Glossary

Standard

ABFO No. 2

The ABFO No. 2 standard reference scale

The ABFO No. 2 is the reference scale of forensic photography. The engineer William Hyzer and the forensic odontologist Thomas Krauss published it in 1988 for the American Board of Forensic Odontology, to photograph bite marks. It is L-shaped, with millimetre graduations on both legs, so that a grid can be projected over the plane the ruler lies in. Three circles 80 mm apart show whether the camera is square to it: a tilted camera draws them as ellipses. A grey patch is part of the specification too.

The specification sets the graduations to ±0.1 mm, or ±1 % at the centimetre marks, the spacing of the circles to ±0.25 %, and the legs square to within two minutes of arc. When the US National Institute of Standards and Technology measured scales from four makers, published in 2013, every length scale met the specification. The circles were often off size, and more than half the rulers were not square; their squareness also drifted over four years.

The scale is accurate in its own plane and says nothing about what lies above or below it. Forensic guidance — the board's own guidelines of 2018, SWGDE's of 2019, ANSI/ASB 050 of 2021 — therefore puts it in the plane of the evidence, with the camera parallel to it. The PHOTARCH method follows the same rule with a caliper on the object's outline; a scale taken at the surface the object rests on is corrected to half the object's height, and what is left is stated as a margin.

On this site Measuring from photographs: a history · Scale in artefact photography

See also Plumb view (F) · Relief displacement · Error margin · GUM

Concept

Access copy

An access copy is a file made from the master for a particular use — a web page, a report, a loan request — and made smaller, compressed and converted to suit it. FADGI calls such files derivatives and lists the other names they go by: service, access, delivery, viewing or output files. They are secondary by nature, not generally a permanent part of the collection, and those made for end users often use lossy compression such as JPEG. Metamorfoze calls them user files, and puts processing such as stitching there rather than in the master.

Because it can always be made again from the master, an access copy carries nothing that cannot be lost. A mistake is put right by making a new copy, not by editing the old one: a JPEG loses a little more each time it is edited and saved.

In the PHOTARCH method the access copy is an 8-bit JPEG in sRGB with its profile embedded, which looks the same in almost every browser and on almost every screen. Any whitening of the ground is done on copies only. The PHOTARCH app exports one beside the master. Downloads from the archive are copies too: each is a new file, with the object's record written into it as XMP at the moment it is made and sealed with C2PA, so it has a checksum of its own, different from the master's.

On this site Guides · Technical requirements · How to photograph archaeological artefacts · Colour accuracy and the ColorChecker · File formats for archival photography · Manual · Export — and what an image is fit for

See also Master · JPEG · Colour space (sRGB, Adobe RGB) · Lossless and lossy compression

Standard

ARK

Archival Resource Key

An ARK, or Archival Resource Key, is a persistent identifier: a name for an object that is meant to keep leading to it after web addresses have changed. It has the form ark:/NAAN/name — in PHOTARCH, for example, ark:/73195/pa-00055. The five digits after ark:/ are the NAAN, the Name Assigning Authority Number, which says which organisation assigned the name; once registered, a NAAN is never registered again to anyone else. Whatever stands in front of ark: — the address of a resolver such as n2t.net — can change without the identifier changing.

The scheme is open. Any stable memory organisation can obtain a NAAN at no cost, and since 2001 more than 1,700 organisations have assigned ARKs, the Louvre and the Smithsonian among them. Its premise is that persistence is a matter of service, not of syntax: an ARK lasts as long as someone keeps it pointing at the object. Add ?info to a working ARK and it returns a short metadata record instead of the object.

Every object published in the PHOTARCH archive is given an ARK under NAAN 73195. It resolves through the archive's own resolver and through n2t.net, is shown on the object's page beside the inventory number, and is written into the Dublin Core and LIDO exports and into the XMP of every downloaded picture. To cite an object in a publication, use its ARK rather than the page's address.

On this site Standards · How to photograph archaeological artefacts · Manual · What is PHOTARCH? · Manual · What an account gets you · Manual · Object pages — what you see · Manual · Edit › Record — the catalogue record · Manual · Keeping the archive in step (the PHOTARCH app) · Manual · IIIF — deep zoom and manifests · Manual · Dublin Core XML export · Manual · LIDO XML export · Manual · ARK — persistent identifiers · Manual · XMP — embedded metadata in image files · Manual · Citing objects from PHOTARCH

See also Dublin Core · LIDO · IIIF · C2PA

Concept

Bit depth

Bit depth is the number of bits used to record each pixel in one colour channel, and so the number of tones that channel can hold between black and white. Every extra bit doubles them: 8 bits give 256 steps, 16 bits 65,536. Three 8-bit channels are often called a 24-bit image: the same thing, counted another way. FADGI adds that bit depth defines how many shades a file can hold, not the quality of the information in it.

More bits matter when tones are stretched — lifting shadows, correcting exposure or colour — where too few steps show as bands in smooth gradients. They do not add detail or lower noise. Image software reads and writes 8 or 16 bits per channel, so a camera's 12- or 14-bit raw is written into a 16-bit file, which holds all of it without loss.

PHOTARCH grades a picture by the bits the camera recorded, read from the raw file's own header rather than from EXIF, which in a raw file describes the small preview embedded in it. A study asks for at least 10 bits, a reproduction for at least 12, written into a 16-bit TIFF. The depth is never changed: a picture photographed in 8 bits is exported in 8, anything deeper in 16, with the recorded number in the file. For flat originals, FADGI and Metamorfoze judge the file itself: FADGI asks for 16 bits for paintings at three and four stars, Metamorfoze for 16 at Full.

On this site Guides · Standards · Technical requirements · How to photograph archaeological artefacts · Bit depth for archival photography · Choosing a camera for artefact photography · A white background for object photography · File formats for archival photography · FADGI · ISO 19264-1 · Metamorfoze · Manual · Export — what the levels mean

See also Raw · TIFF · Clipping · FADGI · Metamorfoze

Standard

C2PA

Coalition for Content Provenance and Authenticity

C2PA is an open technical standard for recording where a digital file came from and what has been done to it, published by the Coalition for Content Provenance and Authenticity, whose steering committee includes Adobe, the BBC, Google, Microsoft and Sony. The record is a C2PA manifest, in plain language a Content Credential: statements such as when the picture was created and which edits were made, gathered in a claim and signed with the signer's private key. A hash of the file's contents is part of what is signed, so a validator that finds the file changed rejects the manifest. A time-stamp from an independent authority shows when the signature was made, so it can still be checked after the signing certificate has expired.

A Content Credential vouches for provenance, not for truth. The specification judges only whether the statements belong to the file, are correctly formed and have not been tampered with.

In PHOTARCH every published picture, and every file downloaded from the archive, carries a manifest saying how the picture was made — photographed, colour adjusted, cropped, published — with the scale in millimetres per pixel inside it, where it cannot be altered, only broken. The master is deliberately not sealed: its SHA-256 is its proof. Until PHOTARCH has been through C2PA's Conformance Program its certificate is its own, so a validator reports a valid signature by an unknown signer. The seal is not a claim of ownership and carries no licence.

On this site What is PHOTARCH? · Standards · Metadata · How you can trust the picture · Technical requirements · Copyright & Open Access · How to photograph archaeological artefacts · Manual · What PHOTARCH stores for each object · Manual · Image Scale Metadata (ISM) — the scale inside the file

See also SHA-256 · ISM · XMP · Master

Method

Camera lucida

The camera lucida is a small prism on a stand, described by William Hyde Wollaston in 1807. The draughtsman looks down past the edge of the prism, so that part of the pupil sees the subject and the rest sees the paper and the pencil: the subject seems to lie on the paper and can be traced. Unlike the camera obscura it is not a dark box and casts no image; the picture exists only in the eye at the prism.

Wollaston set out how it gives size. A drawing placed as far in front of the instrument as the paper lies beneath it is copied at the same size; one placed proportionally further away is reduced in that proportion, so that the instrument can do the work of a pantograph. That holds for a flat original at one known distance. A subject with depth is drawn in perspective from one point, nearer parts larger, as in a photograph.

Archaeologists used it before photography and beside it: Robert Hay's draughtsmen in Egypt in the 1820s and 1830s; Frederick Catherwood at Copán in 1839, who at first failed with it and then drew every outline with it, dividing his paper into sections to keep the proportions; Augustus Pitt-Rivers, who drew skulls with it at half size and warned in 1887 that, because of perspective, the drawings could not always be relied upon for measurement. It has the photograph's plane problem built in: a scale taken from such a drawing holds in one plane only.

On this site Measuring from photographs: a history

See also Telecentric lens · Relief displacement · Plumb view (F)

Concept

CIELAB (L*)

CIE 1976 L*a*b* colour space

CIELAB, properly the CIE 1976 L*a*b* colour space, writes a colour as three numbers, laid out so that equal distances come roughly as equal differences to the eye. It was defined by the CIE, the International Commission on Illumination, and is standardised as ISO/CIE 11664-4. L* is lightness, from 0 for black to 100 for white. a* and b* carry the colour on two opposing axes, roughly green to red and blue to yellow, from which chroma and hue are worked out; a neutral grey has both at zero.

Because CIELAB is calculated from measurements of light rather than from a camera's or a screen's own numbers, it is the common ground of colour measurement. Colour charts come with reference values in it, and since its 2023 edition FADGI states its tolerances in it rather than in RGB values. It is only approximately uniform, which is why colour differences are now measured with refined formulas such as ΔE2000 rather than as plain distances.

In the method, the grey patches of the ColorChecker are judged by how far each lands from neutral and from its reference lightness — the lightness difference, written ΔL* — and colour accuracy by ΔE2000 over all the patches, both computed in CIELAB. The standards for flat originals also use L* to say how deep into the shadows tone is checked: Metamorfoze Full down to L* 5, Light to about L* 20.

On this site Principles & Setup · Colour accuracy and the ColorChecker · FADGI · Metamorfoze

See also ΔE2000 · White balance · Colour space (sRGB, Adobe RGB) · ICC profile

Concept

Clipping

A sensor and a file can record tones only up to a ceiling and down to a floor. Anything brighter than the ceiling is recorded at the ceiling, anything darker at the floor, and the differences between those tones are lost: they are clipped. In an 8-bit picture, clipped highlights read 255 in at least one colour channel and clipped shadows 0. A clipped area holds nothing that can be recovered.

Metamorfoze, written for flat originals, treats highlight clipping as a fault at every one of its levels. It asks that no point in the picture exceed L* 98, which in Adobe RGB is 249 of 255.

On a white ground, clipping shows at once in the histogram. The method exposes for the white: the brightest part of the ground sits just under the point where it clips, around 250 of 255 in an 8-bit view, which the camera's highlight warning or a zebra shows. A ground left two or three stops lower wastes the top of the file, and the correction that lifts it afterwards lifts the noise with it. Polished metal and glaze can be brighter than the ground, so the warning is checked on the object too. A study allows at most 1 % of the object clipped, a reproduction 0.1 %.

On this site Principles & Setup · Guides · Standards · Technical requirements · How to photograph archaeological artefacts · Choosing a lens for artefact photography · Lighting archaeological objects for photography · A white background for object photography · Colour accuracy and the ColorChecker · FADGI · Manual · Capture — the iPhone as the camera · Manual · Edit › Exposure — the reference card · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean

See also Bit depth · Raw · White balance · Metamorfoze

Concept

ColorChecker

A ColorChecker is a colour chart: a card of coloured patches whose reference values are published. Photographed in the same light as an object, it lets the picture's colour be corrected, and lets anyone measure how far the result lies from the reference. The ColorChecker Classic has 24 patches: six greys, the primaries, and colours chosen to stand for natural things such as skin, foliage and blue sky. The ColorChecker Digital SG has 140. Both were made by X-Rite; the Classic is now sold by Calibrite.

The reference values belong to the card's generation. X-Rite changed the colourants of both charts for regulatory reasons, and cards made after November 2014 need the newer reference data. The number of patches matters too: FADGI notes that a chart with few patches gives a good check of consistency but too little for building an ICC profile, for which 100 patches or more are better. Metamorfoze measures colour on the Digital SG, and a 2021 study found that much of the colour error measured on that chart comes from its 15 black patches.

In the method the Classic is enough. It is photographed at the start of every sitting, where the object will lie and in the same light, as a frame of its own; a new light is a new sitting. Its greys set the white balance, and the mean and 90th-percentile ΔE2000 over its patches decide whether the colour is fit for a study or a reproduction.

On this site Guides · Standards · Technical requirements · How to photograph archaeological artefacts · Colour accuracy and the ColorChecker · ISO 19264-1 · Metamorfoze · Manual · The PHOTARCH photographic method · Manual · Edit › Exposure — the reference card · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean

See also ΔE2000 · White balance · ICC profile · Metamorfoze

Concept

Colour space (sRGB, Adobe RGB)

A colour space is the defined set of colours that an image's numbers refer to. The same RGB values mean different colours in different spaces, so a file has to say which space it is in, normally by carrying its ICC profile. Spaces differ in their gamut, the range of colours they can hold.

sRGB, standardised as IEC 61966-2-1 in 1999, is the default RGB space of computers and the web, and the one most screens can show. Adobe RGB (1998), defined by Adobe, has a larger gamut that takes in typical printing-press colours. eciRGB v2 and ProPhoto RGB are further wide spaces named by the standards for flat originals. FADGI recommends one well-described standard space for all masters; for paintings it accepts sRGB at one and two stars and asks for Adobe RGB, ProPhoto or eciRGB v2 at three and four. Metamorfoze asks for eciRGB v2 at Full, allows Adobe RGB at Light, and sRGB only at Extra Light.

In the method the master is a TIFF in Adobe RGB or wider, with its profile embedded, and a reproduction asks for a 16-bit TIFF in Adobe RGB or wider. Access copies are 8-bit JPEGs in sRGB, which look the same on almost every screen and in every browser.

On this site Guides · Standards · Technical requirements · How to photograph archaeological artefacts · Colour accuracy and the ColorChecker · File formats for archival photography · FADGI · ISO 19264-1 · Metamorfoze · Manual · Export — what the levels mean

See also ICC profile · CIELAB (L*) · Master · Access copy

Concept

CRI

Colour rendering index

The colour rendering index, CRI, says how faithfully a lamp shows the colours of objects compared with a reference light. The CIE's method, CIE 13.3-1995, works out how eight test colours look under the lamp and under a reference of the same colour temperature — the light of a glowing body below about 5,000 K, daylight above — and averages the differences into the general index, Ra. A perfect match gives 100. Special indices for further colours, such as a strong red, can be added.

It matters because a lamp with gaps in its spectrum can make two colours look alike that are not, and a colour chart cannot repair a poor spectrum afterwards. FADGI regards a CRI above 90 as appropriate for most cultural heritage imaging, and asks for 90 or better in the booth where originals are compared with the screen. The index has known limits: for LEDs in particular, Ra does not always match what people see. In 2017 the CIE published a more accurate fidelity index, Rf, with 99 test colours, but it does not recommend Rf as a replacement for Ra in rating products.

The method asks for one kind of light, with a direction. One LED panel is enough to follow it; a high-CRI LED, or a strobe with a modifier, gives better colour and a cleaner result.

