U of S | Mailing List Archive | alt-photo-process-l | Re: Two tricolor prints

Re: Two tricolor prints

What this reminds me of, a little bit, is regression statistics. Every spreadsheet program nowadays comes with a regression program built in, so it's easy to run any kind of junk data through a regression program, and people are always doing that and thinking that what comes out the other end must mean something, if it's got a significant beta attached to it. But if you take time to get into the data and look at the descriptive statistics, for example, you often find that the outcome of the regression program has no logical connection to the actual data, but is an artifact, a result of the fact that much if not most available data produce unstable betas that bounce around randomly and sometimes fall on the winning number by accident.

Not that I think curves are bogus, or a matter of chance; I'm not carrying the analogy that far. I only mean that in the same way, I think people sometimes crank their way through a system for producing curves, and assume that the curve that results is a good one. But if you really look at the print tones themselves, you see what you sacrifice to gain something else; it's a matter of deciding whether the sacrifice is worth the gain. And obviously, though my calibration works well for individual layers, it doesn't work very well for tricolor gums, which was the whole point of this demonstration.