Photographing negatives: using a mirror to align camera and film in parallel

The film holder is set up, the macro lens is in place, everything in the center of the frame is razor sharp, and the lower right corner stays mushy. So you stop down again. Still. At some point you start suspecting the lens.

Most of the time it's not the lens. It's the angle.

Why one corner goes soft

When photographing negatives, you're imaging a plane onto a plane. For that to work across the whole frame, the sensor plane and the film plane have to be parallel to each other. If either one tilts, the plane of focus cuts through the film diagonally: in the center it still sits right, at one edge it lies above the film, at the opposite edge below it.

You can calculate how much tilt this tolerates. For the close-up range, the following applies approximately:

Depth of field ≈ 2 × f-number × circle of confusion × (1 + magnification) ÷ magnification²

A 35mm negative on a full-frame sensor means a magnification of 1:1. At f/8 and with the classic circle of confusion of 0.03 mm, you get just under 0.96 mm, which is the entire range in which something looks sharp, in front of and behind the focal plane combined. Across the 36 mm width of a 35mm negative, that corresponds to a tilt of about one and a half degrees.

The catch is that this classic circle of confusion is intended for postcard-sized prints. If you look at your scan at 100 percent on the monitor afterwards, and everyone does when digitizing, you end up closer to 0.01 mm. Then the depth of field shrinks to 0.32 mm and the permissible tilt to half a degree.

Half a degree is not much. A ball head clamp that slips ever so slightly when you tighten it already uses that up.

Stopping down is not a fix

The obvious reflex is to stop down further. That really does increase the depth of field, but it gives the gain back through diffraction, especially quickly at 1:1, because the effective aperture doubles there. A set f/16 effectively becomes f/32, and the diffraction disc grows to around 43 µm. On a 24-megapixel full-frame camera, the pixel pitch is about 6 µm. So the disc covers a good seven pixels. You're trading two soft corners for an image that's soft throughout.

Why a spirit level isn't enough

The usual attempt: a bubble level on the film holder, a bubble level on the camera, get both level, done. The problem isn't the idea, it's the chain. You measure twice against gravity and hope that this gives you the right angle between the two planes. Both measurements have an error, and the errors add up.

On top of that, there's a reference problem. The sensor plane sits inside the body and isn't necessarily exactly parallel to the tripod plate the bubble level rests on. The film holder doesn't have to be exactly parallel to the tabletop. So the spirit level measures two things that both matter to you only indirectly. Manufacturers of scanning accessories classify the built-in electronic levels accordingly: as a rough check and fallback, not as a reference.

The mirror directly measures what actually matters, the angle between sensor and film.

The mirror trick

Place a small mirror where the film would normally lie and look through the camera. What you see is your own lens.

If the optical axis is perpendicular to the mirror, the light path returns on itself and the reflection sits exactly in the center of the frame. If the camera is tilted, it drifts out. That's all there is to it. And because the sensor is by design perpendicular to the optical axis, at that moment it is parallel to the mirror and therefore parallel to the film.

The real trick lies in the law of reflection: if you tilt the mirror by an angle, the reflected beam rotates by twice that angle. So the method shows you the error twice as large as it actually is. That's exactly why it resolves angles where a bubble level fails.

Step by step

1. Set up the film holder exactly as you will when scanning later: same height, same mask, same base. Place the mirror in the center of the film stage.
2. Camera straight above it, switch on Live View, turn on the grid lines. Almost every camera has them in the display menu.
3. Focus on the reflection, not on the mirror. What should be sharp is the front element of your own lens that you see in it. For this, the focus sits at roughly twice the distance.
4. Stop down, f/8 to f/16. The aperture opening gets smaller and therefore more precise as a target point.
5. Adjust the tripod head until the aperture opening sits exactly in the center of the frame. A geared head is more pleasant here than a ball head, because you can fine-tune each axis separately.
6. Remove the mirror, insert the film, refocus on the film. After that, don't touch the tripod head again.

If you can barely make anything out: place a small lamp to the side of the mirror, shining upward. The lens gets lit and stands out clearly in the dark Live View. A phone with its flashlight switched on is all you need.

The cross-check

Aligned doesn't mean proven. Instead of the negative, insert a test chart, a USAF-1951 target or, failing that, a printed page with fine text, take a shot and view it on the computer at 100 percent with sharpening turned off. Check the center and all four corners. If the corners are even, the angle is right. If one corner is worse than the diagonally opposite one, the camera is still tilting in that direction.

What tends to go wrong

• The mirror isn't lying flat. A speck of dust under the edge, a wavy adhesive film, a spring-loaded film stage, and you end up aligning against a tilted reference. Wipe the surface beforehand.
• The mirror isn't where the film is. With multi-level holders, what counts is the level where the negative will later sit, not the base plate underneath.
• Turning the tilt head after aligning. Adjusting the height is not critical as long as the column is truly vertical. Turning the tilt head undoes the work.
• Aligned once, aligned forever. Not true. Check again after every rebuild, every transport and every lens change. The check takes a minute.

What you need

Almost nothing. A small mirror is enough as long as it is flat and extends across the center of the frame; an edge length of 40 to 50 mm is comfortable to handle. A pocket mirror from the drugstore will do, and so will a piece of mirror acrylic from a craft shop.

Full disclosure: we build film holders and copy stands ourselves, and we recently added an alignment mirror to our shelves because the question reached us often enough. You don't need it; the trick works with any flat mirror. What matters is that you actually do it at least once.

Further reading and watching

• DSLR film scanner mirror leveling method, just under a minute, someone demonstrates it with a geared head. In English and almost without words, but you can see exactly what matters.
• Valoi: How to scan with a digital camera, the camera scanning workflow from start to finish, with the mirror step as part of the routine.
• Camera Scanning Knowledge Center: How do I guarantee the sensor and film are perfectly parallel?, the same method plus a cross-check with a test chart.
• pixl-latr: Alignment and focus for film digitisation. A variant for setups without a stand column.
• Exaframe: Guide for batch scanning film with a digital camera, for when you work through whole rolls in one go and the alignment has to stay put.
• Photrio: Alignment for film digitization. A discussion among people who do this every day, including the edge cases.