Wavy gravy – UV wavelengths for carbon DAS revisited

About this whole highlight business with DAS carbon transfer – it just never ends, does it? The last installment on this story had a loose end that I need to stitch up here. It’s about the nature of the light. Let’s have a look at how UV wavelength affects the curve, and highlight performance in particular.

To point out the bleeding obvious in my previous blog about the contrast ceiling thing:

On the left, you’re looking at the same tissue, top and bottom. On the right, same story: same tissue, top and bottom. I also did not modify the step tablets, nor did I sneakily slide in a different negative. So how come the top row has overall totally different contrast from the bottom row?

The answer, of course, is in the light used for the exposure. I had been playing with this before. My conclusion back then boiled down to “you need at least a little 365nm wavelength exposure to make DAS carbon work.” This resulted from earlier and quite disappointing tests with 395nm exposure units – which worked great for dichromate carbon transfer as well as other alternative printing processes. Just not for DAS carbon.

The story at that point in time ended with the purchase of a set of still relatively cheap & cheerful UV floodlights consisting of a mix of 365nm and 395nm LEDs. I’ve been using that setup ever since and overall it has worked quite well.

Except for one little thing: collimation could be better. This is really food for yet another blog, but at some point I was messing about with inkjet digital negatives. Sure enough, the floodlight array is lightyears (heh) better than a bank of UV tubes. Still, I think better collimation than the floodlights can give me, is necessary to get better dot quality.

So I ordered some COB LEDs, cooler, LED drivers and some lenses to see how that would work. I’ve not touched upon the collimation issue yet. Since these are COB LEDs, I had to decide which wavelength to go with, so I decided to get one 395nm unit as well as a 365nm unit.

Here’s the test setup. You’re looking at two 100W COB LEDs (365nm on the left, 395nm on the right – not that you can see the difference like this) hiding beneath (actually above, in the photo) two 60 degree lenses. The COB LEDs are mounted on a cooler with two fans, and there are two LED drivers hidden behind the light source. It all plugs into the light integrator I made last year. Convenient. I can flip a switch on each unit to select 365nm and/or 395nm.

Here it is, doing its thing. Light integrator/timer unit in the upper right corner. Note the probe sitting on the contact frame. The 395nm LED is on for the photo. Note how the beam angle is pretty narrow. At this distance, it covers my regular 12x15cm test area – but nothing more. I can drop away the worktop and place a bigger frame a little lower, in which case it does cover 8×10. Of course, exposures are much (!) longer that way – same amount of light, larger surface area.

Anyway, back to the question: what does the light source do in terms of DAS carbon? Well, turns out it does a heck of a lot indeed. See the illustration prints at the top of the article. Summarized:

• 395nm gives far more contrast than 365nm.
• 365nm prints a lot faster if you look at highlights and midtones – dmax comes quicker with 395nm however.
• 365nm gives much better tonal threshold performance; the transition from tone to paper white is much smoother (other factors as touched upon in earlier blogs still apply).

Well heck, I can put more words here, but a set of curves tells the whole story, really:

By ‘comparable exposures’, I mean I set a fairly random number of UV units on my light integrator and then made two exposures for the number of units, one with 395nm only and the other with 365nm. To achieve the same number of units, the 395nm exposure was something like 8 minutes, while the 365nm exposure was done in around 3m30s if memory serves. This makes good sense given the peak absorption of DAS, which is a little under 350nm and drops quite rapidly towards 400nm.

The contrast difference is likewise easily visible in the steepness of the blue curve, representing the 395nm exposure. Note that it also has a bit of an S-shape going on. The 365nm curve is more linear, but also a lot shallower. While the 395nm exposure reached a fair dmax (for a low-pigment tissue) of around 1.65logD, the 365nm exposure pooped out at a feeble 1.2logD.

Finally, note how the densities of the 395nm exposure drop away rapidly into the highlights and sort of bump into paper white – representing a harsh tonal threshold. Indeed, the highlights on a pure 395nm exposure look harsh, brittle and overall…yuckie. The 365nm exposure on the other hand gently touches down into paper white.

The obvious question then becomes…what happens if you combine them? Unsurprisingly, this results in a best of both worlds situation:

The pink curve is a plain 365nm exposure for 3312 UV units. The light blue curve is a 395nm exposure worth 1312 UV units. Coincidentally, these exposures are both 8 minutes – the time I happened to pick for testing. Hence the odd UV choice of UV units (not really a choice at all, but a coincidence). Note how everything I said above about these single-wavelength exposures features in these plots just the same.

The purple plot with the blue markers represents a mixed exposure of 1500 units 365nm and 2000 units 395nm. This ratio was arbitrarily picked, but happens to be a quite nice compromise. Note how this mixed exposure tracks more or less perfectly with the high-contrast 395nm curve in the shadows. That is, except for a little superimposed bulge around step 3-4 that is also recognizable in the 365nm exposure. From the crossover point around step 7 onwards, the mixed exposure curve starts to lean towards the 365nm single-wavelength curve.

The net effect is strong contrast in the shadows, and at the same time a very gentle touchdown in the highlights and tonal threshold behavior. It’s very much what you’d expect to get when printing a negative that was processed to give significant compensation, or subtle pre-flashing.

Here’s the example image. The pigment concentration in this tissue is quite low; it’s 0.6% Kremer XSL black (a dry pigment that disperses easily in water). Note that I express pigment concentration as a percentage of the dry gelatin weight. Hence, a 1 liter batch of glop with a 10% gelatin load would contain 100g gelatin and 0.6% of that weight in pigment, or 0.6g. A little goes a long way!

