U of S | Mailing List Archive | alt-photo-process-l | Re: gum curves

Re: gum curves

I'm beginning to gather that a person could get lost in the minutia of this curve generation stuff and never get out again; now I'm being told (elsewhere) that it's possible my scanner may be incapable of accurately capturing the highlight end of the scale and for that reason ChartThrob may not be giving me an accurate curve. I hope this isn't true, because I paid rather a lot of money for it as scanners go these days, but at any rate I think I've reached the point of diminishing returns on this issue for the moment. Maybe when the open source curve generator materializes, I'll give it another shot, but in the meantime, I'd rather be printing gum.

On Nov 12, 2006, at 10:12 AM, Katharine Thayer wrote:

On Oct 31, 2006, at 5:24 AM, Loris Medici wrote:

Thanks for that Sandy. All that you've wrote tells me that I'm in the
right direction. I've just started to adjust my gum / pigment /
dichromate / exposure amnts. aiming to get the same amnt. of steps (or
close to each other) on step tablet test prints for each color I've
chosen - hoping this will equalize the contrast of each color layer.
Then I will start to calibrate a curve for each printing protocol. Doing
so, I hope to get a fairly balanced result in the end... Absolutely no
problems if I don't; it will be a good exercise anyway! :)

Yes, it will be a good exercise, and I'll be interested to see if it gets you a good color balance in the end. It depends to some extent on the yellow you choose, but I really doubt that any yellow can be made to print with the same contrast and range as a cyan, no matter how you manipulate other variables. A pigment can only do what a pigment can do. And if you (borrowing Judy's image) tie two hands behind your back by limiting the tonal scale of your blue and red to match the yellow, then I don't see how you could get anything but a pale, anemic tricolor print. But I stand ready, as always, to be persuaded by evidence to the contrary.

As to whether the right curves (profile) can save an unbalanced pigment set and make it produce true color, certainly that's possible: the default CMYK profile in Photoshop is designed to correct the lack of balance in process inks for commercial printing (although in that case it also requires the black to bring everything into balance). But to generate a profile like that would require much more sophisticated techniques than we're talking about here. The SWOP profile was developed for all the inks in combination with each other; that's how they learned that a profile that is skewed toward cyan is the right profile to give true color with those particular inks. But in our case we're talking about developing profiles for each of the colors individually and expecting magically that this will result in true color balance when the colors are printed together; I don't see how this could be a realistic expectation. You can hold back color density with a curve, but you can't add color density beyond what a given mix can print from a clear negative, so the only way you can possibly balance an unbalanced set is to balance to the lowest common denominator, and I don't see how that could result in balanced color.

What I've learned from my short but intense study of curves is what I've always known; the curves simply provide graphic representation of well-established gum facts. For example, when you get beyond a certain pigment concentration, no curve will allow you to get good separation in the darks and highlights simultaneously. Which is just another way of saying what we've always known: you can get a good DMax with a heavy pigment concentration of certain pigments, but such a pigment mix will always give you a very contrasty print. As I always say, you can have drama or subtlety, but not both in the same printing. This is true with or without curves; the curves simply verify the fact. With somewhat less pigment, you can get better separation at the ends of the scale by steepening the ends of the curve, but to do so you must necessarily flatten the midtones to accomplish this, so you sacrifice some midtone separation, and you've already sacrificed DMax with the dilution of the pigment. There is a concentration at which you can get good separation throughout the scale (with some pigments, not all) but again, you have to trade DMax for separation. There is a point of paleness of pigment where the midtones are already as flat as they can get; to steepen the ends of the curve any further would reverse the midtones, so at that point, the ends of the curve have to come in, in order to produce a viable curve. So pale pigments, like most yellows, or dilute concentrations of darker pigments, will have a fairly flat curve where one or both of the endpoints don't go all the way to the corner. But this is just another way of saying that pale pigment mixes will print fairly flat and won't give us much DMax, and we knew that without going to all the trouble of generating a curve to show us that. In each case you can verify the truth of the curve, and of the pigment mix, by applying the curve to the calibration chart and printing that, as Kees helpfully recommended. Where the ends of the curve have been steepened to improve tonal separation at the ends of the scale, the midtones will be flatter, both in the curve and in the tonal representations on the calibration chart, and so forth. You never get something without giving up something else. I've come to these conclusions not only from my own observations, but from looking at the curves and calibration charts that have been sent me by others that were generated from other systems. Whether in the end what you sacrifice is worth what you gain is a matter of personal choice and preference; at this point I'm on the fence about it. But I will say that after two weeks of printing these comparison images, the flatter curved prints have come to look better to me than the more contrasty uncurved prints, which wasn't true in the beginning.

So as not to lose the forest for the trees, I should probably say again explicitly that the main thing I've learned from this is that I still haven't seen a combination of pigment/concentration and curve that would allow one to print a dark DMax and nicely spaced tonal gradation throughout a long tonal scale, in one gum coat.

The comments above relate to monochrome printing only; I haven't begun to try the individual curves in tricolor printing, to test the hypothesis that one can automatically produce truer color with curves than without.


-----Original Message-----
From: Sandy King [mailto:sanking@clemson.edu]
Sent: 31 Ekim 2006 Salő 15:07
To: alt-photo-process-l@usask.ca
Subject: Re: gum curves


One of the advantages of making separations with digital negatives is
that you can actually make curves that will give true color from what
are essentially unbalanced tissue, pigment or dye sets. However, this
involves making curves for each color that correct or compensate for
the higher or lower contrast of the sets. I would not claim that this
can not be done without a sophisticated method of making curves, but
it would certainly be very hard to do so.

  • References: