYK profile does not, as I discovered in another project: my photobook on the covid lockdown in 2020. So, I got that anniversarey book printed the same way, resulting in nice deep blues. I posted it on LinkedIn, with that specific information: having used XCMYK. One of the people commenting, was dr. Andreas Kraushaar, who is a well-known color researcher and head of the Prepress Technology Department at FOGRA. He suggested that I use the eciCMYK v2 profile for my next book, the European equivalent for XCMYK. I didn’t know that profile yet, so I downloaded it and checked it in Chromix ColorThink. If you look at the gamut, it’s quite similar to the XCMYK profile. Nice! Until I converted my cover photo to that profile: again a (slight) blue to purple shift… I assume the eciCMYK v2 profile was tested against the Roman 16 test set, and that this color shift wasn’t visible, or at least not noticed. This clearly shows why you need challenging images from real jobs to check new stuff. The 3D graphs and the numbers for the eciCMYKv2 profile looked great, but my first conversion already showed a flaw…

On the left is the original, in AdobeRGB. Top right is the conversion with the XCMYK profile, bottom right the conversion with the eciCMYK v2 profile, same settings. To see the full effect, you should test this kind of conversion yourself, on a calibrated monitor. This screenshot is an approximation. If you like to test it yourself, on a calibrated and capable monitor, here you can download the file.

As they say, the proof of the pudding is in the eating. It’s not in the recipe and ingredients, not even in a thorough theoretical analysis of the recipe and ingredients. That’s why you need to evaluate color transformations with real files, the most challenging customer files are the way to go.

The real why…
It was a few days after our first e-mail exchange that I remembered the real reason behind those high TAC profiles… And we have to go back a few decades in time, to a printing ‘industry’ that many – if not most – will not remember.

When I started in the printing industry in 1988, my first visit to a print shop was to a one-person shop. It was a very informative visit, and I spent at least half a day, maybe even more, with the owner. He did all kinds of small print jobs, including 4-color jobs. Which he printed on a 1-color press… Yes, in those days, that was certainly not exceptional. In the 1980’s, 4-color presses were very rare in sheetfed offset printing (added 01/06/2026). Most 4-color jobs in commercial printing (added 01/06/2026) were done on 2-color presses. For the record: most jobs in commercial printing (added 01/06/2026) weren’t 4-color in those days, usually either 1 color, or black plus a spot color. This was a very different printing industry from today’s. (added 31/05/2026: Please keep in mind that this is from my experience, which is commercial printing in Belgium. In other types of printing, or other geographical markets, this might have been a bit different.)

While I was there, he was doing a reprint of a small job, an invitation, maybe a few hundred copies, maximum 1000 copies. First, he printed the yellow, then washed his machine for a full ink change, and went on to the next color. Color control? That was purely visual. Except for his trained, experienced eyes, there was no measurement device in his shop.

To visually check every printed color, especially that yellow print, there has to be enough ‘image’ on the plate and on the paper. And that’s where it gets difficult when you apply GCR (grey component replacement), which replaces the overlapping CMY percentage, that would be viewed as grey, by K ink. There isn’t much image left: the percentages of Yellow will become very low and more difficult, if not impossible, to visually check. With Cyan and Magenta, it’s a bit easier to view, but still challenging.

So, that’s one reason why TAC – or better: the CMY percentages – used to be high: visual control is extremely challenging when you are printing the four colors one by one and CMY percentages are low.

Secondly, those were the days before computer-to-plate (CTP). The best-case scenario was that you would start with a full-sized film that was exposed via computer-to-film (CTF). Worst-case scenario, before CTF, you had separate typesetting and repro for the photos and artwork, and you had to tape everything onto a larger sheet of film… With CTF and the separate typesetting/repro, you needed to convert the images on the film to the plate. Which wasn’t an exact process, especially the smaller dots were influenced by this process. Sometimes those smaller dots even disappeared, e.g., when there was still a tiny bit of air left between the film and the plate. And that’s the second reason why GCR, with tiny dots in C, M, and Y, wasn’t popular: color dots would disappear… Which could result in color shifts.

As example you can see the Yellow plate of the test image with chocolates, left the 220% TAC version, right the 320% TAC version.

And here’s the important lesson: with CTP, everything changed. With CTP, you can reliably expose small dots onto a plate. Next to that, with measurement devices, you can objectively check the printed image (e.g., density control of the solids). With 4-color presses, you can immediately check the final result. So all the reasons why GCR was not a good idea in the past had disappeared. But old habits die hard. And nobody took advantage of these important changes in printing to recheck those old habits, those old truths.

I don’t think I already knew the real why when we launched those low TAC ICC profiles. In any case, this information certainly wasn’t in the press release. We only explained that a lower TAC was possible, not that the old reasons to go for a high TAC had disappeared when moving to CTP, measurement devices and 4-color presses. If I had included that in the press release, maybe more people might have considered looking into those profiles.

BTW: digital printing again changed a lot, showing that you can’t immediately transfer old truths from offset (or flexo) to digital printing. You should recheck the very basics of every part of the flow. And that’s an important lesson on innovation too… E.g., when Océ (now Canon) launched their 7-color CPS-700 digital press, everybody immediately assumed it had a wide gamut, with those 7 colors. Which was not the case: it needed those 7 colors to be able to produce the colors in the standard 4 color offset gamut.

Why is this important?
Old habits die hard, even when the context has changed. Even when the core of that old habit has disappeared, when steps or elements of a workflow (from photography and design to print finishing) undergo a fundamental change or even disappear, people will still work the old way. However, at those moments, the complete flow and its requirements should be revisited.

And innovation doesn’t happen overnight. Even if you make a solid improvement, there can still be a lot of hurdles to overcome, one of which is convincing people to change their behavior and ditch old truths and habits. Even with convincing visual proofs, it took about a decade before the printing industry started taking ink reduction seriously. But I’m happy that this finally happened.

If you are doing test, surveys or research: never forget that participants can very easily be influenced and create a bias. That priming is real: if you ask people if they see a color difference, they will look for one. And might even find one when there is none…

Related posts:

Deep blues, with just 4 colors? Yes, it can be done! Not all black is designed equal Designing black for food packaging (not all black is designed equal…)