It’s been awfully quiet on the color carbon front, hasn’t it? Well, that’s partly accurate. I haven’t done as many test strips these past two weeks as I’d been doing before, and the reason is that I’m at a crossroads with this project now. Having learned lots, it’s also becoming clearer now what I’m dealing with. The question is – how to proceed? Let’s start with exploring some of the challenges I’m currently facing, which all happen to revolve around consistency and linearity.
Starting with the ‘important and gratuitous’ thing, in case you’re wondering: this is one of those situations where I’m just ‘discovering’ what I already knew, because I’d been warned. Nothing in this blog is going to be really new knowledge. It’s just stuff I had already read about and that now falls into place. Much of it comes from Calvin Grier, who also wrote these priceless words:
I swear that if someone had told me what I’m about to tell you when I started printing, I would have saved about 5000 euros and six months of wasted time.
Calvin Grier, Calibration for Alternative Photographic Processes, Guide 2:Linearization, page 9
Well, he’s right. I’m not yet those six months in and certainly not the €5k, but the traps he apparently ran into, I didn’t avoid either. Mind you, I was aware of the sentiment expressed above as well as the main warnings Grier had to share, as they’re also addressed in his Guide to Gum Printing. I ran into them anyway, head first. I knew that I had to experience all this myself for it to really sink in.
Consistency is the name of the game
So what’s this all about? Simple things, really, and they all boil down to the same thing: consistency. And by extension, probably: linearity. And there are a few ways in which it turns out to be a challenge.
Firstly, let’s revisit those negatives again. I confessed before that I’m not very fond of inkjet digital negatives and I’m afraid this hasn’t improved all that much. Yes, finding more suitable transparency materials has made a massive difference! I’m now mostly using very cheap, generic screen printing film that looks suspiciously like Fixxons. Real Fixxons does hold a bit more ink before it goes wonky, so the screen printing film is somewhat inferior, but it turns out to hold more than enough ink to create sufficient density for carbon printing. So far, so good.
What is not so good is that inkjet printers really aren’t all that great of an option when making digital negatives with pigment printing processes such as carbon transfer. This has to do with a couple of ways in which both sides (inkjet and pigment processes) tend to misbehave in reality, and this in turn has everything to do with linearity.
Linearity is simply having e.g. 10% density in a digital file and that printing as 10% density on an inkjet printer or any other printing process. And the same for 90% etc. etc. Ideally, x% in (on the digital negative as created in digital computer space) equals x% out (on the end-result, physical printing process). In reality, this doesn’t happen readily, and you have to compensate digital files so that they accommodate all the unique ways in which non-linearities emerge in the chain of printing steps. So you define, let’s say, (x minus 10)% in to get (x%) out.
Problem #1: Dot gain in high and low densities
One of those sources of non-linearities is the dot gain that’s inherent to inkjet printing and several other processes – including carbon transfer. Dot gain simply means that when you intend to print a dot of e.g. 10um (micrometer) diameter, it may end up as a bigger dot, let’s say 15um. It’s easy to see how this would happen with inkjet: as the droplet of ink hits the paper or transparency, some of the ink will splash about and bleed out, making a bigger dot than intended. In a large part of the tonal scale, this isn’t particularly problematic, but it tends to be nasty especially when printing high densities.
High densities in an inkjet print consist of many dots spaced very closely to each other. If the dots are then bigger than they theoretically should be, they are no longer closely spaced, but they start to overlap. As a result, if you print a ‘linear’ tonal scale with inkjet, what you really get is a sizeable part of the chart being solid black before the dots start to separate and the black turns into shades of grey. If you try to compensate for this in a compensation curve, it tends to create a steep kink in the curve to escape from this solid black part of the curve.
Have a look at the scan of a digital negative step wedge above. It’s an uncompensated print made from QuadTone Rip on an Epson 3880 printer, scanned on an Epson 4990 scanner. It looks pretty much OK at first glance, but let’s zoom in on the darkest patches:
In the image above, you can see the 100% density bar at the top – actually the border of the step wedge negative, but it’s identical to the 100% density patch. The patch on the right is the 90% one, to the left is 95%. Note how there isn’t any pure white between the individual dots, that have more or less fused together. For a continuous tone negative, this might not be problematic, but we’ll see later on why this does present problems for carbon transfer specifically.
Mind you, at the other end of the scale, i.e. the low densities, something similar happens. To make a very light tone, an inkjet printer deposits tiny dots of ink widely spaced from each other. If the dots are all black, they need to be spaced very widely to create very light tones. Inkjet manufacturers solved this issue years ago by not making all dots pitch black, but add lighter shades of ‘black’ as well. My Epson 3880 has Black, Light Black and Light Light Black, so three shades of ‘black’, and that’s still a common topology in today’s printers.
The advantage of these lighter shades of black is that making a light tone can be done with a light shade of black (=grey) and have the dots spaced a little closer to each other. This makes the individual dots less apparent, and overall it creates a smoother tone.
Let’s have a look at the lightest tones in our step wedge digital negative:
Note how the dots in the 5% patch come in pairs: there’s a darker one below a lighter one in each pair; these are two ink channels working together. More importantly, note how far apart the dots still are. To create this light tone, which looks even enough in the dramatically scaled down initial scan, we actually have a pretty coarse tone if you look up close at the negative.
Actually, I plotted the measured densities from the scan above and it’s clear that things aren’t as nicely linear as they might be right from the start:
So before we even get to make a carbon print, the output of the inkjet printer is not very linear to begin with. Note the little ‘toe’ in the shadows at the top right of the curve, with the 95% patch being darker than it should be due to dot gain. Something wonky is also going on in the low densities for reasons unknown to me, probably poor linearization of the QTR inkjet curve, resulting in the low density patches all being too low in density. Yes, I might linearize the QTR profile to begin with, but (1) why bother if we are going to do that with the carbon transfer processes appended to it anyway – the goal is not to make linear negatives after all, but linear carbon prints. And (2) this illustration as it is shows neatly that the world of inkjet isn’t perfect, and we have to keep in mind its inherent properties when using inkjet for making digital negatives.
Now, the above was about inkjet, but dot gain also plays a role in carbon transfer printing. Actually, it tends to be even more problematic, because the tiny spaces between individual inkjet dots tend to bleed up very badly on a contact printing process where perfect negative-to-tissue contact and perfect collimation of a light source are nearly impossible, and it’s also impossible (especially with color pigments) to entirely eradicate light piping and halation issues within the carbon tissue itself.
As a result, there will always be some dot gain, although it can be minimized. Note e.g. that Calvin Grier works with a light source placed at a pretty large distance from his contact printing frame, he uses as far as I can deduce fairly high pigment concentrations and hence thin tissues and overall has spent a lot of time on printing perfect dots, i.e. reducing dot gain as well as dot loss. In my current setup, dot gain will be substantially more than in Grier’s.
The end result of the dot gain in carbon tissue is that those low-density patches in an inkjet negative will easily bleed all into black, requiring very steep extremes in adjustment curves to compensate for, like the one below. It’s a screencap from QTR, stretched vertically to obtain a more intuitive aspect ratio as the original incorporates a hefty ink density limit which squashes the curve into a low-profile shape. Mind you, this in itself also has obvious drawbacks for color resolution, potentially causing posterization. An issue I’ll not discuss further here, but that’s worthwhile keeping in mind as well.
Problem #2: Low densities in carbon transfer prints
Inkjet printing isn’t ideal, and carbon printing is no different. And that’s not only about the dot gain issue! One well-known issue with carbon transfer prints is achieving a smooth transition between paper white and low densities. A carbon print is pigment trapped in a gelatin matrix, and optical density is the result of the thickness of this gelatin layer. The thicker the layer, the more pigment is stacked on top of each other, the higher the optical density.
To print very low densities, i.e. subtle highlights, the gelatin layer needs to be very thin. And given the nature of the carbon printing process, this means this layer will be very fragile and difficult to handle especially when it’s being developed in the warm water bath where the unexposed gelatin washes away. The thinnest layers of exposed gelatin will generally also wash away. That means that there’s always a distinct ‘bump’ between paper white (no gelatin) and the lightest printable tone (the thinnest gelatin layer that will reliably survive processing). Calvin Grier calls this the ‘tonal threshold’ in his publications (e.g. in his Gum Printing manual, see page 73).
