Re: uncurve/curve comparison on Loris' website
My mistake then. It sounds like you have been able to calibrate a curve that works well for you.
Sorry, Mark, I don't always get your posts, so I have to go to the archives to see your responses and I don't always remember to do that.
Although the post of mine you referred back to wasn't about any specific curve of mine but was a general discussion about the tradeoffs involved in constructing curves for gum, and although the general shape of the curves I generate is just as a curve "should" be, I do appreciate your offer to diagnose any curve problem, and will be sure to take advantage of that kind offer in future if I think I could use your advice about curves. At any rate, I figured out some time back that the problem was that the printer settings were wrong, as you suggested and as we've discussed.
But I need to say a word about that earlier post you referred back to: I want to make sure everyone understands that I wasn't criticizing the curves that I was describing, that had been graciously sent to me by others; I was simply using them to illustrate the tradeoffs that must be considered when constructing curves for gum, that we all face. My point was that with gum you always have tradeoffs. Given a middling level of pigment saturation, you can have good separation in the midtones at the sacrifice of separation in the highlights and shadows, or you can have good separation in highlights and shadows, by sacrificing separation in the midtones. The examples of curves I mentioned were from people who had elected for the latter option. I don't think there's any right or wrong about which way to go; I wasn't condemning those curves or those who created them for being "wrong;" I was only using them as an example of the tradeoffs that are necessary, whether we are conscious of them or not, and as a way of saying we're all in this difficulty together.
The comparison offered at the top of this thread serves as an illustration of the other approach, to concentrate on separating midtones and let the highlights and shadows go, although in this case the difficulty about highlights and shadows was apparently resolved by pushing most of the outside values, the shadow values especially, into the midtones (I'm speaking absolute values here, not relative values, and print tones rather than negative densities). As I said, I don't think either approach is right or wrong, they're just different ways to deal with the same problem. I would probably approach that image differently. I wouldn't lighten the shadows, because the print isn't too dark overall, in the sense that I mean by "too dark" (and that I illustrated in the print that I showed). In this print it's mainly the midtones that are too dark; the shadows are fine right where they are. I like the tires still black but somewhat lighter than the darkest tones, as they are in the "before" print, and would want to keep them there rather than to lighten them into the midtones. But, that's neither here nor there, because it's a very subjective call and as I've said, there's no right answer.
As I said before, I find the effect interesting, and the comparison useful as an example of two very different ways to print the same image. But I don't find it compelling as proof of the superiority of a particular approach to curves for tricolor printing, since the "after" print has a pronounced color cast and doesn't cover a full tonal range (in the absolute sense rather than the relative sense; in other words to my mind a good set of tricolor curves, for an image that includes a full tonal range as it appears this one does, should result in a print that also includes a full tonal range, including shadows that run from, say reflection density 1.40 or so to .9 or so, and highlights that run from say .20 up, as well as a full range of midtones).
The tradeoff I discussed above holds for fairly common mixes of a middling color saturation; different mixes introduce different tradeoffs, as there's a complicated interaction between negative densities and pigment saturations as they affect print tonalities.
As for tricolor, my calibration for each of the pigment mixes I'm using in my current explorations of tricolor (Prussian blue, PV19 gamma, and PY97) found the curves all the same, a deep curve that looks identical across the three pigments, although I haven't superimposed them to make sure they correspond at every point. My study of curves and pigments suggests that differences in curves are more a function of differences in color saturation than of the colors per se; a less-pigmented mix generates a shallower curve than a more- pigmented mix of the same pigment; a less-saturated color generates a shallower curve than a more-saturated color, regardless of the pigment. So it makes sense that since I use my colors for tricolor in equivalent saturations (note, I don't mean equal pigment concentration; I mean equal color saturation, which is something different) in order to ensure color balance, that the curves would be the same. Whether these curves work well for tricolor is yet to be seen, since I haven't had a chance to print those separations yet, and since as I keep saying, tricolor is more a matter of how the layers interact than of calibrating the individual layers.
Looking back over this thread on the archives, I see I owe you an apology, Mark; you weren't the first person to drag my curves into this thread that I thought had nothing to do with me. There was an earlier post I hadn't seen before, that addressed me by name and stated rather imperiously that my curves, or the system I use to calibrate my curves, must be wrong. That post also suggested that my purpose in posting the two prints I posted in another thread was to show that curves are worthless or useless, which of course is just silly. If I thought curves were worthless or useless I wouldn't have undertaken a study of curves as part of my ongoing study of negative densities and pigment concentrations as they affect print tonalities. The two prints were posted in a lighthearted expression of solidarity with a person who was struggling with curves for tricolor gum, and if there was a point I intended to illustrate with it, it was simply the point that what works for three separate layers doesn't always work for three superimposed layers, and that what we need is a three-dimensional calibration. The comparison was simply posted as a sharing of work in process between friends; I'm sorry it was misinterpreted by others.
On Sep 30, 2007, at 4:59 PM, Katharine Thayer wrote:
Mark wrote:Maybe I misunderstood one of your ealier posts.... I don't have time to search back right now for it, but I thought you had been working on a calibration and the results had been what seemed to be ok differentiation in the deep shadows and highlights, but very flat midtones.... I thought if you could post what the curve liked like when shown in the Photoshop Curves Window that it would be helpful to see what is going on. Usually you can tell by the shape of the curve what the problem is.Yes, you did misunderstand completely. That was in a different thread altogether; I was discussing curves for gum in general, and to illustrate a point, I was describing some curves and charts people have sent to me over time as examples of "good" gum curves, that fit that description. They were generated, actually, from the PDN system (although I didn't mention that because I didn't think it was germane to the discussion). But no, I wasn't describing any curve of my own, nor do my curves fit that description.