Reducing the number of variables

For the moment I have decided to scan all my black and white negatives with a simple rule. Turning off the scanner software's auto exposure, the bottom of the histogram will be 0 and the top 255, with an gamma of 1.25 and the tone curve set to linear.

This allows me to 1) reduce the number of variables so that I can learn more easily the effects of manipulation in GIMP, and 2) should, in theory, be a more constant, predictable and and perhaps accurate representation of the characteristics of the film itself.

Another comparison

A small portion of the same scene.

First the LX3:

then the RZ:

Colour printing

I'm glad I'm not printing this in the darkroom: I've tried over 20 variation of the scan settings, which, if this were colour paper, would have cost a small fortune. I'd have been ducking in and out of the darkroom to see what I'd done, and since it takes a while for my eyes to adjust, this takes time and introduces error. I also couldn't have been writing this at the same time as the print was developing (the scanner is scanning at the moment).

Here are the thumbnails of some of the variants. The differences between them are generated, in some cases, by very small changes in the scan settings (e.g., from 236 to 237 in one of the blue channel parameters).  In just one panel - the AD coding 'histogram', there are 5 settings per channel, so there are a lot of variables to mess with.     


This is the final product. It probably has slightly more blue in the highlights than the original scene, at least as I remember it (it was late evening as the sun was setting and the light was fairly yellow), but it seemed more pleasing to me than a possibly more accurate representation.

Fuji Reala 100

Two rolls of color negative film came back today. It's much harder to work with than black & white because the film itself is so dense which means the information is compressed into a much smaller range. That makes the tonal gradation rougher and reduces the margin of error. This is visible in the 'sudden' transitions between adjacent tones. Here is a comparison with my LX-3.  First, LX3 image:

Next the film version (Fuji Reala 100).

I'm not too happy about either image to be honest. The lower one seems to me to have a little to much cyan, while the upper is a shade too yellow. The lower image could also be a little flatter. But this was my first attempt and it's much harder than black and white - obviously - since there are many more variables. But this is all about learning the craft. The road is long, but the path is clear (sounds like a quote but Google won't tell me whose).

P.S. Seems I left the white balance setting on the LX3 to "house with shadow", which according to the user's manual is for "when taking pictures outdoors in the shade", which wasn't the case here. So the odd tint in the LX3 is my fault. More tests (and care) are clearly warranted. 

Digital Analog(Digital) Comparison

Two images of the same scene. The top one was take with my Lumix LX3, the lower image with the RZ-67 and Acros 100.

The film was scanned at 1200 dpi which gives it about the same sized file and resolution as the digital end-to-end below.

Putting the 'S' in

The Epson V700's software produces what appear at a glance to be good 'exposures' when set on automatic, but on closer inspection have some serious problems. The highlights are very flat and often blown out and shadows compressed to oblivion. This means doing the A-D mapping from emulsion to bits by hand.

Without intervention, the scanner uses a logarithmic function for reading negatives. The result over emphasizes shadow and highlight detail while compressing the mid tones (see below).

Straight from the scanner

While all the information is here, the balance between the mid gradients and the ends (light and dark) doesn't look right so some tweaking of tone curve is needed. The mid tones need separating and this generally means squashing the highlights and the shadows, making an 'S' curve (or something like a cumulative normal distribution).     

In the first, the contrast in the sky is flattened and a little in the shadows too, while the the tree trunks are opened up.

    
Here's another interpretation of the image, a little more somber. The shadow detail is still preserved but the tree on the left in the sun is darker. The sky also retains slightly more contrast. This comes at the expense of the mid tone separation.

Whether the sky was or should be portrayed as that dramatic is a choice. The upper picture emphasizes the light on the tree on the right which was caught when the clouds opened temporarily. The lower one has more texture and a more threatening sky. I think my preference is for this last one.   

I'd probably crop some of the sky if I were printing this, (not least because of the streaking at the top which I think was from lack of agitation in processing). 6x7 is squarer than 8x10, and some cropping would be expected, but the slightly unusual squareness has an odd appeal. Perhaps that's just because it's late and I'm tired.     

My light meter

I am still debating whether I should take my Honeywell Pentax spot meter with me. The answer is probably yes but for the moment I've been using my trusty Lumix LX3 as a meter. Since I take it with me everywhere, and the results are fine, it's one less thing to carry. This means  giving up the whole idea of pre-visualization and zone placing and relying to some extent on the latitude of the film and 'post processing'. But since I'm moving away from the zone system which 1) I never mastered and 2) is largely about properties of emulsion and chemical film developing which I no longer do, this may not be too much of a problem. We shall see.