There are several ways of dealing with this issue. Commonly, a low pigment concentration is chosen, which results in a low density still requiring a somewhat thicker (and sturdier) gelatin layer. Very careful processing is another approach, using a warm water development temperature that’s just above the melting point of gelatin, around 40C or so. This results in a longer development time, but it will reduce the thinnest exposed gelatin bits flaking off.
Such things do help…a bit. Because in reality, they only reduce the problem without actually resolving it. There will always remain a minimum gelatin layer thickness that’s required to survive processing. And that means that the transition from paper white to the lightest printable tone will never be perfectly smooth. It may be very, very close, and it may not seem like a problem especially in B&W prints and depending on the image content. But the problem will be there, and in certain prints (skies with wispy clouds, for instance) it will still pop up. As well as in color prints, because there it doesn’t just cause a coarseness in the highest tones – it will create very visible color shifts, where each color layer suffers from this high-tone problem to a slightly different extent.
Problem #3a: Tissue sensitization – drying and (not) waiting
As with digital negatives, I’ve made significant strides in sensitizing tissue as well. Not that I’m an expert carbon printer all of a sudden – far from it. But learning tends to come in big strides especially at first, and I’ve made some of these strides in sensitization in particular.
So far, I’ve printed using dichromate – the classic approach employing the toxic, carcinogenic and downright nasty chromium (IV) salt. Very nasty indeed, but it works very well, too, and most carbon transfer literature is based on it. The trick with sensitization is to get the dichromate into the tissue just before exposing the print. Leave the sensitized tissue to sit around for too long, and the so-called dark-reaction will induce fog. This limits the practical lifetime of dichromate-sensitized carbon tissue to a day or so at room temperature (provided it’s stored in absence of any UV or blue light!)
But for the print to come out well, the tissue needs to be sufficiently dry, too, so there’s also a minimum time it needs to sit around and dry before it can be exposed. Printing with too moist tissue creates various problems, such as uneven densities in especially high tones, showing up as cloud-like patterns. The tissue may (will) also stick to the negative; especially inkjet negatives tend to be prone to this and are generally destroyed if this happens. This issue can be alleviated somewhat by placing a very thin, non-stick transparent film between the tissue and the negative. I use a double-sided siliconized boPET film for this.
But printing with moist tissue is also a problem in terms of consistency. The speed and contrast of the sensitized tissue not only depend on dichromate concentration, but also on remaining moisture content within the tissue. A moist tissue will generally print slower and with lower contrast than a perfectly dry tissue. When tissue is so dry that it doesn’t feel tacky anymore (a sharp object like a fingernail won’t leave a very apparent dent in it, easily) it tends to be on a ‘plateau of stability’ where it’s reasonably consistent in terms of printing speed and contrast. Since it’s very difficult (neigh impossible) to accurately verify the exact moisture content of the tissue, the only feasible way to get consistent prints is to dry the tissue completely after sensitizing. But not so long that the dark reaction kicks in.
And the sensitization story doesn’t end here…
Problem #3b: Tissue sensitization – consistency in concentration
Sensitizing the tissue equals mixing dichromate (or…well, we’ll get to that) into the already existing, dry tissue, in a perfectly even way and at a known concentration. And it so happens that those two requirements are awfully difficult to manage.
Consider the well-known ways of dichromate sensitization: soaking the tissue in a watery solution of dichromate for a set amount of time, brushing a mixture of water and alcohol and/or acetone with a known concentration of dichromate onto the tissue, or doing the same not with a brush, but with a foam roller.
Soaking has the potential for great evenness, but has a few obvious drawbacks. Firstly, it’s horribly messy and wasteful, and when it comes to handling a highly dangerous compound, those aren’t just inconvenient drawbacks, but huge risks. Secondly, the rate of diffusion of the dichromate solution into the tissue will depend strongly on the temperature of the solution, with the pre-existing moisture content in the tissue probably playing a minor role as well. This means that to have good control over the final concentration of dichromate, tight temperature and timing control are necessary. And a safe way of handling and drying tissues soaked in a dichromate solution needs to be devices. Sure, it can be done and indeed has been done (it was the default approach back in the early days of carbon transfer), but it has never been my cup of tea. Yes, I tried it once or twice and quickly decided this was NOT a good idea.
Brushing comes next and in my experience works quite well for small tissues, i.e. significantly smaller than 8×10″. With larger tissues, it seems that evenness quickly becomes a problem. I just never found a bulletproof way to brush-sensitize an 8×10″ tissue with decent evenness using the regularly recommended foam brush.