Note in the example image that there is differentiation throughout the entire range of the Stouffer T2115 step wedge: a range of 3.15logD! In fact, the range of the print extends further; note the white border around the image, which is true paper white, and step 21 has a little density above this level.

Here’s another example, on a different type of tissue and with a different ratio of 365nm (800 units) to 395nm (1500 units) exposure.

It’s a tissue with far more pigment (2% vs. 0.6% – more than 3x as much!), so evidently it’s a lot punchier. Looking at the curve shape of this one and the previous combined exposure, the difference in exposure ratio is visible:

The curve on top with the diamond markers is the second example with the overall higher contrast. Note how this curve more closely follows the pattern of the pure 395nm exposure I showed earlier (the blue dashed line in the plot above). Yet, it does taper off a little towards the highlights due to the added 365nm exposure, making the highlights a little smoother.

Note also that this second example doesn’t show a very nice tonal threshold behavior. The ‘contrast ceiling’ argumentation still holds – it’s a high-contrast tissue, so it’ll be challenging to get smooth highlights from it to begin with. Moreover, this tissue was one out of my test batches with thin tissue, and as I concluded before, that doesn’t work well for highlight rendition. The thin tissue also explains the wavy edge on the right; this is actually flagging caused by exposing almost all the way through the tissue and close to the even thinner edge, causing a transfer problem. It’s a preventable issue, but it does show once more how complex interactions in carbon transfer printing can be.

There’s still an elephant in the room of course. The fact that the 365nm exposure is substantially faster than the 395nm (judging by highlights and upper midtones) follows from the spectral absorption of DAS. But that doesn’t explain the huge difference in contrast.

I think for the contrast to be explained, two factors need to be taken into account: penetration of the UV light into the tissue, and self-masking behavior of DAS.

Concerning the latter (self-masking), there’s an insightful post on Groups.io based on a useful bit of testing by David Pitcher that shows how DAS stain increases with exposure. This stain emerges during actual exposure, and David’s data consist of UV transmission values. The gist is that UV density of the gelatin tissue increases during the actual exposure. This well-known effect is referred to as self-masking, since the created stain in heavily exposed areas makes it more difficult for additional UV to make it through, effectively holding back exposure in those areas. This results in compensation in the shadow areas of the print.

However, the self-masking phenomenon in itself still doesn’t explain why there’s such a difference between different wavelengths. For this, I think we need to take into account the fact that shorter wavelengths tend to be attenuated more strongly in a medium than longer wavelengths. Or, put differently, longer wavelengths travel further. Compare it with the party a couple of apartments away or a music festival a few miles down the road: what remains, are the booming bass tones, while the higher frequencies are gone.

Something similar happens in a carbon tissue. The smaller 365nm wavelength light is absorbed in the superficial zone of the carbon tissue, where it either forms image density (as it’s absorbed by DAS molecules), or is simply extinguished as it hits other materials. The longer 395nm wavelengths by contrast penetrate deeper into the tissue as they are less likely to be blocked by its constituents. The result of this difference is that the 365nm wavelength exposure will create a lot of DAS stain in the upper part of the tissue. This then blocks further exposure quite effectively.

This would explain very well why the shorter wavelengths work well for rendering delicate highlights, as well as the limited shadow density created by those same wavelengths. Short-wavelength hardening happens close to the tissue surface, making a nice and somewhat contiguous layer of hardened gelatin that better survives warm water processing. At the same time, the DAS stain formed in the upper layer of the tissue creates an increasing threshold for the buildup of density, which would have to happen also in deeper regions of the tissue.

One implication I take from this is that when combining exposures, it’s probably more effective to start with the shadow exposure. After all, the short-wavelength exposure for the highlights creates a threshold for shadow density to build up, and while longer wavelengths will still more easily penetrate the tissue than shorter wavelengths, DAS stain in the upper tissue area will still block some of it. I the tests I showed above, I started with the short-wavelength exposure – which in hindsight was not the most appropriate route. I’ve not yet done densitometry on an A-B vs. B-A comparison.

All this leaves one somewhat puzzling observation with regard to the example this article started with. It’s the difference in shadow density between the 365nm exposures on a thick vs. a thin tissue.

Here are the curves of those two prints:

Note how the highlight densities track fairly well between both tissues, but the thin tissue starts to bulge out towards higher densities in the midtones. This is counter-intuitive, since the thin tissue actually builds more shadow density. What’s up with that?

I suspect it has something to do with halation, or at least light bouncing around in the tissue itself. On the thin tissue, some of the exposure will simply go straight through the tissue and bounce back against the white (so fairly reflectant) Yupo substrate. This happens on the thicker tissue as well, but naturally less so – it’s thicker, after all, so it’ll attenuate especially shorter wavelengths stronger. I admit it’s a bit of a guess, so I’m open to better explanations. One thing that bugs me is that the thin-tissue exposure still flattens off into the highest densities – maybe self-masking catches up with the light-bouncing effect at some point. I really don’t know.

In any case, I’ll probably continue to exploit a dual exposure approach for continuous tone negatives to control contrast and highlight rendering. The technique seems to be pretty effective. It’s probably a good idea to trade the simplicity of my previous dual-wavelength light source (which did not allow for separate exposures) for the better control of a dual light source. And maybe the better collimation could help in some way, too. I hope to come back to that aspect in the future.