Eric Radman : a Journal

Scanning Film: B&W, Color, and Reversal

At one time the measure of a good negative was the ease or difficulty it provided in making a paper print with an enlarger. Today we still value a negative that prints easily, but in most cases ease of scanning is even more important because the digital representation is the first and maybe only way that other people will see the photograph.

Scanning with a Digital Camera

For 35mm and medium format I prefer a digital camera over a consumer-grade scanner for several reasons: speed and control over exposure.

To get started with macro photography you'll need a tripod with a reversing post such as the Slik 700DX Pro and a light pad with a high color reproducability such as the Kaiser Slimlite Plano 8x11.

Always use the 2-second self-timer to get to get a sharp image. You probably need to use manual focus, but contrast-based systems are also good enough. I usually take macro shots at f5.6 or f8.

B&W Negatives

Black and white negatives capture a relatively wide tonal range which does not map the linearly to the digital camera's dynamic range. This means that in post you will almost certainly want to adjust the shadows and highlights using an S curve which vaguely mimics the restrainer in the highlights and depletion of halides of shadows on photo paper.

Converting the negative to a positive is easy, but applying the tone curve is a problem because tones that are stretched can show signs of banding. The only way to avoid visible degradation is experimentation. I would say that it is imperative to leave a margin at the white and black points during your initial curve adjustment so that highlight and shadow detail are preserved curing subsequent curve adjustments.

Color Negatives

The orange mask on color negative film is a hindrance to capture, not because it shifts the color but because it causes clipping in the red and blue channels. To bring each channel closer to the center I use a tungsten-to-dalight (blue 80A) filter. The following histograms shows the difference

80A filter colorcube comparison

(You loose two stops of light using this filter, so skip this if you need to.)

Color adjustment is the second step. If you are in a hurry you can use the filter such as Auto White-Balance (as it's called in Gimp) which will stretch each channel. The problem with this is that you may clip areas of low and high intensity. To counter-act this you may have to set each channel's white and back point manually. Oh course we can use both methods: adjust manually and then use and Auto-something filter on a second layer who's opacity can be adjusted.

Final color correction is hard. In Gimp you might get close by using Mid-point eyedropper in Levels. More likely you will need to adjust curves. I have found the following envalope to work reasonably well with Fuji Pro 400H

Color Negative Adjustment curves

This is a preset I save called my Color Envalope gives you six control points, two for each channel that work by giving you separate control over compression of hilights and shadows as well as relative saturation.

Color Reversal

Slide film is a real joy to inspect by eye, but requires creativity to scan because it has very fine details embedded in the shadows, and because even the best digital cannot capture over-exposed colors with any reasonable degree of fidelity. This second point provides a clue about why color film can still be so compelling.

Now on to scanning: I always auto-bracket by 1 stop. If you're in a hurry you can simply pick the best of the three for each frame. This isn't usually what happens. More commonly two of these exposures will be selected for a selective merge using a luminosity mask. Using luminosity masks we can create a kind of high-dynamic range image that hopefully comes close to what we see with our eyes.

I have no formula for this step, but in often I start with darker image and add an overexposed image to bring back the shadow detail. I may use any combination of masks ranging covering very light to very dark, but I try to stick with three layers or less for mental simplicity.

Some Scripts

There are a few tasks to do with photo processing can be automated. The first is the removal of extra channels that Gimp leaves in a TIFF after editing

#!/bin/sh -ex
mkdir _
for f in *.tiff; do
   convert $f -background black -alpha off _/$f
mv _/* .
rmdir _

Always write images to a temporary location in case something in my script runs off the rails mid-process. Some others

 # Convert to B&W
 convert -colorspace Gray $src _/$dst

 # Invert
 convert -negate $src _/$dst

 # Resize to 2800px wide, preserving aspect ratio
 convert -resize 2800 $src _/$dst

 # Rotate
 convert -rotate 180 $src _/$dst

 # Rename images sequentially (DSCF888.tiff -> 01-888.tiff)
 for img in DSCF*.tiff; do
     let n+=1
     mv $img $(printf '%02d' $n)-${img##DSCF}

Notes about 645

35mm negatives are famously sized at exactly 36x24mm, but the 645 format is does not have obvious dimentions. I find that the actual usable area is about 56x41mm (not quite 56x42). After cropping I typically scale the image down to the equivalent of 50 or 60 pixels/mm

Pixels/mm 56 41 MP
30 1680 1230 2.1
40 2240 1640 3.7
50 2800 2050 5.7
60 3360 2460 8.3
80 4480 3280 14.7
100 5600 4100 23.0

Last updated on October 13, 2018