Data Recovery Case File · Mac & Apple Systems · Originals Among the Derivatives

A Photo Library Holds Several Versions of Every Picture, and Only One Is Yours

Her enquiry asks for something more specific than a recovery. A backup of a photo library held on an external drive, where she would like "the larger photos, not all the copies the application made of individual faces and so on." She has identified a real problem in advance: a library contains thousands of derivative files alongside the originals, and a recovery that returns everything returns mostly the wrong thing.

The decode: what a library actually contains, and how the sets are separated

Why one photograph becomes many files: the application generates renditions for its own use. A thumbnail for grid views, a medium version for previewing, a display-sized copy, and often cropped extracts for faces — each stored as a separate file alongside the original.

Why that produces the ratio she has noticed: the derivatives outnumber the originals several times over. A library of a few thousand photographs can contain tens of thousands of files, most of which nobody would want back.

Why a naive recovery returns the wrong result: a signature search finds every image file equally. It cannot tell an original from a thumbnail, because both are valid images with the same internal format — so the output is dominated by material that is worthless.

Why that matters beyond tidiness: it makes the result unusable. Finding a few thousand originals among tens of thousands of near-duplicates is work she would have to do by hand, which is why asking beforehand was the right move.

The first way the sets are separated: by location within the library. Applications store originals and generated renditions in different internal folders, so where the structure survives, the distinction is simply a matter of which folder a file came from.

The second, where the structure does not survive: by dimensions. Thumbnails and previews are small by definition, and originals carry the camera's full pixel dimensions — a threshold separates them reliably without any folder information.

The third, and the most precise: by capture metadata. An original from a camera carries the exposure settings, lens and capture time embedded in it, while generated renditions typically do not — so presence of that data identifies an original almost definitively.

Why the drive not being in the platform's own format helps here: the backup was written to a drive formatted for general use. That makes the library a straightforward directory tree rather than a platform-specific container, which is simpler to work through.

Why the library's own catalogue is still worth recovering: it records albums and arrangement. Even where she only wants originals, the catalogue tells us which file is which — and can restore names and dates alongside the images.

What should be agreed before the work: the size threshold, and whether near-duplicates should be excluded entirely or delivered separately. Those choices shape the deliverable, and they are cheap to settle in advance.

On the bench

Library structure was mapped to distinguish originals from generated renditions — applications producing thumbnails, previews, display copies and cropped extracts as separate files alongside each original, so derivatives outnumber originals several times over and a signature search cannot distinguish them, all being valid images of the same format. Originals were identified by internal library location, pixel dimensions and capture metadata, the last being present in camera originals and typically absent from generated renditions.

The outcome

Library structure mapped to separate originals from derivatives, the drive imaged before extraction, and derivative renditions excluded from delivery. Free assessment, one fixed written figure including VAT; where a drive has to be opened, 50% of parts and labour is payable upfront with the balance only on success — otherwise no recovery, no fee. The decode: you spotted the real problem. A library holds several versions of every picture, and a search cannot tell a thumbnail from an original — so the separation is made by folder, by dimensions, and by capture metadata.

Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.