Digitising film archives

Scanner, drum or camera: three ways to capture film

Hedgerows divide green and golden fields into irregular shapes.
Fields from the air.Photograph by Robert Harper. Robert Harper Photograph Library, Archive.London. From the archive, shown for illustration.

Every method of digitising film is a light source, a lens or sensor, and a way of holding the film flat. The differences are in speed, in handling and in what the numbers on the box leave out.

Every way of digitising film does the same three things: it shines light through the film, it records that light with a sensor, and it holds the film still and flat while it does so. Three families of equipment have grown up around those tasks. CCD scanners move a line of sensors past the film. Drum scanners spin the film past a single point of light. Cameras record the whole frame at once. An archive choosing between them is choosing how much time, handling and operator skill to spend on each frame.

That choice is rarely all or nothing. A collection like the analogue side of the Robert Harper Photograph Library, which mixes medium-format transparencies, black-and-white negatives, larger-format film, 35mm slides and contact sheets, may well call for more than one method.

How do flatbed and dedicated film scanners work?

A CCD scanner lights the film from one side and moves a linear sensor, or the film, in small steps, building the image line by line. Flatbeds with a transparency lid take film in plastic holders laid on the glass. Dedicated film scanners feed strips or mounted slides through a narrower, purpose-built light path.

Their advantages are cost and familiarity. One operator with a desktop scanner can work through mixed formats without specialist training, and many scanners offer infrared dust detection for colour film. The Canadian Heritage Information Network's guide, published on 25 May 2023, notes that this infrared cleaning has to be applied at the time of scanning and cannot be added afterwards.

Their weaknesses are speed and the gap between claimed and delivered resolution. The same guide tells operators to measure effective resolution with a test target, because results often fall short of the manufacturer's figure, and to disregard "interpolated" or "software enhanced" numbers. It suggests a scanner density range of at least 3.5, with 4.0 ideal and uncommon in affordable machines. That matters for transparencies, whose shadows are far denser than those of a negative.

Professional labs run higher-grade CCD equipment than a desktop flatbed. Metro Imaging in London, for example, lists CCD film scans from medium-format and 35mm frames alongside its drum service, and describes them on its scanning page as "not as high quality as our drum scans".

What is a drum scan, and when is it worth it?

A drum scanner mounts the film on a clear cylinder, spins it at speed and reads it one point at a time with a photomultiplier tube. The film is taped or fluid-mounted to the drum by hand, which keeps it flat and can suppress the visibility of fine scratches. Each frame is set up individually.

The result is a method with a long-standing reputation for shadow detail and tonal smoothness, and a cost structure to match. Labs price drum scans one by one, by output file size. Metro Imaging's published list (viewed 17 November 2026) starts at £47.90 for a 20 to 50 MB file and reaches £129.90 for files over 350 MB, taking the lower of the two figures shown for each band; the higher figure is 20 per cent more, consistent with VAT. Its CCD film scans on the same list start at £9.90 per frame.

Those prices tell you where drum scanning fits. It is a tool for the exhibition print, the book cover and the handful of frames where the last fraction of shadow detail is the point. It is not a way to work through a room of boxes.

How does camera scanning work?

Camera capture, often called camera scanning, places the film in a holder over an even, colour-accurate light source and photographs it with a digital camera and a flat-field macro lens on a rigid column. The whole frame is recorded in a single exposure.

Speed is the attraction. With no scan head to travel, capture time per frame is short and most of the cycle is film handling. This is why many volume programmes now use cameras. Heritage equipment makers now sell complete rigs: DT Heritage, for instance, describes a film scanning kit with magnetic-mask carriers for 35mm, 120 and sheet film, and states that it reaches FADGI four-star performance and meets or exceeds Metamorfoze requirements. Those are the manufacturer's claims for its own product, and they should be read as such.

The costs are less obvious. Resolution is fixed by the sensor, so a larger piece of film gets fewer pixels per inch unless it is shot in sections and stitched. Film flatness, alignment and stray light all depend on the rig and the operator. Colour negatives need converting after capture, since the camera records the orange-masked negative as it is. FADGI's guidelines accept either route for negatives, inverting at capture or capturing as a positive and inverting afterwards, and note that colour negatives need dedicated conversion. For small jobs, the browser-based negative to positive converter removes the mask and inverts a camera-scanned frame locally.