On this site Technical requirements · How to photograph archaeological artefacts · Lighting archaeological objects for photography

See also ISO 3664 · White balance · ColorChecker

Method

Cross-polarisation

Cross-polarisation is a way of photographing without surface reflections. A polarising film over the lamp lets through only light that vibrates in one direction. A second polariser on the lens is turned until it stands at right angles to the first. Light reflected from a shiny surface keeps the lamp's polarisation and is blocked; light scattered diffusely has lost it, and passes. Glare from glaze, varnish and polished metal largely disappears, and colours look more saturated. The filters also cost light, so exposures are longer or apertures wider.

Heritage photographers use it for glazed ceramics, metal and other glossy objects, and in photogrammetry, where specular highlights compromise the model and its texture.

In the PHOTARCH method it is a deliberate choice, not a default. It removes much of what makes metal look like metal, and it changes colour; so it is used for a whole sitting, colour card included, and recorded with the pictures. Without it, the method handles shiny objects with a larger diffuser in front of the key light, which spreads a hard highlight into a broad, softer one, and a white card opposite that gives polished surfaces something to reflect on the shadow side. The highlight warning is then checked on the object itself, not only on the ground.

On this site Lighting archaeological objects for photography

See also Sitting · ColorChecker · White balance · Clipping

Concept

Depth of field

A lens focuses on one distance only. Depth of field is the zone in front of and behind that distance which still looks sharp. Sharpness does not stop at its edges but falls away gradually, so the zone depends on how much blur is accepted as sharp — the circle of confusion. Closing the aperture deepens it roughly in proportion to the f-number: two stops, from f/5.6 to f/11, about doubles it. Close up, it depends mainly on the magnification and hardly on the focal length.

Close to a small object it becomes very thin. The method's lens guide works it out as roughly 2 × f-number × c × (1 + m) ÷ m², with c, the accepted blur, taken as two pixels of 0.006 mm. At f/8 that is about 0.4 mm at life size and 3.8 mm at a quarter life size. The PHOTARCH Venus is 47.7 mm high; at a quarter life size and f/8, a band of about 3.8 mm of her is in focus — most of her outline, and little else.

The method therefore focuses by hand on the object's outline, which is what is measured, and then checks the highest point. It stops down to f/8–f/11 and no further: past that, on most cameras, diffraction softens more than the extra depth gains. An object deeper than that can hold is photographed as a focus stack. A study asks for at least 80 % of the object in focus, a reproduction for 95 %; both figures are provisional.

On this site Choosing a camera for artefact photography · Choosing a lens for artefact photography

See also Diffraction · Focus stacking · Magnification

Concept

Diffraction

Light bends a little as it passes the edge of an opening. Because of it, a lens never images a point as a point but as a small disc of light, the Airy disk, about 2.44 × the wavelength × the f-number across. The smaller the aperture, the larger the disc, and the softer everything in the picture becomes, including what is in focus. Lenses therefore tend to have an aperture where they are sharpest: wider than that, the lens's own aberrations dominate; smaller, diffraction does.

Close up, the lens works at a higher f-number than the one set. The method's lens guide takes this effective aperture as the f-number × (1 + m), so that at life size f/8 behaves as f/16; how far a real lens departs from that depends on its construction. At f/16 the disc for green light, with a wavelength of 0.00055 mm, is about 2.44 × 0.00055 × 16 ≈ 0.021 mm across: more than three pixels on a sensor with 0.006 mm pixels. Pixels are then not the same as detail, since diffraction blurs what they could record.

Diffraction is why stopping down cannot be pushed far to gain depth of field. On most cameras f/8 to f/11 is where the gain in depth starts to cost sharpness, and the method goes no smaller. An object deeper than f/11 can hold sharp is photographed as a focus stack instead.

On this site How to photograph archaeological artefacts · Choosing a camera for artefact photography · Choosing a lens for artefact photography

See also Depth of field · Focus stacking · Magnification · Pixels per millimetre

Format

DNG

Digital Negative

DNG, the Digital Negative, is a raw format that Adobe introduced in 2004 and whose specification it makes available to the public. It is a kind of TIFF: the sensor data sits in a TIFF structure with the camera's metadata in TIFF, EXIF, IPTC and XMP tags, and the maker's own private data can travel with it. A DNG can even carry a copy of the original raw file. Some cameras write DNG directly, and Apple ProRAW, from an iPhone, is a DNG.

Because the format is documented and widely read, converting a maker's proprietary raw to DNG is a common archival step. The Library of Congress accepts DNG for digital photographs, and its staff have discussed it as a middle way between a camera's own raw and an uncompressed TIFF.

In the PHOTARCH method a DNG plays the part of any raw. It is the camera original, kept as written, and the preservation master is developed from it as a TIFF. Whether the maker's original raw is kept beside a converted DNG is a decision for each collection; keeping it costs storage and loses nothing. An iPhone's ProRAW records 10 bits: enough for a PHOTARCH study, not for a reproduction, which asks for 12.

On this site Guides · Standards · Bit depth for archival photography · Choosing a camera for artefact photography · File formats for archival photography · Manual · Export — what the levels mean

See also Raw · TIFF · Bit depth · XMP

Standard

Dublin Core

Dublin Core is a small, general vocabulary for describing a resource. Its core is fifteen elements: title, creator, subject, description, publisher, contributor, date, type, format, identifier, source, language, relation, coverage and rights. It is maintained by the Dublin Core Metadata Initiative (DCMI), and the fifteen elements are also an ISO standard, ISO 15836-1:2017. DCMI Metadata Terms adds several dozen further properties and classes.

Its strength is that nearly every catalogue and repository understands it; its limit is that it is flat, a list of fields without the events and relationships a museum record holds. OAI-PMH, the protocol aggregators use to harvest records, requires every repository to offer unqualified Dublin Core, under the prefix oai_dc.

In the PHOTARCH method Dublin Core is the base layer of the record. It is written into every exported picture's XMP — title, creator, rights, source — so the file can explain itself in any program. Every object can also be exported as Dublin Core XML, with its title, type, material, period, institution, find site, photographer, capture date, licence and ARK identifier mapped to the elements, and the archive's OAI-PMH interface serves every published object in oai_dc. Where a richer record built around events is needed, the same object is exported as LIDO.

On this site What is PHOTARCH? · Standards · Metadata · Manual · Bringing photographs in · Manual · Export — and what an image is fit for · Manual · OAI-PMH — harvesting the archive · Manual · Linked Art — the record as CIDOC-CRM · Manual · Dublin Core XML export · Manual · LIDO XML export · Manual · XMP — embedded metadata in image files

See also LIDO · XMP · ARK · IIIF

Concept

Error margin

Every measurement may be a little off, and the error margin says by how much. Two numbers are behind it, and they are easy to confuse.

The standard uncertainty is the basic one: a standard deviation, worked out from everything that can make the measurement wrong. On its own it covers only about two cases in three. The margin that is stated — the expanded uncertainty — is the standard uncertainty multiplied by a coverage factor, usually 2, so that it holds about 95 times in 100. A 25 mm bead photographed from 500 mm on the board, corrected to its middle height, has a standard uncertainty of about 1.4 %; its error margin at 95 % is about ± 2.9 %.

Two sizes can be told apart when they differ by more than both margins together. A margin also says what a picture is fit for: a scale that is uncertain by more than 1 % cannot carry a reproduction.

In PHOTARCH the ± beside a size is the expanded uncertainty, at about 95 %. The file records the relative standard uncertainty, as Image Scale Metadata asks, so that any reader can expand it as the GUM says.

On this site How to photograph archaeological artefacts · Scale in artefact photography

See also GUM · Relief displacement · Pixels per millimetre · ISM

Standard

EXIF

Exchangeable image file format

EXIF — Exif in the standard's own spelling, short for Exchangeable image file format — is the set of tags a camera writes into a picture about how it was taken: the date and time, the camera model, exposure time, focal length, a GPS position where there is one, and much more. It is a standard published jointly by CIPA, the Camera & Imaging Products Association, and JEITA, and it defines how the tags are stored in JPEG and TIFF files. The current edition is Exif 3.1, from 2026; CIPA also publishes a companion standard, Exif metadata for XMP.

