Every published object is open access.◆No gatekeeping. All objects published openly.◆PHOTARCH — research infrastructure for archaeology.◆Every published object is open access.◆No gatekeeping. All objects published openly.◆PHOTARCH — research infrastructure for archaeology.◆

Metadata

Why a file should be able to explain itself

In archaeological documentation, an image without context is not data — it is decoration. A photograph that leaves its record behind in a database is one database migration away from meaning nothing. PHOTARCH's answer is that the file carries its own documentation: what the object is, who photographed it, what may be done with the picture, how large the object is, and whether anyone has altered the file since.

What a PHOTARCH file carries

SHA-256 C2PA manifest what was done · the scale, signed · a timestamp XMP packet dc · Iptc4xmpExt · photarch · ism Pixel data the picture, untouched by all of the above The scale is written twice In the XMP, where any program can read it — and any program can change it. Inside the seal, where it cannot be changed at all, only broken, and a reader is told which. A program that knows none of this sees the bottom band and shows the picture. That is the point of putting it in the file.
The SHA-256 is taken over the whole master file and recorded in the archive, so the master can be checked years later. A downloaded copy has its record written into it and gives another checksum; its seal is what proves it.

The record: Dublin Core, IPTC, and PHOTARCH's own fields

Every exported picture carries an XMP packet. Dublin Core gives the base layer any system understands — title, creator, rights, source. IPTC Extension carries the place and the holding institution. PHOTARCH's own namespace carries what the others have no room for: inventory number, period, material, object type, technique, condition, and the object's dimensions in millimetres.

Camera-generated EXIF is preserved alongside it. PHOTARCH reads EXIF and never rewrites it: what the camera recorded is what the camera recorded.

The measurement: Image Scale Metadata

A photograph only shows an object's size if the viewer knows how large one pixel is on the object. That number is written into every exported file of a measured object as Image Scale Metadata: millimetres per pixel, the pixel grid it applies to, how it was measured — from the board's markers, from a ruler in the frame, or with calipers on the object — and an uncertainty figure with the assumption behind it spelled out.

The grid matters as much as the number. A scale that describes the master's pixels is wrong on a smaller copy of it, so every file states the grid its own scale belongs to. A reader can then rescale correctly when the file is resized, and recognise a crop to another shape and refuse to guess.

The seal: Content Credentials

A file from the archive, or exported under a paid licence, also carries a C2PA manifest — a signed statement of how the picture was made, with the scale inside it, countersigned in time by an independent authority. The difference from the XMP is the whole reason both exist: the XMP can be corrected, improved and also quietly falsified; the manifest can only be broken, and a reader is told when it has been.

The fingerprint: SHA-256

Every master is given a SHA-256 checksum when it enters the archive, shown on the object's record and included in bulk exports. It answers one question — is this file byte for byte the one that was published? — and it answers it without needing us.

Structured fields for human and machine

The fields PHOTARCH asks for are the ones that make an image usable as evidence rather than illustration:

  • Site, find context, and stratigraphic unit
  • Typology, material, period, culture — mapped to the Getty AAT where a term exists
  • Object dimensions in millimetres, and the scale of the picture itself
  • Photographer, institution, and the licence the image travels under
  • Camera, lens, light source and light angle

Because they are embedded rather than referenced, they survive the journey: into a collection system, onto a hard drive, into a publication, and out again.

References

  • schemaDublin Core Metadata Initiative. DCMI Metadata Terms.
  • standardISO 16684-1:2019. Extensible metadata platform (XMP).The packet the record travels in, inside JPEG, TIFF and PNG.
  • formatImage Scale Metadata (ISM) 0.1, draft specification.Conformance levels A, B and C: the scale, what it is worth, and where the object lies on the grid.
  • specificationC2PA. Content Credentials specification.
  • vocabularyGetty Research Institute. Art & Architecture Thesaurus.
← StandardsTrust & Provenance →