Three documents govern how the world's libraries judge a scan. Only one of them has published figures for film, and it marks several of those figures provisional. That is worth knowing before a star rating goes into a contract.
A star rating describes what a scanning system could do on the morning it was tested, and says nothing about frame 31,407 on a wet Thursday.
A FADGI star rating is a grade, from one to four, for how accurately an imaging system reproduces a test target. Metamorfoze is the Dutch equivalent, with three levels. ISO 19264 is the international standard that defines the measurements both rely on. Explanations usually present them as a ladder: pick a rung and write it into the contract. Our angle is the small print. These documents grade a system and not each image, they say little about film compared with paper, and the one that does publish figures for film marks some of them provisional. Anyone commissioning the scanning of an archive of transparencies, as we are preparing to do, needs to know where the standards are firm and where they are still being written.
What are FADGI, Metamorfoze and ISO 19264?
| FADGI | Metamorfoze | ISO 19264-1 | |
|---|---|---|---|
| Full name | Technical Guidelines for Digitizing Cultural Heritage Materials | Metamorfoze Preservation Imaging Guidelines | Photography, Archiving systems, Imaging systems quality analysis, Part 1: Reflective originals |
| Who publishes it | Federal Agencies Digital Guidelines Initiative, a United States interagency group | The Metamorfoze office, the Dutch national programme for preserving paper heritage, based at the national library | International Organization for Standardization |
| Current version | Third edition, published 10 May 2023 | Version 1.0, January 2012 | First edition, June 2021; a revision is at draft stage |
| Levels | One to four stars | Metamorfoze, Metamorfoze Light, Metamorfoze Extra Light | A, B and C |
| Written for | Many material types, with separate tables, including photographic film | Paper-based heritage: manuscripts, maps, books, newspapers, photo collections | Reflective originals only |
| Cost | Free | Free | Sold by ISO |
They are relatives, not rivals. Zeutschel, a scanner manufacturer, notes in its account of the history that the ISO work grew out of testing with a universal test target and out of the existing guidelines, Metamorfoze and FADGI among them, with a first ISO document in 2017. FADGI's third edition says its guidelines align with the ISO committee's work. All three share one method: photograph a calibrated target on the production system, analyse the file with software, and compare the numbers with published tolerances.
What does a FADGI star rating actually measure?
Not megapixels. The third edition lists the criteria that are evaluated from the target, and they describe how an imaging system behaves, not how large its files are:
- Sampling frequency: the true pixels per inch at the original.
- Spatial frequency response: how much real detail those pixels resolve, which is not the same thing. A soft lens at 4,000 ppi can resolve less than a sharp one at 2,500.
- Tone response: whether light and dark values land where they should.
- White balance error and colour accuracy.
- Lightness uniformity: evenness of illumination across the frame.
- Colour channel misregistration: colour fringing.
- Noise, sharpening (whether any has been applied) and reproduction scale accuracy.
- Skew, rotation and geometric distortion.
Some of these are required for a star level and some are reported for information only. The guidelines also specify, by material type, a master file format, bit depth and colour space for each level, and the third edition added a section on the professional staff a programme needs.
FADGI's own descriptions of the levels are refreshingly blunt.
| Level | FADGI's description | Nominal ISO 19264 level |
|---|---|---|
| One star | For reference to locate the original, or when higher levels are not possible; "for reference use only" | None; does not conform |
| Two stars | The minimally acceptable rating for most professional digitisation | C |
| Three stars | "A very good professional image that is appropriate for most uses" | B |
| Four stars | The best imaging practical today; often suited to smaller projects, and not advised for all but the most sophisticated programmes | A |
How do the standards treat film?
Unequally, and this is the gap that matters to a photographic archive.
FADGI has tables for transparencies and negatives. For film from 35mm up to 4 x 5 inches the levels step from about 1,000 ppi at one star through 2,000 and 3,000 to about 4,000 ppi at four stars, with 16 bits per channel at three and four stars. For film of 4 x 5 inches and larger the top figure is about 2,000 ppi, because the film itself is larger. Our scan resolution calculator turns those figures into pixel dimensions for any frame size.
But the third edition is candid that its transmissive figures are less settled than its reflective ones. It describes the tone response values for transmissive materials as provisional, along with white balance and lightness uniformity for those types. Its measures of highlight and shadow tolerance are called an emerging area with informational values, on the reasonable ground that a transparency is itself an interpretation of a scene whose true colours are gone. Dynamic range is informative only. The guidelines do give firm practical direction, for instance that light sources for transparencies should be 5,000 K with a colour rendering index above 90.
ISO 19264-1:2021 does not cover film at all. Its title says reflective originals, and ISO's summary states that "a future part two will address imaging of transparent originals".
Metamorfoze was written for a paper preservation programme. Its top level is intended for originals treated as works of art, a category that includes photo collections, but its criteria and targets are those of reflective capture.
The consequence is simple. For film, FADGI's tables are the most specific published benchmark an archive can cite, and parts of that benchmark are explicitly provisional. A brief should say so, not paper over it.
What do the standards not measure?
A conformance report tells you that the camera or scanner, as set up on the day of the test, could meet a level. It does not tell you about anything else, and most of what goes wrong in a large scanning job is something else.
- Each individual frame. The target is typically captured at set-up and at intervals, not with every exposure. Metamorfoze's two higher levels are the exception in requiring a chart in every image, which is practical for a manuscript page and awkward for a 35mm slide.
- Focus and flatness on the day. A buckled transparency or a holder a fraction out of true will soften a scan from a four-star system.
- Dust, scratches and fingerprints.
- Completeness and order. Whether every frame in the sleeve was captured, the right way round, and named correctly.
- The condition of the original. A faded slide scanned accurately is an accurate record of a faded slide. Restoration is a separate, later decision, discussed in restoring faded slides digitally.
