Documentation

A Photo Archive Workflow, Step by Step: From Import to Restore

Inventory, review duplicates, consolidate, describe, archive, restore: a photo archive workflow for an artist's work, step by step, with the trap hidden in each one.

Organization

An artist’s archive isn’t sorted in one go

An artist’s photo archive rarely arrives in good order. Several external drives, copies of memory cards, exports made for printing, negative scans, and somewhere a “to sort” folder that is ten years old. The same images exist in three copies, some exist only once, and nobody remembers which.

A sound photo archive workflow doesn’t try to fix everything at once. It runs through steps in a set order, because each one prepares the next: you don’t describe files you are about to move, you don’t archive what you haven’t sorted, and you don’t trust a copy you have never restored. Here is that path, step by step, with what can go wrong at each stage.

Before you start: three rules

  • Nothing is deleted while sorting. Decisions come first, files move afterwards, and whatever leaves goes to the Trash, not into the void.
  • Every file gets a fingerprint. A SHA-256 fingerprint is computed from the exact content of a file. Two files with the same fingerprint are identical; a file whose fingerprint changes has been modified or damaged.
  • Every step leaves a written record. A manifest, a report, a log: what will let someone, twenty years from now, know where each file came from.

Step 1: take inventory without touching anything

The first step is to know what you own. Every drive has to be read, every file listed and fingerprinted, without writing or moving anything.

In Obscura Trace, this is the job of the Scan Assistant. You declare your drives or folders as sources, choose file types and folders to exclude, then the analyses: metadata extraction, exact duplicates (same SHA-256 fingerprint) and visual duplicates (very close images, such as a re-saved JPEG or a black-and-white conversion). The analysis only reads your files.

The trap. By default, only photographic formats (RAW, JPEG, TIFF, PSD…) are inventoried. XMP edit files, videos and PDFs are optional. If your RAW files have XMP sidecars, tick that type: otherwise they won’t travel with their images in step 3.

Step 2: decide on duplicates yourself

Software can find duplicates. It can’t know which one to keep for you. Is the 2009 scan worse than the 2015 one, or is it another reading of the same negative?

Obscura Trace’s Review Center (Centre de revue) shows each group of matched files with a recommended master and the reasoning behind it, criterion by criterion: resolution, bit depth, format, metadata, color profile. Confidence is stated: “established” for verified identical content, “estimated” for visual similarity. You then decide, file by file: master, keep, another version, or duplicate. None of these decisions modifies a file.

The trap. Visual duplicate detection recognizes neither cropping nor rotation. Two scans of the same negative, framed differently, will generally not be grouped. Our article on image deduplication covers these limits.

Step 3: consolidate into a single folder

Once decisions are made, what remains has to be gathered into a clean library on a drive you choose.

Obscura Trace’s Transfer Center (Centre de transfert) copies (the default) or moves the contents of a source to a destination. It always goes through a plan you read, optionally a dry run that can never be executed, then a confirmation that sorts each operation by whether it can be undone. Each copy is written under a temporary name, verified by its fingerprint, and only then given its final name. When moving, the source goes to the Trash, which the app never empties. A file marked “duplicate” stays in the source. A RAW file and its XMP leave together or not at all.

At the end, Trace writes a manifest in the destination: a readable JSON file that gives, for each file, its original path, its new path and its fingerprint. It is the memory of the consolidation.

The trap. A file the analysis didn’t inventory isn’t transferred. Before consolidating, check that step 1 covered every type you want to take along.

Step 4: describe the work

The consolidated folder is now stable. This is the right moment to describe it, and not before.

Obscura Catalog imports a folder without copying or moving it: each image is referenced where it is, with its SHA-256 fingerprint. Catalog reads EXIF, IPTC and XMP metadata. By default, it never writes into your files: your edits go into the catalog and into an XMP sidecar placed alongside, which other software reads. Beyond files, it keeps the catalogue raisonné: works distinct from files, each with a number that is never reassigned, attributions, prints and provenance.

The trap. Catalog recognizes a folder by its path. A folder moved or renamed in the Finder, then imported again, becomes a new project, and its files show up as duplicates. That’s why consolidation comes before description.

Step 5: preserve, and prove it

Describing doesn’t protect anything. You still need verifiable copies, on several media, one of them off-site.

Obscura Vault takes files in two ways.

  • A folder. Vault imports it as a project, recording each file’s fingerprint, then packs it into a signed .obf archive, optionally encrypted, which is read back as soon as it is written.
  • A Catalog package. In Catalog, “Prepare a Package for Vault…” writes a folder holding a backup of the library, an export in open formats and the list of originals, identified by fingerprint. Vault verifies the package, finds each original by its fingerprint, archives it, and reports the missing ones without rejecting the package. Catalog and Vault share no database and no settings: the package is their only exchange.

Then Vault checks that everything stays intact: fixity audits, on demand, weekly or monthly, re-read every file and compare its fingerprint with the one taken at import. The storage media inventory turns the 3-2-1 rule into three levels (working, local archive, off-site) and gives each drive a replacement date. Offsite copies go to S3 or compatible storage, a remote folder or a synced folder, and Vault verifies the local archive before any upload. For the reasoning behind three copies, see the 3-2-1 rule.

The trap. Keys live in the Mac’s keychain, not in the library. Without the encryption key, an encrypted archive can no longer be opened by anyone. Export your keys as soon as they are created, and store them away from the archives.

Step 6: restore

The last step is the one people test least, and the one that justifies all the others.

To restore an archive, Vault verifies it first, refuses to extract it if it fails, never overwrites a file already present, then reads back each restored file and compares it with the manifest. The same applies to an offsite copy. A Catalog package is given back byte for byte: its backup folder reopens in Catalog, and the originals, put back in their original location, are found again by the library. Finally, an unencrypted .obf archive remains readable without Vault: its content is an ordinary tar or zip archive, and its inventory a text file.

The trap. A restore you have never tried is a hypothesis. Run one in a test folder before you need it: see why you should test a restore.

Alongside: the studio’s documents

An artist’s archive isn’t only images: letters, contracts, scanned notebooks, press files. These aren’t described like photographs, but like archives.

Memory Atlas arranges them to the ISAD(G) standard, from fonds down to item. Files stay where they are: Atlas keeps a link to each one, without copying or modifying it, and the link follows a file renamed or moved on the same drive. A file that has been deleted, or copied to another drive and then erased, is flagged as missing: Locate… finds it again, and Atlas offers to relocate the other files from the same folder in one go. As with Catalog, it’s best to link files once consolidation is done. The description exports to EAD 2002 or EAD3, or as a PDF or HTML finding aid.

The workflow in one table

StepAppWhat goes inWhat comes out
Inventory, reviewObscura Traceyour drivesdecisions, no file modified
ConsolidateObscura Tracean analyzed sourcea clean folder and its manifest
DescribeObscura Cataloga folderthe catalog, XMP sidecars, a package for Vault
PreserveObscura Vaulta folder or a Catalog packageverified .obf archives, offsite copies
RestoreObscura Vaultan archive or a packagefiles read back and checked against the manifest
Arrange documentsMemory Atlasfiles, left in placean ISAD(G) description, an EAD export

Each app does one thing, and they hand each other files: a folder, a manifest, a package. You can start with just one.

Where to begin

If your photos are spread across several drives today, start with Obscura Trace: the inventory is the basis for everything else, and it touches nothing. The suite’s software page presents each step, and the documentation brings together the guides of all four apps.