In the PHOTARCH method EXIF is the camera's own record, and it is read but never rewritten: what the camera recorded stays as the camera wrote it. The descriptive record — title, caption, rights, photographer — is written separately, as XMP, into every copy that is downloaded or exported.

Some EXIF figures cannot be used for measurement. The resolution fields, XResolution and YResolution, are read by most software as print settings, not as the size of a pixel on the object, which is why the scale travels as Image Scale Metadata. The focus distance has no defined reference point or accuracy: good enough for a rough distance, not for setting millimetres. In a raw file the ordinary tags tend to describe the small embedded preview, so the bits the camera recorded are read from the raw image's own header instead.

On this site Standards · Metadata · Technical requirements · Scale in artefact photography · Bit depth for archival photography · Measuring objects from photographs · Manual · Export — what the levels mean · Manual · Image Scale Metadata (ISM) — the scale inside the file

See also XMP · Raw · Bit depth · ISM

Standard

FADGI

Federal Agencies Digital Guidelines Initiative

FADGI, the Federal Agencies Digital Guidelines Initiative, is a collaboration of United States federal agencies, hosted by the Library of Congress. Its Technical Guidelines for Digitizing Cultural Heritage Materials, free to read and now in a third edition of May 2023, describe what a good digital copy of a book, document, map, photograph or painting is, and how to prove it by measurement.

The guidelines grade on four stars. Four, three and two stars correspond to Levels A, B and C of ISO 19264-1; one star is offered for reference only. Each level sets figures per kind of material — pixels per inch of the original, bit depth, colour space, and how far colour may deviate as a mean ΔE2000 — measured with test targets and analysis software. A star is reached through image analysis, not declared, and without a managed quality-assurance programme no claim of FADGI conformance can be made.

None of its fifteen kinds of material is a three-dimensional object. The nearest, paintings and other two-dimensional art (§3.13), asks for 300 ppi at one and two stars, 450 at three and 600 at four, and a mean ΔE2000 below 6.5, 5, 3.5 and 2. PHOTARCH assesses objects by its own levels instead — a study asks for 12 px/mm at the object, about 300 ppi, and a reproduction for 24, about 600 — and applies FADGI only to flat originals such as paintings, documents and prints.

On this site What is PHOTARCH? · Background · Measuring from photographs: a history · Principles & Setup · Standards · Technical requirements · How to photograph archaeological artefacts · Scale in artefact photography · Bit depth for archival photography · Choosing a camera for artefact photography · Choosing a lens for artefact photography · Lighting archaeological objects for photography · Colour accuracy and the ColorChecker · File formats for archival photography · FADGI · ISO 19264-1 · Metamorfoze · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean · Manual · An object on the phone

See also ISO 19264-1 · Metamorfoze · ΔE2000 · ColorChecker

Method

Focus stacking

A focus stack is a series of photographs of the same object from the same place, each focused a little deeper than the last, blended in software into one picture in which the whole object is sharp. It is the way past the limit of depth of field: close up, a small, deep object needs more depth than any aperture gives before diffraction softens the detail. Natural-history collections use it for small specimens, with programs such as Helicon Focus and Zerene Stacker.

Each step must be smaller than the depth one frame holds, so that the sharp slices overlap. A step too large leaves a band that no frame holds sharp, and stacking cannot make it sharp afterwards. Only the focus moves; camera, object and lamp stay put. In the method's example the Venus, 47.7 mm deep, is photographed at a quarter life size and f/8 in fifteen frames, each holding 3.8 mm and each 3.1 mm deeper than the last.

The standards differ on whether a stack can be a master. Metamorfoze keeps stacking out of the preservation master and allows it in the user files made from it. FADGI does not mention it, but treats a stitched composite as a production master and its captures as the archival master. PHOTARCH accepts the stack — a TIFF, or a PSD with the frames as layers — as the finished picture, judged like any other, with the raw frames kept beside it and the record saying it is a stack.

On this site Measuring from photographs: a history · Standards · Technical requirements · Bit depth for archival photography · Choosing a lens for artefact photography · File formats for archival photography · Metamorfoze · Manual · Export — what the levels mean

See also Depth of field · Diffraction · Master · Metamorfoze · FADGI

Standard

GUM

Guide to the expression of uncertainty in measurement

The GUM is the international rulebook for saying how far a measurement can be trusted. It is published by the Joint Committee for Guides in Metrology — the BIPM, ISO, IEC and five other international bodies — and is free to download. Its current edition is JCGM 100:2008, the 1995 guide with minor corrections.

Its central idea is that a result is not complete without its uncertainty: a number for how widely the true value may lie around the one measured. The GUM expresses it as a standard uncertainty, a standard deviation, and combines the uncertainties of several sources by adding their squares and taking the square root. A known, systematic error is not hidden inside the margin: the result is corrected for it, and only what is left uncertain about the correction is added (§3.2.4). When a quantity is known only to lie somewhere between two bounds, with no reason to prefer any point between them, the GUM takes its middle as the value and the width divided by √12 as its standard uncertainty (§4.3.7). To state a result, the standard uncertainty is usually doubled — a coverage factor k = 2 — which gives an interval that holds about 95 % of the time (§6.3.3).

PHOTARCH follows it for measurements on an object with height. The outline lies somewhere between the surface the object rests on and its full depth H, so a scale taken at the surface is corrected to H ÷ 2, and the spread, H ÷ (√12 × D) for a camera D away, is added to the scale's own uncertainty. The result is shown as ± at about 95 %.

On this site Measuring from photographs: a history · Scale in artefact photography

See also Error margin · Relief displacement · ISM

Format

ICC profile

International Color Consortium profile

An ICC profile is a small file, in a format defined by the International Color Consortium, that says what an image's numbers mean as colour. It relates the colour values of a device — a camera, a screen, a printer — or of a colour space such as Adobe RGB to a device-independent reference based on CIE colorimetry, the profile connection space, so that colour-management software can carry colours from one to another. The format is published as ICC.1:2022 and as the international standard ISO 15076-1. A profile can be embedded in an image file, so that its colours travel with it to any computer.

Two uses are easily confused. An input profile describes a camera: how its raw numbers, in this light, become colours. It can be built from a colour chart photographed in the session, and FADGI recommends exactly that — a profile for each session, applied to its pictures, which are then converted to one standard colour space for the master. The output colour space is what the finished file is written in, and its profile goes with the file.

In the method the ColorChecker photographed in each sitting gives that sitting's colour profile. The master is a TIFF in Adobe RGB with its profile embedded, and the access copies are JPEGs in sRGB, embedded likewise, so that every program reads the numbers the same way.

On this site What is PHOTARCH? · Principles & Setup · Standards · Technical requirements · Colour accuracy and the ColorChecker · File formats for archival photography · Manual · The PHOTARCH photographic method · Manual · What an account gets you · Manual · Object pages — what you see · Manual · Edit › Exposure — the reference card · Manual · Export — what the levels mean · Manual · Bulk dataset export (CSV) · Manual · ICC colour profiles

See also Colour space (sRGB, Adobe RGB) · ColorChecker · CIELAB (L*) · TIFF

Standard

IIIF

International Image Interoperability Framework

IIIF, the International Image Interoperability Framework — pronounced "triple-eye-eff" — is a set of open standards for delivering images and their descriptions on the web, used by libraries, archives and museums around the world. Two of them matter most here. The Image API gives every region of an image, at any size, rotation, quality and format, an address of its own, so a viewer fetches only the tiles it needs to zoom into a very large picture. The Presentation API describes an object as a manifest: its images, their order and size, and its title, metadata and licence. Any IIIF viewer, such as Mirador or the Universal Viewer, can open any collection's manifests, side by side.

