Preservation
Why an audit, and not a hunch
Most photographers have a feeling about the state of their archive: “everything is on the NAS”, “there’s a backup drive somewhere”. That is a hunch, not an assessment. A photo archive audit replaces the hunch with dated facts: how many drives, how old, how many copies of each series, how many files were read back and found intact.
An audit fixes nothing. It tells you where the risk is, and in what order to deal with it. That is why it has to be written down: an audit you cannot reread next year is only useful once.
The six risks to measure
Ageing media
Hard drives, SSDs, memory cards and optical discs all have a service life. You cannot know it in advance, but it is there. The first risk in an archive is often a ten-year-old drive in a drawer holding the only copy of a series. For each medium, record its kind, the date it went into service (or “unknown”), and where it is kept.
Missing copies
The usual yardstick is the 3-2-1 rule: three copies, on two different media, one of them offsite. The audit doesn’t ask “do I have a backup?” but “how many separate copies of this series exist, and where?” Three drives sitting on the same desk are not an offsite copy.
Duplicates that blur the inventory
Duplicates are not only wasted space. They make the inventory wrong: you believe you hold three copies of a series when you actually hold three different exports, or you believe a file is unique when four copies of it are scattered around. Only a byte-for-byte comparison of content can say two files are identical. Name, date or size are not enough.
Corrupted or unreadable files
A file can be present and still be lost: a few altered bytes, an interrupted copy, a weakening sector. Nothing tells you until you open it. The article on detecting corrupted files covers the mechanisms. For the audit, keep one rule: a file is only sound once it has been read in full and compared with a reference fingerprint.
Formats that close
A proprietary format readable today may not be tomorrow. The audit lists the formats present (each camera’s RAW, TIFF, JPEG, PSD, XMP edit files) and flags those that depend on a single program. Which formats to keep is discussed in Should you archive only RAW files?.
No proof of integrity
This is the least visible risk. Without fingerprints recorded when the files entered the archive, there is no way to prove they haven’t changed. You can see that a file opens, not that it is still the same. An archive without reference fingerprints is an archive you cannot fully audit.
The method in five steps
- List the media. Connect each drive, one at a time. For each one: a name written on its label, its kind, its date of entry into service, its location. Drives that no longer mount go to the top of the list.
- Inventory the contents. List what each medium holds, without moving or deleting anything during the audit. An audit that starts “tidying up” stops being a record of what was there.
- Count the copies. For each series or project, write down how many identical copies exist, and on which media. This is where exact duplicates become useful: they show you where the copies are.
- Read the files back. Compute or recompute fingerprints and compare them with the reference ones. Keep a separate list of files you could not read: they are neither sound nor corrupted, they need examining.
- Record and rank. Date the findings, rank the risks, decide on actions. The next audit is compared with this one.
The audit checklist
| Risk | Question to ask | Warning sign | Priority |
|---|---|---|---|
| Ageing medium | How old is this medium, and when should it be replaced? | Unknown date, drive that no longer mounts | High |
| Missing copy | How many separate copies, and how many offsite? | A single copy, or all copies in one place | High |
| Duplicates | Are these files identical in content? | Same names, different contents | Medium |
| Corrupted or unreadable file | Has it been read in full and compared with a reference? | File won’t open, fingerprint differs | High |
| Closed format | Can this format be read by more than one program? | Single-program format, no open derivative | Medium |
| No proof of integrity | Is there a dated reference fingerprint? | No fingerprints, or fingerprints never re-checked | High |
Fill in one row per medium and per series, not one row for the whole archive. Risk hides in the exceptions.
Common traps
- Counting a missing file as sound. A disconnected drive proves nothing. It doesn’t prove the files are lost either. It is a third answer, to be kept separate.
- Mistaking presence for a copy. A synced folder that faithfully reproduces a mistake is not an independent copy.
- Fixing things during the audit. Deleting duplicates before counting copies risks erasing the last good one.
- Never auditing again. A finding has a date. Without the next one, you can’t tell what degraded in between.
What the Obscura Flow suite does
Inventorying drives with Obscura Trace
Obscura Trace covers steps 2 and 3 of the method. Its Scan Assistant analyzes one or more drives declared as sources, and the analysis only reads: it writes nothing, moves nothing and deletes nothing. For each source, the list shows the number of files already indexed, “never analyzed” where relevant, or “root unreachable” if the drive is absent.
Trace only treats two files as exact duplicates when their full SHA-256 fingerprints match; name, date or size alone are never enough. Visual duplicates (a re-saved JPEG, a resized version, a black-and-white conversion) are proposed separately, as matches to confirm. The Dashboard totals indexed files, volume, exact duplicates and the images that belong to no group.
Files the analysis could not read are logged with their reason and gathered under the “Corrupted files” filter of the Review Center. Trace neither modifies nor deletes them. Running the analysis again later only rereads new or modified files, and detects files that have disappeared.
Measuring integrity with Obscura Vault
Obscura Vault provides the proof of integrity most archives lack. A fixity audit reads every file in full and compares its fingerprint with the one taken at import. The result comes down to four numbers: Verified, Altered, Unverifiable, Never checked. They always add up to the library’s total: a file that was never checked is never counted as verified.
The Understand the verdicts chapter explains why “unverifiable” is neither “verified” nor “altered”, which is exactly the trap described above. Each file read adds a line to the fixity journal, which only grows: the moment a file stopped conforming remains known. An automatic audit can be scheduled weekly or monthly; a stopped or incomplete audit does not move the deadline.
Keeping track of media and the 3-2-1 rule
The Storage media screen matches step 1. Each medium is recorded with its name, kind, tier, commissioning date and location. Vault translates the 3-2-1 rule into three tiers (T1 working, T2 local archive, T3 offsite) and flags any tier without an active medium: “the 3-2-1 rule is not met”. Three new disks at the same tier make only one copy.
Each medium gets a replacement deadline and a state: In Service, Replace Soon, Overdue or Deadline Unknown. If the commissioning date is unknown, Vault doesn’t make one up. You can also verify an archive held on a specific medium; the verdict is then recorded for that medium. Vault’s Dashboard brings it all together and names each gap, for instance “no offsite copy”.
Where to start
You can run your first audit by hand: the checklist above is enough. To repeat it without effort, start by inventorying your drives with Obscura Trace, then entrust your originals to Obscura Vault, which reads back, dates and records every check. The four applications of the suite are presented on the software page, and their guides are gathered in the documentation.