Sunday, June 1, 2014

Calibration

Most of the work I've been doing recently is about process standardization and reproducibility (quality in the operations management lexicon). Last year I systematized the time/temperature calculation creating a web-based app that uses data on the film, developer, dilution and temperature to calculate  the appropriate development time. Hit and miss is (thankfully) a thing of the past. More recently I've been focusing on calibrating the film, much as one calibrates a display or a printer.

Each emulsion responds to light in a unique way. Thankfully Kodak's process control is such that I haven't yet found differences between batches (although I'm not sure the films I've shot weren't actually from the same batch).

To begin I create a negative with bands of different exposure raging from seven stops under exposed to seven stops over. This is about the dynamic range of TMX-100. I develop the film and scan it with a linear tone curve, setting the black and white ends of the output to the darkest and lightest bands on the negative.

I then use GIMP to measure the level of grey the scan has generated for each level of illumination on the negative. From this I generate a series of tone curves (49 in all) for different ranges of emulsion density to compensate for the emulsion's characteristics at different exposures.

For example, one curve might correspond to a negative in which the shadows were 3 stops below middle gray and the highlights 5 stops above middle gray. Rather than labeling them in stops over and under, I label each curve for the range of measured emulsion densities. So in this example, the tone curve would be the one that captures emulsion densities from 143 to 232. (These are all out of 255, the 8 bit tonal representation in the digital image).Doing so allows me to choose the appropriate curve simply by looking at the histogram of the each negative.


TMX-100 measurements


The red line on the chart on the right shows the density of the TMX-100 emulsion for different EV levels. The response is clearly non-linear. At light levels above middle gray, (EV>0) there are only small differences in density; at low light levels the differences in density increase at each step until 5 stops under when they decline sharply.


This suggests first that underexposed areas, the shadows, will be hard to deal with unless the response curve is exactly matched to the exposure; and it reenforces the dictum - "expose for the shadows, develop for the highlights", which helps ensure one doesn't have to deal with the non-linear behavior of the emulsion for very low light levels (-5 EV and below).

The curved blue line shows the intensity of the scanned image for each exposure level. Because the negative was scanned with a gamma of 1.5 the line curves upwards rather than being straight.

Using Microsoft Excel, the tone curves are written into directly into the Windows registry keys for the Epson scanner; when the scanning software is opened the tone curve drop-down box is populated with the generated tone curves. Using them is fairly simple. Looking at the histogram for each negative, the levels for the darkest and lightest points are determined - since each tone curve covers a different range of emulsion densities this measurement determines which generated tone curve to use.