The Image API has three compliance levels. At level 0 a server need only serve the sizes and tiles it declares, so it can be a set of files prepared in advance.

The PHOTARCH archive publishes every object this way: Image API 3 at level 0, with tiles made once when a picture is published, and a Presentation API 3 manifest at the master's full resolution, open without an account. The scale goes with it through IIIF's Physical Dimensions service, a factor that turns the canvas's pixels into a physical size, here in millimetres. IIIF itself warns that such figures are often missing or unreliable, which is why the method records how its scale was measured and how certain it is.

On this site Measuring from photographs: a history · Standards · Manual · Object pages — what you see · Manual · Export — and what an image is fit for · Manual · What PHOTARCH stores for each object · Manual · Linked Art — the record as CIDOC-CRM · Manual · IIIF — deep zoom and manifests · Manual · Image Scale Metadata (ISM) — the scale inside the file

See also ISM · Master · LIDO · Dublin Core

Concept

Inverse square law

Light from a small source spreads out as it travels, over an area that grows with the square of the distance. The light reaching each part of a surface therefore falls with the square of the distance: twice as far gives a quarter of the light, two stops less; three times as far a ninth, about 3.2 stops less. The law holds closely once the lamp is a few times its own size away. A large softbox close to an object falls off more slowly than this near it.

At the copy stand it has two consequences. The first is exposure: a lamp moved from 50 to 100 cm needs two stops more, which is why its distance belongs in the record of a sitting and stays the same for a whole series. The second is the ground. With the key light at the upper left, the near side of the ground is closer to the lamp than the far side, and brighter. Across 30 cm of ground the far side receives, from distance alone, about 1.1 stops less light with the lamp at 50 cm, 0.6 at 100 cm and 0.3 at 200 cm: doubling the distance roughly halves the fall-off.

The method does not try to remove that fall-off: it belongs to a light with a direction, and the ground is taken to white in the published copies. Moving the lamp back evens the ground, but costs light and gives a harder shadow edge.

On this site Lighting archaeological objects for photography

See also Sitting · Clipping · White balance

Format

ISM

Image Scale Metadata

ISM, Image Scale Metadata, is an open format for carrying a picture's true scale inside the image file. A handful of XMP fields say how many millimetres one pixel covers on the object, on which pixel grid that holds, how the scale was obtained — a ruler, calipers, a calibration target, the optics, or other — and how certain it is, as a relative standard uncertainty. It is a draft, version 0.1 of September 2026, published under CC BY 4.0, with reference libraries in JavaScript and Python under MIT, test files, a validator and a viewer. The idea comes from medical imaging, where a DICOM file states how large a pixel is and on which grid.

The grid keeps the number honest. A reader that meets a copy resized in proportion rescales the scale; a copy cropped to another shape is recognised, and no scale is shown. The scale holds in one plane only, which the file can name. Three conformance levels build on each other: A gives the scale, B what it is worth, C where the object lies on the grid.

ISM was developed for the PHOTARCH method's measurements and is published independently of it. The method writes it into every exported picture of a measured object, sets the TIFF resolution fields to match, and carries the same figure in the archive's IIIF manifests. A reproduction asks for a scale from calipers or the board, not a ruler, with an uncertainty of 1 % or less.

On this site What is PHOTARCH? · Background · Measuring from photographs: a history · Principles & Setup · Standards · Metadata · How you can trust the picture · Copyright & Open Access · How to photograph archaeological artefacts · Scale in artefact photography · Measuring objects from photographs · Manual · The PHOTARCH photographic method · Manual · How measurement works · Manual · Image Scale Metadata (ISM) — the scale inside the file

See also XMP · Pixels per millimetre · Error margin · IIIF · C2PA

Standard

ISO 19264-1

ISO 19264-1 is the international standard for analysing the image quality of the cameras and scanners used to digitise cultural heritage. It works from a single image of a specified test target, and sets out which characteristics are measured — among them tone, white balance, colour, sharpness, noise and evenness of light — how they are measured, and how the results are reported. Part 1 covers reflective originals such as paper, prints and paintings; transparent originals are left to a future part. The current edition was published in June 2021 by ISO's committee for photography, TC 42, replacing a technical specification of 2017, and a revision is under way.

Its tolerances are grouped into three levels, A, B and C, A being the strictest, and are given in an informative annex. FADGI's four, three and two stars correspond to them, and Metamorfoze's highest level, Full, is largely aligned with Level A, which asks for a mean colour difference of 4 or less and a maximum of 10.

The standard qualifies the set-up, not a particular photograph: it says whether camera, lens and light can reproduce faithfully, not how an original should be placed or lit. PHOTARCH does not assess objects by it. For flat originals the method derives an ISO level from the FADGI grade, through FADGI's own correspondence; no test target is analysed under the standard itself.

On this site What is PHOTARCH? · Measuring from photographs: a history · Standards · Technical requirements · How to photograph archaeological artefacts · Bit depth for archival photography · Choosing a lens for artefact photography · FADGI · ISO 19264-1 · Metamorfoze · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean · Manual · An object on the phone

See also FADGI · Metamorfoze · ΔE2000 · ColorChecker

Standard

ISO 3664

ISO 3664 is the international standard for viewing conditions in graphic technology and photography: the light and surroundings under which prints and transparencies are judged and compared. It is made by ISO's committee for photography, TC 42. The 2009 edition was withdrawn when the fourth edition, ISO 3664:2025, appeared in July 2025; the new edition's scope no longer includes images on colour monitors.

Its reference light is CIE illuminant D50, a phase of daylight of about 5,000 K; the 2009 edition also asked for a colour rendering index of 90 or more, with at least 80 for each of the first eight test colours, and for a print's surround to appear neutral. The surround is the area next to the image, in the plane of viewing, that affects how the eye adapts; the standard keeps it distinct from any border that belongs to the image, such as unprinted white paper, and from the side walls of a viewing cabinet. FADGI refers to ISO 3664 for the viewing booth beside the monitor, at 5,000 K with a CRI of 90 or better.

For photographing objects, the method takes from it the honest case for a grey ground — a white surround makes an object look slightly darker and less saturated — and also the limit of that case: the standard governs the room and the frame around a picture, not the background inside it. PHOTARCH photographs on the same white in every picture, so the effect is the same in all of them and cancels in comparison.

On this site Principles & Setup · A white background for object photography

See also CRI · White balance · CIELAB (L*) · FADGI

Format

JPEG

Joint Photographic Experts Group

JPEG is the everyday format for photographs on screens and the web, named after the Joint Photographic Experts Group, which wrote the standard, ISO/IEC 10918, in 1992. In the form cameras and browsers use, it holds 8 bits per channel and uses lossy compression: the picture is cut into blocks of 8 × 8 pixels and detail the eye is unlikely to miss is discarded, typically making the file ten to twenty times smaller. More is lost each time a JPEG is edited and saved again.

That makes it right for publishing and wrong for keeping. FADGI does not list JPEG as a master format at any level. The Library of Congress prefers uncompressed masters and accepts JPEG where an image exists only in lossy form. Metamorfoze assumes an uncompressed TIFF master and allows JPEG only with the client's approval; in sRGB, a master reaches no higher than its lowest level, Extra Light.

In the PHOTARCH method the JPEG is the access copy: 8 bits, sRGB embedded, made from the master and made again rather than edited. In PHOTARCH's own levels a JPEG master is a record shot at most, since a study asks for a raw of at least 10 recorded bits and a lossless master.