How do the three methods compare?

CCD scanner (flatbed or film scanner) Drum scanner Camera capture
How it records Line sensor, stepped Single point, rotating drum Whole frame, one exposure
Film holding Plastic holders or glass Hand-mounted on drum, often with fluid Carrier or glass over light source
Time per frame Slow to moderate Slowest; individual set-up Fastest
Operator skill Low to moderate High Moderate to high
Typical role Small volumes, mixed formats, access scans Selected frames for large prints Volume digitisation, survey passes
Main caution Claimed resolution often exceeds measured Cost per frame; wet handling Quality depends on the whole rig

The table deliberately has no "quality" row. Quality is a property of a particular machine, set up by a particular operator, on a particular day, and the only reliable way to state it is to measure it.

How do you test a capture system instead of trusting a label?

The cultural heritage imaging standards all work the same way: photograph a reference target, analyse the file, compare the numbers with tolerances. FADGI's guidelines (third edition, approved 10 May 2023) grade the result from one to four stars. For film from 35mm up to 4x5 inches the four-star level is about 4,000 ppi at 16 bits per channel; for 4x5 inches and larger it is about 2,000 ppi. The guidelines describe performance and say nothing about which family of machine must deliver it.

One gap is worth knowing about. ISO 19264-1:2021, the international standard for analysing cultural heritage imaging systems, covers reflective originals only; ISO's own summary says a future second part will deal with transparent originals. For film, FADGI's tables are therefore the most specific published benchmark an archive can cite.

In practice, ask any vendor for three things: a target-based conformance report from the rig that will do the work, sample files from your own film, and the unprocessed capture as well as the finished file. Then look for yourself. The image histogram tool shows whether highlights or shadows have been clipped, and the scan resolution calculator confirms whether the pixel dimensions match the resolution you were promised for the film size.

Which method suits a mixed archive?

For most collections the sensible answer is a combination: a fast method for the survey pass that makes everything visible, a measured preservation method for the selected frames, and a premium route for the few that will be printed large. Where each line falls is a budget question, explored in what it costs to scan 50,000 negatives, and the surrounding steps are set out in the box-to-file workflow.

The Robert Harper archive has not fixed its method. Its 52,684 analogue photographs, by the inventory's count, largely await scanning, and the mix of formats means the decision should follow a tested pilot. Vendors and imaging specialists who would like to propose an approach, or to run a comparison on a sample batch, are invited to start with the digitisation partners page.

Frequently asked questions

What is the best way to scan film: flatbed, drum or camera?

There is no single best method. Drum scanning is the slowest and most labour-intensive and is priced accordingly, so it suits selected frames. Flatbed and dedicated film scanners suit small volumes and mixed formats. Camera capture is the fastest per frame and is widely used for large collections. Judge any of them by measured results on a test target, not by type.

Is camera scanning good enough for archives?

It can be. Heritage equipment makers sell camera-based film systems that they state meet FADGI four-star and Metamorfoze requirements. Those are vendor claims, and the result depends on the lens, sensor, film holder, light source and operator. Ask for a conformance report produced on the actual rig, using a film-format target, before accepting the claim for your project.

What is drum scanning?

A drum scanner holds film on a rotating transparent cylinder and reads it point by point with a photomultiplier tube instead of a CCD or CMOS sensor. Film is usually mounted to the drum by hand, often with a mounting fluid. The process is slow and skilled, which is why labs price drum scans individually by output file size.

Does scanner dpi equal real resolution?

No. The dpi or ppi figure describes sampling, the number of pixels recorded per inch. Real resolving power depends on optics, focus and film flatness and is usually lower. The Canadian Heritage Information Network advises testing effective resolution with a target and ignoring interpolated or software-enhanced figures entirely.

Sources

Published by Archive.London. Corrections are welcome under our editorial policy.

Help digitise the archive

52,684 analogue originals, mostly medium-format transparencies, are waiting to be scanned and catalogued. We are speaking to scanning, cataloguing and tagging partners.

See the digitisation briefWrite to the archive