Couleur locale – the search for color in carbon

Technically, carbon transfer printing isn’t ‘carbon’ anymore once it becomes color. Yes, for the black/key layer (if used), a carbon pigment is generally used. But for the other colors, evidently carbon falls short. So something like ‘pigmented gelatin printing’ is more accurate. In any case, as color enters, it brings many questions, issues and concerns. Lots of fun stuff, as well. How about this one: which colors should we go for in the first place? Let’s have a look and see if we can make at least a first step.

The elephant in the room, let’s start with that. It’s big. And it’s pink. Or at least, it’s supposed to be, but it doesn’t quite render the correct shade of pink. Cue a drum roll, because here are my first ever color carbon prints, and you’ll immediately recognize the need for this blog post:

The first color checker torture test. Torture succeeded; patient deceased. Scan of print to the left, digital file right.

The torture test appears to work as intended. It’s pure torture, horror and Spanish inquisition all over. Well, it’s certainly clear that there are some, ehm, minor issues to be ironed out. But given this dramatic success, I thought I’d try an actual photograph. Truth be told, I used my second set of color tissues to make the color checker and had exactly one set left I didn’t really need for other tests, so I figured, why not try an actual print and see how bad it comes out. Well, pretty bad, I can tell you that! Here, look:

First ever color carbon photograph I printed. Scan of print left, digital file right.

You know, I’m actually quite happy with it! I mean, it’s sort of recognizable as the same image. It also materialized without too many difficulties; there were no blatant transfer issues, and the printing process was fairly swift. I made the actual prints within an hour or so, not counting the tissue drying time and the time to transfer each individual print onto the final support. I might do an additional blog post on how I am speeding up the printing process; all this testing would be too tedious if it took me a full day to do one set of separations.

There are, of course, pretty blatant color problems here. The banana leaf in the middle (that’s what it is, in case you hadn’t recognized it) suffers from a severe case of magentitis (I made that up, sorry): way too much magenta. But in my previous post, I already noted the apparent balance issue with too much magenta in the color checker, so no surprise it also materialized here. It was kind of a surprise how bad the issue was, though! Wow. Then there are all sorts of curve problems, resulting in an odd contrast range. We’ll settle that, at some point, I think. And evidently, there’s a bit of a problem with the primaries. I included the primary color gradients for testing purposes, but they also illustrate an inconvenient reality here: the real world ‘magenta’ looks nothing like the digital magenta, and the same goes for all the others. Even black isn’t actual black, but more of a very deep chocolate brown.

There are two things I could do at this point – and I’ll need to do both. One: adjust the pigment set so that the colors are chosen in an intelligent fashion. We’ll get into that shortly. Two: I need some kind of profiling and soft-proofing approach so that what I make digitally, is translated to color separations in such a way that it matches the (in)capabilities of the printing process. I haven’t tackled #2 yet, although I might get lucky and be in the position to borrow an actual profiler soon!

Back to #1: the pigment set. I had never given pigments much thought, but now I have to. And it’s fun, actually. First let’s see what we’re trying to do.

Green alien: yellow and blue make Martians

This is the basis: take two pigments and blend them to make other colors. Since this is printing space, we generally work with 3+1 pigments: cyan, magenta, yellow and added black for neutral, deep blacks. There are of course alternatives and extensions to the process, but let’s keep it at this for a moment. Additional spot colors are great, but something I’ll sidestep until I have a firmer grip on the basics. In theory, the combination of three colors that are sufficiently far apart from each other on the color wheel should be able to make all imaginable hues by blending them. Right? Well, not quite. Here’s that color wheel (the CIELAB version at least):

CIELAB color wheel; numbers represent hue angle in degrees

In the circle above, all hues are represented and I’ve indicated the position of the primaries Cyan, Magenta and Yellow (at 225, 315 and 90 degrees). You might also notice that there’s considerable distance to the outside diameter of the circle and its center. In other words: it’s actually a circle, not a dot. The distance from the center of the circle to any point within the circle is the chroma of a hue: it indicates how saturated it is. Hues very close to the center have a low chroma and are closer to the greyscale than to the colorful outside rim. Hues that are closer to the outside rim are more saturated and colorful. This is also why I put the primaries on the outside rim: they’re at the highest theoretical level of chroma – they’re after all mostly theoretical constructs anyway. So these two characteristics, hue (which color is it?) and chroma (how saturated is this color?) turn out to be important considerations in which pigments we’ll choose for our printing process.