On this site Guides · Metadata · How you can trust the picture · Technical requirements · How to photograph archaeological artefacts · Bit depth for archival photography · Choosing a camera for artefact photography · File formats for archival photography · Manual · What an account gets you · Manual · Object pages — what you see · Manual · Bringing photographs in · Manual · Capture — the iPhone as the camera · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean · Manual · Camera — photographing at the stand · Manual · Account settings · Manual · What PHOTARCH stores for each object · Manual · Open access and copyright

See also Access copy · Lossless and lossy compression · Colour space (sRGB, Adobe RGB) · TIFF

Standard

LIDO

Lightweight Information Describing Objects

LIDO, Lightweight Information Describing Objects, is an XML schema for delivering information about museum objects to online collections, portals and aggregators. Version 1.0 was released in November 2010 at the ICOM CIDOC conference in Shanghai. It grew out of CDWA Lite and museumdat, and is an application of the CIDOC Conceptual Reference Model. It is maintained by the LIDO Working Group of ICOM's International Committee for Documentation (CIDOC). Version 1.1 followed in December 2021 and is backward compatible: any valid LIDO 1.0 record is still valid.

Where Dublin Core is a flat list, LIDO organises much of a record around events, each with its type, actors, date, place, method and materials. A record can then say who made an object, when and where, and what happened to it afterwards, as well as who holds it now. One LIDO record can be supplied to many portals without a separate mapping for each.

In the PHOTARCH method every object can be exported as a LIDO 1.0 XML file covering classification, title, repository and inventory number, find site, the production event with period, material and technique, rights, and the ARK identifier as its record ID. The archive's OAI-PMH interface serves every published object in LIDO as well. It suits national heritage aggregators and collection management systems; Europeana itself takes EDM, delivered through an aggregator.

On this site Standards · Manual · Export — and what an image is fit for · Manual · OAI-PMH — harvesting the archive · Manual · Linked Art — the record as CIDOC-CRM · Manual · Dublin Core XML export · Manual · LIDO XML export

See also Dublin Core · IIIF · ARK

Concept

Lossless and lossy compression

Compression makes an image file smaller by storing its data more efficiently, and it comes in two kinds. Lossless compression gives back, when the file is opened, exactly the data that went in, bit for bit. Lossy compression discards information to make the file smaller still, and what comes back is not identical to what went in.

In image files the usual lossless methods are LZW and ZIP inside a TIFF, lossless JPEG 2000, and the lossless form of JPEG that DNG raws can use. The ordinary JPEG is lossy: it codes the picture in blocks of 8 × 8 pixels, discards detail the eye is unlikely to miss, and typically makes the file 10 to 20 times smaller. It loses a little more each time it is edited and saved again. FADGI approves lossless compression for all uses; for some kinds of material it also lets federal agencies keep masters as JPEG 2000 compressed up to 20:1, tested by eye to be "visually lossless" — which is not the same as lossless.

The PHOTARCH method keeps the two apart. The master is uncompressed or losslessly compressed; the access copy may be lossy. A study asks for a lossless master, so a JPEG master is a record shot at most. In the camera, choose an uncompressed or losslessly compressed raw where one is offered.

On this site Standards · How to photograph archaeological artefacts · Bit depth for archival photography · File formats for archival photography · Metamorfoze · Manual · Export — what the levels mean

See also TIFF · JPEG · DNG · Master · Access copy

Concept

Magnification

Magnification, also given as a reproduction ratio, compares the size of an object's image on the sensor with the size of the object itself. At 1:1, or 1.0×, life size, 10 mm of the object covers 10 mm of the sensor; at 1:2, half life size, it covers 5 mm. A lens's specification gives its maximum magnification, reached at its closest focus. Macro lenses for 35 mm cameras reach life size, some further; most other lenses reach only a fraction of it.

Close up, magnification decides much of what a picture can do. It sets the resolution at the object: at magnification m a lens puts m millimetres of sensor on each millimetre of the object, so the pixels per millimetre are m divided by the pixel pitch — about 167 px/mm at life size with 0.006 mm pixels. It sets the working distance: in a simple lens model the object lies at the focal length × (1 + 1/m) from the lens, 200 mm for a 100 mm lens at life size, so a longer lens stands further back at the same magnification. It governs depth of field, more than the focal length does. It also raises the effective f-number, to about the f-number × (1 + m).

For small objects — coins, beads, flint — the limit on how many pixels land on the object is therefore often not the camera's pixel count but how close the lens can get.

On this site Measuring from photographs: a history · Principles & Setup · Choosing a camera for artefact photography · Choosing a lens for artefact photography · Measuring objects from photographs

See also Pixels per millimetre · Depth of field · Diffraction · Telecentric lens

Concept

Master

A master is the file kept for the long term, from which every other copy is made. FADGI defines the archival master as the best copy a digitising organisation produces, kept without loss of its essential features and used as the source for production masters and derivatives; it also accepts primary in place of master. Best practice, FADGI says, is a lossless master, and for paintings and other flat art the master formats it lists are TIFF and JPEG 2000.

Standards draw the line in different places. For Metamorfoze the preservation master is the unprocessed, first-generation camera or scanner file with its colour space embedded; stitching and focus stacking belong in the user files made from it. FADGI calls a picture assembled from several captures a production master, and keeps the captures themselves as the archival master.

For a three-dimensional object, the PHOTARCH method's master is the finished picture: developed from the raw, or a focus stack made from raw frames that are kept beside it. It is a TIFF in Adobe RGB with its profile embedded, uncompressed or losslessly compressed, at the depth the picture was photographed in — 8 bits stay 8, anything deeper is written as 16. Once written it is never rewritten; whitening the ground and every other change go on copies. Its SHA-256 is recorded when it enters the archive, and it is deliberately left unsealed by C2PA, so that it opens in any software.

On this site Measuring from photographs: a history · Principles & Setup · Guides · Standards · Metadata · How you can trust the picture · Technical requirements · How to photograph archaeological artefacts · Bit depth for archival photography · Choosing a lens for artefact photography · A white background for object photography · Colour accuracy and the ColorChecker · File formats for archival photography · FADGI · Metamorfoze · Manual · What an account gets you · Manual · Object pages — what you see · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean · Manual · The organisation’s archive on a reading computer · Manual · What PHOTARCH stores for each object · Manual · IIIF — deep zoom and manifests · Manual · SHA-256 checksums · Manual · Open access and copyright

See also Access copy · Raw · TIFF · Lossless and lossy compression · Focus stacking · SHA-256

Standard

Metamorfoze

Metamorfoze is the Dutch national programme for preserving paper heritage, launched in 1997 by the Ministry of Education, Culture and Science and hosted by the KB, the National Library of the Netherlands. It funds the preservation and digitisation of vulnerable paper heritage, and every project it funds must follow its Preservation Imaging Guidelines by Hans van Dormolen, which are free to read; version 2.0 appeared in April 2025.

Their main aim is that everything visible in the original is also visible in the preservation master, because the originals are withdrawn from use once digitised. There are three levels: Full, for originals regarded as works of art; Light, for books, newspapers and manuscripts; and Extra Light, for text documents on sheet-fed scanners. Full asks for eciRGB v2 and a mean colour difference of 3 or less with no patch above 7, measured on a ColorChecker Digital SG; Light allows a mean of 4 and a maximum of 14. Measurement follows ISO 19264-1 unless the guidelines say otherwise.

The guidelines are written for paper. PHOTARCH assesses objects by its own levels and applies Metamorfoze only to flat originals, where its colour figures are indicative, because the method reads a ColorChecker Classic rather than the Digital SG. The idea of a master that is colour-managed and never altered carries over. Where Metamorfoze keeps stitching and focus stacking out of the master, though, PHOTARCH accepts a focus stack as the master of an object that no single frame can hold sharp.

On this site What is PHOTARCH? · Background · Measuring from photographs: a history · Principles & Setup · Standards · Technical requirements · How to photograph archaeological artefacts · Bit depth for archival photography · Choosing a lens for artefact photography · Lighting archaeological objects for photography · Colour accuracy and the ColorChecker · File formats for archival photography · FADGI · ISO 19264-1 · Metamorfoze · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean · Manual · An object on the phone

See also FADGI · ISO 19264-1 · ColorChecker · Master

Concept

Pixels per millimetre

Pixels per millimetre (px/mm) is the resolution of a picture measured on the object itself: how many pixels fall across one millimetre of it. It is the object's length in pixels divided by its length in millimetres, so it can only be worked out when the picture has a scale. Imaging standards usually give the same quantity in pixels per inch of the original, ppi; FADGI notes that pixels per millimetre are also used. One px/mm is 25.4 ppi.

