gum curves -- corrections
A couple of weeks ago, I posted a comparison between a print made from an uncurved negative and one made from a negative with a curve generated with ChartThrob. The curved print was much flatter and much darker than the uncurved print. It turns out that was a misleading comparison for two reasons, which illustrate two principles I've posted reflections about here recently: (1) the two prints were printed on different days, with very different environmental conditions. Ordinarily I would adjust the exposure to accommodate a change in humidity as a matter of course, but I was caught up in this idea about standardization, which shows what a powerful idea it is, and how misleading it can be. The humidity was 20% higher when I printed the curved print than when I printed the uncurved print, and the curved print was simply way overexposed by using the same exposure time. I owe this insight to an offlist correspondent who kept telling me that my curve was so close to a straight line that the prints should be more like each other, so finally I did a side by side print, using the same negatives as before and printing on the same piece of paper with the same emulsion, under the same environmental conditions, everything the same. But this comparison still favored the uncurved print, as did several others I did the same way, which leads to principle #2: As I said here recently, in any comparison, the protocol that you're calibrated for is the protocol that will win. The protocol I'm calibrated for is uncurved negatives. This light bulb went off when I was studying some comparison charts and curves sent me by another offlist correspondent; in his comparison, the curved print looked better, and the uncurved print didn't look good at all. It was fairly obvious from a consideration my comparison and his together, that the two kinds of negative don't print just the same, and shouldn't be compared using the same protocol. If they are, then the comparison will favor whichever one the protocol is calibrated for. So I did another comparison side by side on the same paper with the same coating, same printing, but this time, instead of keeping the two prints together and following my natural tendency, which would be to use a development protocol that would work well for the uncurved print, but maybe not so well for the curved one, I cut the prints apart and allowed them to develop independently. It's not perfect; I got busy doing something else and left the uncurved print a bit too long, but the main thing to notice is that by standardizing the results by varying the protocol, I got what seems a fairer comparison between the curved and the uncurved print, that shows the advantage of the curved negative. It's a fairly subtle difference, but it's there. http://www.pacifier.com/~kthayer/html/Curvecomp.html I also made a statement in response to Keith's request for people to send him curves, that I now believe to be in error. I was simply repeating what I thought I'd heard here from curve proponents, that every pigment needs a different curve, every specific protocol requires a different curve. But after two weeks of generating a lot of curves, and talking to other people who are familiar with curves, I've come to feel I was mistaken when I repeated that statement without checking it out myself. I'll post more on that with images in due time. Katharine
|