This immediately brings us to a realization: those theoretical values of Cyan, Magenta and are nice, but how do they relate to our real-world colors? Here’s another version of our color wheel, but this time I included a triangle. The three points of the triangle correspond with the pigments that I used for the prints above. I used three paints: for cyan, I used a Winsor & Newton Primary Blue gouache. This is based on the PB15 pigment and I suspect it’s closer to the Red Shade than the Green Shade PB15:3 pigment we’ll come back to later. For magenta, I used a Talens Permanent Rose Magenta gouache, which is a dual pigment mix consisting of PV19 and PR122 (these are fairly close to each other in the color wheel). For yellow, I used Van Gogh (which is also Talens) Transparent Yellow Medium, which is a PY128 pigment. For the PY128 I had to take a stab at where it is in the color wheel. I suspect it’s pretty close to 90 degrees hue angle and quite likely pretty darn high up in chroma as well. I think I put it a bit low where it is; chroma might even be higher. Anyway, here’s what it looks like on the color wheel:

My first pigment set: a Winsor & Newton ‘primary blue’ PB15, Talens ‘permanent rose magenta’ PV19+PR122 and a Van Gogh/Talens ‘transparent yellow medium’ PY128.

Now, in the pale yellow triangle I drew, I have simplified something to the point of being dangerous, but it’s sort of good enough for an approximation, for now: I’ve drawn straight lines between the pigments as they are located on the color wheel. Why is this dangerous? Well, the triangle effectively encompasses all hues that can be made by mixing the pigments in various ratios. Consequently, any hue outside this triangle is simply impossible to be made by mixing these pigments. This is called the ‘saturation cost’ of mixing colors.

If I mix a green by throwing some PB15 and some PY128 together, I end up somewhere on the straight line between them – which happens to be very close to the center of the circle. In other words: I end up with green, but it’s muted green that has low saturation or chroma. At least, in theory. In practice, this isn’t entirely correct: sometimes, you can actually get a bit more out of a mix in terms of saturation – or a bit less. In terms of the triangle: the lines aren’t straight, but they’re actually curved. How much, in what angles etc. – I guess only actual testing can figure that one out.

Remember that triangle – it’s the gamut of all hues that we can print with our pigment selection. As you can see, the gamut of my first pigment set should do a pretty neat job at rendering reds, some violets and oranges, and blues and yellows will do just fine. One thing my gamut doesn’t do particularly well, is greens, teals and limes. Pretty much anything to the left side of the color wheel is something I cannot really do very well with these pigments. Is that bad? I don’t know. Look at the little picture of the green Martian alien earlier. It was made by mixing some cyan/blue and yellow that are very close to the ones in the triangle above. Still, it gives a fairly decent green. And then there’s the photo of the banana leaves. Those leaves to the left and right of that image are pretty green, aren’t they? They sure are. But not a particularly vibrant green, and certainly not as vibrant as in the digital original (don’t be fooled by the left image simply being darker – that’s not what I mean. The green is actually less of a toxic, eye-watering green than it might have been). That may have in part to do with how I overlayed the key/black layer on top of the color layers, but it’s also due to the harsh reality of the gamut as I created it. Unwittingly, but hey, I learn!

Up to this point, I had basically been selecting paints/pigments by following my usual ‘seat of the pants, let’s wing it and how hard can it be’ approach that I take early on in a project, until I hit my head hard enough to kick the grey matter into gear. I mostly used Calvin Grier’s ( pigment selection, which I gleaned from is Gum Printing book. If it’s good enough for him, it sure is good enough for me. After learning some more about it, I still think so. But I now understand much better why it is so. I like that feeling. It also helps me to envision alternatives. Grier’s approach is excellent. And there are other approaches that are interesting as well. He explores them himself as well, as evidenced by his more cent work with spot colors and earth pigments. So let’s have a look at that Calvin Grier pigment selection.

(By the way, if you wonder, why not just get his color pastes and be done with it? There’s several answers to this, which vary from “my local store doesn’t stock them and I like to buy local when I can” to “why ship stuff halfway across Europe if I may get something similar around the corner” to “he’s sold out on magenta” (he is at the time of writing; not an excuse for yellow and cyan, of course) to “what gives that he’s got a wheel that’s round, I can make a wheel too and even if it’s not round, it’ll be my wobble and I happen to like a little wobble to my ride.” It all boils down to “just let me do it my way and when I fail, I can buy into the existing solution knowing exactly why it’s better in every way.” It’s just fun shopping for stuff and trying it out! Oh, one final thing: Grier’s pigments are way cheaper than getting watercolors. Trust me, he’s not lying on that part. In fact, if anything, he’s understating this!)