- Colour negatives. Inverting an orange-masked negative into a positive involves interpretation that no target fully governs.
- Metadata. Nothing in an imaging standard checks that a caption is right.
Colour management of film has its own tools and its own limits, which we cover in IT8 targets and colour management for film.
Which level should you ask for?
The honest answer for most film archives is three stars for the collection and four for selected frames, and FADGI's own wording supports it. Three stars is "appropriate for most uses". Four is "often suited to smaller projects".
There are costs on both sides of that line. Moving from 3,000 to 4,000 ppi nearly doubles the pixel count of every file, with the storage and transfer that follow. Meeting the tighter colour and noise tolerances can mean slower capture. Digital Transitions Heritage, an equipment supplier, summarises the top-level tolerances as an average colour difference of 3 and a maximum of 6, measured as Delta E 2000, for FADGI four star, against an average of 4 and a maximum of 10, measured as Delta E 1976, for the strictest Metamorfoze level. The two use different formulas and are not directly comparable. The same summary maps Metamorfoze Extra Light, Light and full Metamorfoze roughly onto FADGI two, three and four stars. Those are a vendor's characterisations and should be checked against the source documents before they are relied on.
One real difference in philosophy is worth using. FADGI allows a project to set criteria independently: three-star resolution with four-star colour, say, where colour fidelity matters more than enlargement. Metamorfoze's levels are applied as a whole.
How do you cite a standard in a brief without promising the impossible?
By being specific about five things and modest about one.
- Name the document and edition. "FADGI Technical Guidelines for Digitizing Cultural Heritage Materials, third edition, May 2023."
- Name the material category and level. "Photographic transparencies, 35mm to 4 x 5 inches, three star."
- Restate the numbers you mean, so nobody has to interpret the reference: resolution, bit depth, colour space, file format.
- Say how conformance will be shown. A target captured on the production equipment, analysed with software that implements the ISO 19264 methods, with the report supplied at set-up and at agreed intervals. FADGI permits any tools that meet ISO 19264 and provide the required analysis.
- Add what the standard does not cover. Handling, dust, completeness, naming, metadata and acceptance sampling need their own clauses.
- Acknowledge what is provisional. For film, state which FADGI criteria you are treating as requirements and which as informational, since the guideline itself draws that line.
And avoid three phrases. "FADGI compliant", with no level or category, means nothing. "ISO 19264 certified film scanning" cites a standard that does not yet cover film. "Four star throughout" for tens of thousands of frames commits both parties to a level its authors caution against at scale. A full template is in how to write a digitisation tender brief.
Why this matters to people who will never hold the film
A capture standard is also documentation. A picture desk, a publisher or a data team receiving scans cannot inspect the scanner, but it can read a statement that the files were made to a named level, on a named date, with a report attached. For image datasets in particular, capture conditions are part of provenance: resolution, bit depth, colour space and whether any sharpening or restoration was applied all affect what a model learns from. That is why capture details belong in a dataset card alongside rights and counts.
Where the Robert Harper archive stands
We should be exact about our own position. The inventory records one folder of 842 transparencies as already scanned. More than half of the archive is still on film and has yet to be scanned, and the principal analogue format is 6 x 7 colour transparency film, with sheet film, 35mm slides, black-and-white negatives and 4,005 Polaroid prints alongside. No capture standard has yet been applied across the collection, and we make no claim that existing scans meet any star level. The folder-by-folder figures are in the archive inventory.
That is the practical reason for this article. When the specification for digitising the library is written, it will have to cite these documents accurately. If you are a digitisation company or an institutional imaging studio with experience of working to FADGI levels on film, we would like to hear how you demonstrate conformance; the page for digitisation partners explains how to reach us.
Frequently asked questions
What does a FADGI star rating mean?
It is a grade for the measured performance of an imaging system, from one star to four. FADGI calls three stars a very good professional image appropriate for most uses, and four stars the best imaging practical today. The grade comes from analysing a photographed test target against tolerances for resolution, tone, colour, noise and uniformity.
Is FADGI four star necessary for film scanning?
Usually not for a whole archive. FADGI itself says four stars is often suited to smaller projects and cautions against it for all but the most sophisticated programmes. Three stars, about 3,000 ppi at 16 bits per channel for film up to 4 x 5 inches, is described as appropriate for most uses.
Does ISO 19264 apply to film and transparencies?
Not yet. ISO 19264-1:2021 is titled for reflective originals, such as prints, documents and artwork. ISO's summary states that a future part two will address the imaging of transparent originals. A contract for film scanning that cites ISO 19264 alone is therefore citing a standard that does not cover the material.
What is the difference between FADGI and Metamorfoze?
FADGI is a United States federal guideline with four star levels and tables for many material types, including film. Metamorfoze is a Dutch guideline written for paper heritage, with three fixed levels. Both rest on the same measurement methods. FADGI lets a project set different levels for different criteria; Metamorfoze levels are applied whole.
Can a scanning company be FADGI certified?
FADGI publishes guidelines and free analysis tools; as far as we can establish it does not certify companies or machines. What a vendor can show is a conformance report: an analysis of a test target captured on its production equipment, with the edition, the star level and the material category stated. Ask for the report, not the adjective.
Sources
- FADGI: Technical Guidelines for Digitizing Cultural Heritage Materials, third edition (May 2023)
- Library of Congress, The Signal: FADGI third edition announcement (May 2023)
- ISO 19264-1:2021 Photography, Archiving systems, Imaging systems quality analysis, Part 1: Reflective originals
- Zeutschel: Metamorfoze, background, goals and content
- Digital Transitions Heritage: Overview of FADGI and Metamorfoze guidelines
Published by Archive.London. Corrections are welcome under our editorial policy.