So how about rolling with a high-density foam roller? Well, that seems to work quite nicely, for small as well as bigger tissues. Indeed, so far I’ve had the best luck with this approach. Hey, if it works for Sandy King whom I ‘stole’ this technique from, it should be plenty good enough for me.
Or would it? The thing that irks me with both brush and roller sensitization, and especially with the latter, is that there’s just no way of knowing how much of the dichromate ends up in the tissue and how much remains inside the brush or roller. I noticed issues with this when sensitizing several tissues in a row (e.g. cyan, magenta, yellow and black right after each other). It would take different (increasing) amounts of time to finish sensitizing, and there was/is the obvious issue of starting out with a freshly cleaned roller, perhaps moistened (or not? who’s to say?) and doing consecutive tissues with a roller that already has some (how much?) dichromate embedded inside of it. So there are obvious issues with controlling sensitizer concentration in the tissue even with roller sensitization and good technique – although here, just as with highlight control, finetuning technique may reduce the problem without ever truly eliminating it.
Finally, I noticed when printing color step wedges and measuring them with an XRite i1 Pro that even with my best attempts at roller sensitization, I got measurable density differences across the same sheet. The only conclusion I could draw was that the sensitization wasn’t perfectly even. Mind you, it is perfectly even enough for making very nice B&W prints that don’t exhibit any apparent unevenness in density. But if I measure it, it does appear to be there. And again, what I can get away with in B&W, I won’t necessarily be able to get away with in color.
See the example above; this is of a tissue that was apparently very badly affected by uneven sensitization, or at least the magenta one. See how especially the a* coordinates seem to follow a staircase-like pattern? The step wedge pattern I used consisted of rows of patches and you can see the difference in sensitization between individual rows – a typical result of poor roller-sensitization. Sadly, the problem is difficult to eliminate entirely using this method of sensitization, making the process somewhat unpredictable and very sensitive to technique.
So sensitization of the tissue remains an issue!
‘Problem’ #4: Tissue pigment load
This is a major and a minor issue at the same time. It’s major because pigment concentration in the carbon tissue is really the most effective control of contrast and/or density. I’ve done a lot of experimentation with different pigment loads and it’s far, far more effective in adjusting contrast than adjusting e.g. dichromate concentration. The downside is that for good consistency, the pigment load of the carbon tissue needs to be tightly controlled. A small difference in pigment load can translate in a major difference in output density.
This issue is exacerbated when working with dry pigments, since with most modern pigments, the required amount is really tiny. To give an impression, for black carbon tissue, I use around 0.15% w/v Kremer PBk7 in my glop (with 8% w/v gelatin) to get similar performance to 1% Talens India ink in a similar formulation. That means only 150mg of Pbk7 in small test batch of 100ml glop. And it’s really difficult to accurately control such a tiny concentration.
The issue is actually minor, because the solution is kind of obvious. With e.g. India ink or watercolor paints, it’s already easier to have good control over pigment concentration, as they are far easier to measure out and disperse into the glop than a powdered pigment. The obvious solution then is to make something similar to India ink or watercolor paint with the pigments I’m using. Which is what I did; I got myself a little rock tumbler and made some pigment dispersions with it in varying concentrations between 1% and 10%, the latter being the highest concentration that is easy enough to make with my equipment in small batches. And 10% ends up being plenty concentrated enough.
So let’s consider this problem as one that’s easy enough to deal with. Well, the way I look at it now, that is. Perhaps I’ll have to revisit this bit of hubris in the future.
Conclusion and outlook
Where does this leave us? Well, there are apparently inherent problems with inkjet printing (for digital negatives) and carbon transfer printing in itself related to how both media print dots (or fail to do so), and how carbon transfer deals with delicate highlights (or refuses to do so). And there are additional challenges with how the tissue is made and processed that pose a threat to consistency.
Combined, these challenges have made it pretty frustrating for me to linearize test strips. I have learned a lot along the way, but also realized that the process as I was practicing it, was/is just too prone to variation in all sorts of parameters to make it sufficiently consistent for decent color printing.
As a result, to continue, the challenges outlined above need to be faced. Fortunately, there are opportunities for this, but they come with their own challenges. I’ll go into these in a next blog post, so stay tuned!