Grier selected after ample testing for himself the following pigment set: for Cyan, he uses PB15:3, which is beta copper phthalocyanine. The “:3” addition is important, as the PB15:3 pigments lean towards green (and thus cyan, as they’re basically blue to begin with), whereas :0, :1 and :6 lean towards red. For Magenta, he prefers PR122, which is quinacridone magenta. For yellow, he uses PY155, a benzimidazolone pigment. For magenta and cyan, I linked to the excellent pages of (Bruce MacEvoy) who has taken it upon himself to write and test just about everything concerning watercolors, with a high emphasis on color and pigments. But it so happens that he’s apparently not omnipotent, and I cannot find anything about PY155 on his website. I therefore assume (the mother of all fuckups…) that PY155 is pretty close to PY151 and PY154 (also yellow benzimidazolones, and pretty darn close on the color wheel) in terms of hue and chroma. Why Grier didn’t just pick either PY151 or PY154, which seem to be more common? It may have something to do with staining, transparency, toxicity or dispersion stability. If I were to make an educated guess, I’d say it’s probably staining. You’ll have to ask him. Anyway, let’s see what the gamut of this Grier/thewetprint pigment set looks like:

The ‘Grier gamut’ (purple) and my first test gamut (pale yellow). Note I took PY154 instead of PY155 for the yellow of Grier’s set, but the difference for this illustration is probably negligible.

There’s a clear difference concerning the yellows, but I’m not sure this is meaningful. I assumed that Grier’s PY155 is close to PY155; I’m quite confident this is pretty much correct. I also assumed that PY128 that I used is ‘somewhere up there as well’. This is much more uncertain. In fact, I suspect that PY128 is probably very close to PY154/155 and that consequently this difference just isn’t very meaningful.

In terms of the magenta, we both used PR122, but mine was mixed with P19. Because PV19 is very close to PR122, I don’t think it necessarily impacts the gamut all that much. But for one thing: we assumed earlier that the lines between the pigment positions are straight, but also established this assumption isn’t accurate. I can very well imagine that a PV19/PR122 mixes in a different way with other hues, and that this difference might in fact result in the gamut being smaller than if a pure PR122 is used. Or maybe the opposite? I just don’t know, but I’m pretty sure it does something. So there is some kind of difficult to quantify difference here.

Then there’s cyan, or rather the blue side of things. I suspect my mix uses one of the other PB15’s, such as PB15:1, which is more red than Grier’s PB15:3. This should impact the Achilles’ heel of both gamuts: the green side. There’s probably a reason why dedicates a lengthy page all to the problem of mixing greens! I have yet to read, let alone comprehend, all of it, but I do presume from this that the slight shift of PB15:3 towards green as opposed to the more red-leaning PB15:1 c.s. will be meaningful in this regard.

Concluding then that Grier’s pigment set at the very least has clear merits, and probably is superior to what I ended up buying, can we then translate his set into a shopping list of watercolors I can take to my local shop? Again, I could buy Grier’s pigments, and there’s a good chance I’ll end up doing that at some point as well (sans the magenta, which has run out; I came too late to the party for that one!) Well sure, I could. And it would go something like this (numbers between bracket are the brand’s watercolor identification number):

  • Cyan: pigment PB15:3, hue angle at around 250-260 degrees
    • Winsor & Newton ‘Winsor Blue Green Shade/GS’ (207 or 707)
    • Schmincke ‘Helio Cerulean’ (479)
    • Daniel Smith ‘Phthalo Blue’ (053) – although it might be a slight tad too red
  • Magenta: pigment PR122, hue angle at around 0 – 10 degrees
    • Winsor & Newton ‘Quinacridone Magenta’ (545)
    • Schmincke ‘Purple Magenta’ (367)
    • Daniel Smith ‘Quinacridone Lilac’
  • Yellow: pigment PY154, hue angle at around 85-95 degrees (we take this as a substitute for PY155 as PY154 is common in watercolors, whereas PY155 doesn’t appear to be used much, or at all in this application)
    • Winsor & Newton ‘Winsor Yellow’ (058)
    • Schmincke ‘Pure Yellow’ (216)
    • Daniel Smith ‘Azo Yellow’ (but this is the closely related PY151 instead of PY154 or PY155) or ‘Lemon Yellow’ (which has PY175, again a closely related bennzimidazolone yellow, but it appears to have a somewhat lower chroma than the others)

I actually ordered a kit of Winsor & Newton tubes along the lines above to see how I like it. Testing can be fun – as long as it doesn’t run out of hand.

But, speaking of which…if we can figure out why a certain gamut may work, and how to translate it into a shopping list, could we also figure out an alternative to the solution above? Well, I suppose we could. After all, how hard can it be?

Taking as a starting point two things I don’t feel quite comfortable with in the present solution: the magenta isn’t really magenta, and the cyan isn’t really cyan. For a CMYK set, those two concerns sound pretty plausible, right? Well, in fact, I doubt these concerns are so grave. For instance, look at inkjet inks. Both PB15:3 and PR122 are quite commonly used in inkjet inks, at least as far as I can glean from whitepapers and patent applications. Surely, other blue(ish) pigments are also used, but at least PB15:3 seems to have its merits (page 7, bottom). Likewise, PR122 is a common occurrence in the same application field, so apparently, other people with far more knowledge than I have in this area do not really share my concerns. Still, let’s see where we end up if we tackle these (non-)issues.

For starters, let’s look at an alternative to PR122. As said, it’s a bit too red to be a true magenta, so what could we find in the actual magenta corner? Looking at the very nice chart on, there’s an interesting set of pigments at the bottom right we could investigate. Some of those, such as PV16, PV23 and PV15 (RS) would be interesting, if not for the fact that they’re a bit too close to the center of the circle. In other words, they have only a moderate chroma or saturation. Since we can never mix our way out of a low saturation situation (the opposite, however, is possible), I’d like our primary pigments to be as saturated as possible. This leaves us with something like PV49 (Cobalt violet). Alright, doesn’t seem to be popular, but at least Daniel Smith appear to have a paint based on it, so that makes it feasible.

Then the other one, the cyan part. Just like the magenta segment, the cyan segment does not appear to be very densely populated. Discarding the candidates of low chroma, we end up with something like PG50 (Cobalt teal blue), which happens to be a popular one. Like PV49, it has excellent light-fastness, so it won’t fade much. Now, PG50 comes in a couple of forms that range from something closer to green to something closer to blue. It so happens that the greener shades are also of lower chroma, so instead I’d select something closer to blue – although the greener shade would be interesting as it should contribute to better rendition of green hues. If that is desirable, we could also lean towards PG7 (Phthalo green blue shade), although it has a bit lower chroma than the blue-leaning PG50’s.

For yellow, I simply don’t see anything wrong with the PY154 we already found; it’s a true yellow, it has very high chroma and pretty much no horrible downsides. Why fix something that’s not even close to being broken?

So our alternative pigment set could be like this:

An alternative pigment set based on PG50 for cyan and PV49 for magenta. This set displayed in blue; the Grier set (proven technology!) based on PB15:3 cyan and PR122 magenta in purple.

I don’t really know if the gamut of the alternative set will necessarily be bigger. It depends a lot on how the colors mix, I suppose. There’s probably a reason why the extant pigment sets for inkjet inks seem to lean heavily (or virtually exclusively) on PB15:3 – a reason I don’t know (yet), but that someone who’d try this alternative set would probably find out the hard way! So for the adventurous among us, here’s a shopping list attached:

  • Cyan: PG50 (Cobalt teal blue) or PG7 (Phthalo green blue shade)
    • Winsor & Newton ‘Cobalt Turquoise Light’ (191; PG50) ‘Winsor Green Blue Shade’ (719; PG7)
    • Schmincke ‘Cobalt Turquoise’ (509; PG50)
    • Daniel Smith ‘Cobalt Teal Blue’ (PG50)
  • Magenta: PV49 (Cobalt violet)
    • Daniel Smith ‘Cobalt Violet’
  • Yellow: PY154 (Benzimidazolone yellow)
    • Winsor & Newton ‘Winsor Yellow’ (058)
    • Schmincke ‘Pure Yellow’ (216)
    • Daniel Smith ‘Azo Yellow’ (PY151 instead of PY154) or ‘Lemon Yellow’ (PY175 instead of PY154)

As seen above, in this pigment set, the magenta is a bit of an oddball and indeed it seems to be a rare beast in watercolor pigments. Why? I just don’t know. I do understand there are only one or two suppliers of the pigment, and perhaps paint manufacturers would rather lean towards pigments that are easier to source than to become dependent on a single source, resulting in relationship tension at some point in time. Pure conjecture on my part; it might just as well be the case that PV49 is difficult to disperse or something like that.

Oh, did I say adventurous? Well, did I also mention that in my order of some paints to try out, I actually did include some Winsor & Newton ‘Cobalt Turquoise’ and some Daniel Smith ‘Cobalt Violet’? I’m not saying I will try it, but you know, if the opportunity arises, I just might

Leave a Reply

Your email address will not be published. Required fields are marked *