It is not the camera's megapixels, nor the ppi written in a file's header. A 24-megapixel camera puts 96 px/mm on a 50 mm object spread across 4,800 pixels of the frame, but only 9 px/mm on a 600 mm object across 5,400. Its inverse, millimetres per pixel, is the scale a file records in Image Scale Metadata. More pixels are not always more detail: at high magnification diffraction blurs what the pixels could record, and FADGI notes that resolution does not by itself guarantee the quality of the information.

PHOTARCH's levels ask for at least 4 px/mm (about 100 ppi) for a record shot when the picture has a scale, 12 px/mm (about 300 ppi) for a study — enough to print the object at life size at 300 dpi — and 24 px/mm (about 600 ppi) for a reproduction, which prints at twice life size. Without a scale a picture can be a record shot and nothing more.

On this site What is PHOTARCH? · Principles & Setup · Guides · Standards · Technical requirements · How to photograph archaeological artefacts · Choosing a camera for artefact photography · Choosing a lens for artefact photography · ISO 19264-1 · Metamorfoze · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean

See also Magnification · ISM · Diffraction · FADGI

Method

Plumb view (F)

The plumb view is the photograph taken straight down onto an object as it lies on the board, with the camera on a stand, square to the surface. In the method's names for camera positions it is F, Front: the upper face as the object lies. Back is the face it rests on; Top and Down are the ends at the top and bottom of the plumb view, Left and Right its sides, and ATL, ATR, ADL and ADR the four diagonals.

It is the main picture: the one the archive shows first and the one measured from. F is the one picture every object must have. Other views add to it and never stand in for it, since an object without F has no picture that can be measured and laid beside another. It also orients the rest: the end at the top of the plumb view is Top.

Other fields that measure from photographs place the camera the same way. Forensic laboratory guidelines say the best position is usually directly over the item, with the camera perpendicular to it; aerial mapping calls it a vertical photograph, taken with the optical axis vertical. A plumb view measures the outline — length and width as seen from above — exactly in one plane only, and never shows thickness. That is why the method records the object's depth, measured with a caliper, beside it.

On this site What is PHOTARCH? · Principles & Setup · How to photograph archaeological artefacts · Scale in artefact photography · Lighting archaeological objects for photography · A white background for object photography · Measuring objects from photographs · Manual · The PHOTARCH photographic method · Manual · Object pages — what you see · Manual · Angles

See also Relief displacement · Sitting · Pixels per millimetre · ISM

Format

Raw

A raw file holds what a camera's sensor recorded before it was turned into a picture. Most sensors see colour through a pattern of red, green and blue filters, and the raw keeps those values as they are: demosaicing into full-colour pixels, white balance and the tone curve are all done afterwards, by a raw developer, and can be done again. Most raw formats are the camera maker's own — CR3, NEF, ARW, RAF — and makers publish little about them, so a format risks going unread once nobody supports it.

In the PHOTARCH method the raw is the camera original: photographed in the camera's deepest raw mode and kept exactly as the camera wrote it. It is not the preservation master. FADGI also counts raw files as unsuitable masters, since each would have to be adjusted before it could be used; the master is a picture developed from the raw.

What a raw is worth depends on the bits the camera recorded. Most cameras record 12 or 14, some medium-format cameras 16, and an iPhone's Apple ProRAW 10; some record fewer in continuous or silent shooting. PHOTARCH's levels ask for at least 10 recorded bits for a study and 12 for a reproduction. The figure is read from the raw image's own header, not from the EXIF tags, which in a raw file tend to describe the small preview embedded in it.

On this site Measuring from photographs: a history · Guides · Standards · Technical requirements · How to photograph archaeological artefacts · Bit depth for archival photography · Choosing a camera for artefact photography · Choosing a lens for artefact photography · A white background for object photography · Colour accuracy and the ColorChecker · File formats for archival photography · Metamorfoze · Manual · Bringing photographs in · Manual · Capture — the iPhone as the camera · Manual · Export — what the levels mean · Manual · Camera — photographing at the stand

See also DNG · Bit depth · Master · White balance

Concept

Relief displacement

In a photograph taken straight down, a point that stands above the ground is not drawn where its foot is. It is shifted along a line running out from the centre of the picture — the point straight below the lens — and away from it. Aerial mapping calls this relief displacement. The US Army Corps of Engineers' photogrammetry manual gives its size as d = r × h ÷ H: it grows with the point's distance r from the centre and its height h, and shrinks with the camera's height H. A point in the centre is not displaced at all. Mapping has to account for it in mosaics and orthophotos, and uses it to measure the heights of vertical objects.

On a copy stand the same geometry applies to an object on the board. Its raised edges are pushed outward, on top of the enlargement every raised part already gets for being nearer the lens. In the method's scale guide, a steep-sided object 25 mm high, photographed from 500 mm, measures about 5.3 % too wide in the middle of the frame. Placed 100 mm from the middle, it measures about 16 % too wide along the line to the middle, while across that line it stays at 5.3 %. A 25 mm bead in the same place goes from 2.6 % to 4.7 %.

The method's remedy is to lay the object in the middle of the frame.

On this site Measuring from photographs: a history · Scale in artefact photography

See also Plumb view (F) · Telecentric lens · Error margin · GUM

Standard

SHA-256

Secure Hash Algorithm, 256-bit

SHA-256 is a hash function: a calculation that reads every byte of a file and returns a number 256 bits long, usually written out as 64 hexadecimal characters. It is one of the algorithms of the Secure Hash Standard, FIPS 180-4, published by the US National Institute of Standards and Technology. The same file always gives the same value, and any change to it — one pixel, one line of metadata — will with very high probability give a completely different one. That is why the value is called a checksum, or the file's fingerprint.

Archives use checksums for fixity: the assurance that a file has not changed. A checksum is recorded when a file is taken in, then worked out again after copying, or after years in storage, and compared with the stored value. A match shows that the file is bit for bit the one that was kept. A mismatch shows that something changed, but not what, nor who changed it; a signed record such as C2PA answers those other questions.

In PHOTARCH every master is given a SHA-256 when it enters the archive, shown on the object's record and included in the bulk CSV export. Because the master is never rewritten, its checksum holds for as long as the archive exists. A download is a different file, with the record and the seal written into it, so it has a checksum of its own; what proves a download is its seal.

On this site Standards · Metadata · How you can trust the picture · Technical requirements · File formats for archival photography · Manual · What PHOTARCH stores for each object · Manual · SHA-256 checksums · Manual · Bulk dataset export (CSV)

See also C2PA · Master · Access copy

Method

Sitting

A sitting is one session at the camera in which the light does not change: the same lamp and modifier, at the same place and distance, with the same camera settings. It begins with a frame of the ColorChecker, laid where the objects will lie, in the same light. Every object photographed after it, from every side, is lit the same way — the camera moves between views, never the lamp — so one correction from that card serves every frame of the sitting.

Imaging guidelines call the same thing a session. FADGI recommends imaging a target at the beginning of each session, making an ICC profile for each session and applying it to that session's images, and imaging the target again before resuming when a session is interrupted.

The rule that follows is short: if the light changes, it is a new sitting, and the card is photographed again before the next object. Moving the lamp or changing its modifier starts one. Cross-polarisation, when it is used, is used for a whole sitting, colour card included. The sitting is also what is recorded: the light source, its modifier, its angle and its distance belong with the pictures, since no file can know them unless someone writes them down.

On this site What is PHOTARCH? · Principles & Setup · Guides · Standards · How to photograph archaeological artefacts · Bit depth for archival photography · Choosing a camera for artefact photography · Lighting archaeological objects for photography · A white background for object photography · Colour accuracy and the ColorChecker · ISO 19264-1 · Metamorfoze · Manual · Collections in the app · Manual · Capture — the iPhone as the camera · Manual · Edit › Exposure — the reference card · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean · Manual · The physical archive (Organisation) · Manual · ICC colour profiles

See also ColorChecker · White balance · ICC profile · Plumb view (F) · FADGI

Concept

Telecentric lens

An ordinary lens sees a nearer thing larger: the image of an object grows almost in proportion as it comes closer, so the top of an object is drawn larger than its base. A telecentric lens collects only the light whose central rays run parallel to its axis. Within a range of distances, its magnification then does not change as the object moves nearer or further, and an object with height is drawn at one scale from top to base.

The cost is size. Because the rays it uses are parallel, the front of the lens must be at least as large as the field it sees, which makes telecentric lenses larger, heavier and dearer than ordinary ones, and most cover only a few centimetres. They are used in machine vision and industrial measurement, where parts are gauged from images.

For archaeological objects such a lens would remove the error that height brings to a scale, which comes from perspective. None is known in use for museum objects: the method's scale guide notes that one covering a field of 31 × 21 cm is a metre long and weighs about 39 kg. The method reduces the error by distance instead — a longer lens, further back — which shrinks it but never removes it, and states what is left as an error margin, worked out from the object's depth measured with a caliper.

On this site Measuring from photographs: a history · Scale in artefact photography

See also Relief displacement · Magnification · Error margin · Plumb view (F)

Format

TIFF

Tagged Image File Format

TIFF is an image format developed by Aldus and Microsoft in the 1980s; the specification now belongs to Adobe. Its current version, 6.0, dates from 1992. A TIFF stores the picture together with tags that describe it, and can keep it uncompressed or with lossless compression, LZW or ZIP, which makes the file smaller without changing a single pixel. It can carry an embedded colour profile and metadata.

TIFF has long been the ordinary format for preservation masters. FADGI calls TIFF 6 the go-to master format of still imaging for decades and names it for masters beside JPEG 2000. Metamorfoze assumes an uncompressed TIFF 6.0 master at all three of its levels, and the Library of Congress lists TIFF as a preferred format for digital photographs.

In the PHOTARCH method the preservation master is a TIFF in Adobe RGB, with its colour profile embedded, and it is never rewritten once stored. It is written at the depth the picture was photographed in: 8 bits stay 8, and a deeper capture — 10, 12, 14 or 16 bits — is written as 16, which holds all of it without loss, with the bits the camera recorded stated in the file's metadata. A reproduction asks for a 16-bit TIFF in Adobe RGB or wider.

On this site Measuring from photographs: a history · Guides · Standards · Metadata · How you can trust the picture · Technical requirements · How to photograph archaeological artefacts · Scale in artefact photography · Bit depth for archival photography · Choosing a lens for artefact photography · Colour accuracy and the ColorChecker · File formats for archival photography · Measuring objects from photographs · FADGI · Metamorfoze · Manual · What an account gets you · Manual · Object pages — what you see · Manual · Bringing photographs in · Manual · Export — and what an image is fit for · Manual · Export — what the levels mean · Manual · What PHOTARCH stores for each object · Manual · Image Scale Metadata (ISM) — the scale inside the file · Manual · Open access and copyright

See also Master · Lossless and lossy compression · Bit depth · ICC profile

Concept

White balance

Every light has a colour of its own, and the camera records it. White balance is the correction that makes neutral things neutral again, so that a grey comes out grey. It is set in kelvin, after the colour temperature of the light — a household bulb is about 2,700 K, daylight and flash about 5,500 K — with a second setting, tint, for the lean toward green or magenta of many lamps, LEDs above all. In a raw file the white balance is only a setting stored beside the data, so it can be changed afterwards without loss.

The imaging standards measure it on the grey patches of a chart, as how far they land from neutral. FADGI's metric is ΔE2000 with the lightness left out, written ΔE(a*b*); Metamorfoze uses the same measure and allows at most 3 at its Full and Light levels.

In the method, white balance is set once per sitting, from the grey patches of the ColorChecker photographed where the object will lie, in the same light. One click on a grey patch sets it for every picture of the sitting. That is why the method asks for one kind of light only, with room lights off and windows shaded: two sources of different colour cannot both be corrected by one white balance. A new light makes a new sitting, and a new frame of the card.

On this site Guides · Standards · Lighting archaeological objects for photography · Colour accuracy and the ColorChecker · File formats for archival photography · FADGI · ISO 19264-1 · Metamorfoze · Manual · Camera — photographing at the stand

See also ColorChecker · Sitting · ΔE2000 · CIELAB (L*) · Raw

Standard

XMP

Extensible Metadata Platform

XMP, the Extensible Metadata Platform, is a way of writing information about a file inside the file itself. Adobe created it, and it has also been an ISO standard, ISO 16684-1, since 2012. The information is written as RDF/XML in a block called an XMP packet, and each kind of information has a namespace of its own — Dublin Core for title, creator and rights, for example — so new namespaces can be added without disturbing the others. JPEG, TIFF, PNG and DNG each have a defined place for the packet. Where it cannot be embedded — some raw formats have no room for it — the same packet can be kept beside the file as a sidecar ending in .xmp.

In the PHOTARCH method the XMP packet carries the record that travels with every exported picture: Dublin Core for the base description, IPTC Extension for the place and the holding institution, PHOTARCH's own namespace for inventory number, period, material and the object's dimensions in millimetres, and Image Scale Metadata for the scale. The camera's EXIF is kept beside it, unchanged.

Any program that handles XMP can read the record, and can also change it. That is why the scale in a sealed file is written twice: in the XMP, where anything can read it, and signed inside the file's C2PA manifest, where it cannot be altered, only broken.

On this site What is PHOTARCH? · Standards · Metadata · How you can trust the picture · Scale in artefact photography · Measuring objects from photographs · Manual · What an account gets you · Manual · Object pages — what you see · Manual · Export — and what an image is fit for · Manual · What PHOTARCH stores for each object · Manual · Image Scale Metadata (ISM) — the scale inside the file · Manual · XMP — embedded metadata in image files

See also EXIF · Dublin Core · ISM · C2PA

Method

ΔE2000

CIEDE2000 colour-difference formula

ΔE (Delta E) is a single number for how far one colour lies from another. Both colours are placed in CIELAB and the distance between them is measured. ΔE2000, or CIEDE2000, is the formula the CIE published in 2001, now standardised as ISO/CIE 11664-6. It is built on CIELAB but weights differences of lightness, chroma and hue by how the eye judges them, with further corrections for blues and greys.

In the method's plain reading, 0 means identical, a value around 1 is near the smallest difference a practised eye sees between two patches side by side, and larger numbers are differences anyone can see. A chart photographed in the light gives one ΔE per patch, and the standards report the mean over all patches and the worst ones. White balance is measured with a version of the same formula that leaves lightness out, written ΔE(a*b*). Metamorfoze uses a variant that weighs lightness evenly along the whole grey scale.

PHOTARCH measures ΔE2000 over the patches of the ColorChecker Classic photographed in the sitting. A study asks for a mean of 5 or less; a reproduction for a mean of 3 or less, with the 90th percentile at 6 or less; a record shot asks nothing of the colour. For flat originals, FADGI's category of paintings and two-dimensional art asks for a mean below 6.5, 5, 3.5 and 2, from one star to four.

On this site Guides · Standards · Colour accuracy and the ColorChecker · FADGI · ISO 19264-1 · Metamorfoze · Manual · Export — what the levels mean

See also CIELAB (L*) · ColorChecker · FADGI · Metamorfoze