How to Back Up and Protect Your Reference Library (2026)

Backing up a reference library means protecting two things: the image files themselves, and the index that holds your tags, ratings, descriptions, and folder structure. Miss either one and you either lose images or lose months of curation work. This guide walks through a practical, tool-agnostic backup strategy for artists and designers, then covers the specific considerations for the four most common reference tools: refern, Eagle, PureRef, and BeeRef.

The two things you can lose

Before getting into steps, it helps to understand exactly what is at risk.

Your image files. These are the actual JPGs, PNGs, and other files you have collected. A hardware failure, accidental deletion, or ransomware attack can wipe them permanently if no copy exists elsewhere.

Your index. This is the metadata your tool builds around those files: folder organization, tags, ratings, descriptions, source URLs, color labels, and smart folders. Even if you recover your image files, a lost index means rebuilding months or years of curation from scratch. For a library of thousands of images, that curation effort can be more valuable than the images themselves.

Different tools store these two things very differently, and that difference determines how hard or easy your backup situation is.

How different tools store your library

Understanding what is on disk tells you what you need to back up.

Files-on-disk model (refern)

refern is a local-first desktop reference manager that indexes existing workspace files in place without moving them. Your images stay exactly where you put them, in a normal folder you already own. refern writes two sidecar items into that folder: a SQLite database (refern-db.sqlite) that holds all your metadata, tags, ratings, and search index, and a refern-thumbnails/ folder for fast grid previews.

This model has a significant backup advantage. To back up everything, you back up the folder. One operation covers your images and your entire index together. There is no proprietary container to worry about, no special export step, and no second location that can get out of sync.

Managed library folder (Eagle)

Eagle ($34.95 one-time as of August 31, 2026) uses standard import to copy media into its own managed .library folder along with its metadata database. The upside is that the managed library is in one place. If you retain the source originals, the source media bytes exist in both locations, although total storage also depends on generated data and later file handling. Eagle has no built-in cross-device cloud sync, so protect the .library folder with a third-party backup tool or external drive. Back up retained source originals separately if you still rely on them.

Proprietary file format (PureRef)

PureRef 2.x scenes use .pur files. Local images are embedded by default, but PureRef can instead link them to existing disk paths. Back up every .pur scene and any linked source folders. A scene backup alone does not protect linked media.

A .pur scene is an important project file and belongs in a versioned backup routine. Also test a restore that includes any linked source paths. Do not rely on a single scene file or fallback copy as your only backup.

PureRef's current handbook does not document a library-wide tag database or cross-board content search. Back up every .pur scene and any linked source folders.

Proprietary file format (BeeRef)

BeeRef (free, open source) works similarly to PureRef: it saves each board as a .bee file with images embedded as PNG or JPG data inside a SQLite container. Larger boards with high-resolution references can result in very large files. Back up your .bee files as you would any important document. BeeRef has no library, no tags, and no search, so there is no separate index to protect.

Before you start: audit what you have

Before setting up a backup strategy, spend five minutes understanding what you actually need to protect.

1. Find where your reference files live. In refern, this is the workspace folder you pointed at on setup. In Eagle, look in Preferences for your library location. For PureRef and BeeRef, this is wherever you save your .pur and .bee files.
2. Estimate the total size. A folder with 10,000 images at typical web resolution is roughly 20 to 40 GB. High-resolution photography or RAW files push this higher. Your backup destination needs to hold at least this much space.
3. Identify the index and generated-data location, if any. A refern workspace uses a SQLite database and generated thumbnails alongside, or in a configured data location associated with, the workspace files. For Eagle, the managed .library folder contains the database and imported media. For PureRef and BeeRef, there is no separate index.
4. Check how long since your last backup. If the answer is "never" or "I don't remember," start now.

Step 1: Choose your backup destinations

A reliable backup strategy uses at least two destinations: one local and one offsite.

Local backup (external drive). An external USB or NAS drive gives you fast recovery after a hardware failure and works without internet access. Use a dedicated drive for backups, not the same drive you work from.

Offsite backup (cloud storage). Cloud backup services such as Backblaze Personal Backup ($99/year, unlimited storage), Google Drive, iCloud, or Dropbox give you protection against fire, theft, or physical damage to your home or studio. For large reference libraries (50 GB or more), Backblaze is significantly cheaper per gigabyte than Dropbox or Google Drive.

The rule of thumb for important data is the 3-2-1 rule: three copies total, on two different media types, with one copy offsite. For a reference library, "three copies" might be: the original on your working drive, a copy on an external drive, and a copy in the cloud.

Step 2: Set up automated backup

Manual backups fail because they rely on remembering. Automate as much as possible.

On Windows, use File History (built into Windows 10 and 11) to continuously back up a folder to an external drive. Set it to run hourly or daily. Alternatively, a dedicated tool like FreeFileSync can schedule folder-to-folder copies.

On macOS, Time Machine backs up your entire drive (or selected folders) to an external drive automatically. For cloud backup, Time Machine can back up to a NAS or server; many macOS users add Backblaze as a second cloud layer.

On Linux, rsync with a cron job is the standard approach. A simple rsync -av --delete /path/to/references /path/to/backup scheduled daily covers most cases.

For the cloud layer on any platform, a service like Backblaze Personal Backup runs continuously in the background and backs up everything below your size/type filters automatically, with no manual action needed after setup.

Step 3: Back up the index alongside the images

This step is where most artists make a mistake. They back up the image files but forget the metadata.

refern: Back up the workspace media root and its associated data location. With the default layout, the SQLite database and generated data may be alongside the workspace files; with a custom data folder, that location must be included separately. Confirm both paths in the workspace settings before relying on a backup.

Eagle: Back up the entire .library folder. The metadata database is stored inside the managed library alongside the imported media. If you retained and still rely on source originals outside Eagle, include those source folders in your backup plan too.

PureRef: Each .pur file is self-contained. Back up the folder where you store your .pur files. There is no separate database.

BeeRef: Same as PureRef. Back up the folder where your .bee files live.

Step 4: Test a restore

A backup you have never tested is not a backup you can rely on.

Once you have set up your backup, simulate a restore with a small set of files:

1. Copy a folder of 10 to 20 reference images and the associated index or board file to an isolated temporary location.
2. Leave the working originals untouched.
3. Restore the test set from your backup into a second isolated location.
4. Open the restored test copy and verify that the images and metadata appear correctly.

For refern, this means pointing a workspace at the restored folder and confirming that your tags, ratings, and folder structure come back intact. For Eagle, it means opening the restored .library folder and verifying that your library appears as expected. For PureRef and BeeRef, it means opening a restored .pur or .bee file and confirming the images are present.

If the restore works, you have confidence. If it fails, you catch the problem before a real data loss event forces you to find it.

Step 5: Build a routine

Backups degrade over time if you stop running them. Build a simple maintenance habit.

• Weekly: Check that your automated backup ran successfully. Most backup tools send a status notification; look for it.
• Monthly: Verify the backup destination has enough space. A full backup drive stops making new copies silently in some tools.
• Before a major reorganization: Run a manual backup before doing a large tag cleanup, folder restructure, or library migration. If something goes wrong mid-operation, you have a clean snapshot to restore from.
• After importing a large batch: If you import 500 new images and spend an hour tagging them, trigger an immediate backup before closing the session.

Comparison: backup difficulty by tool

How refern handles backup differently

During normal workspace indexing, refern reads existing files in place without duplicating or moving them. Explicit imports and copy actions can create files. Standard backup tools can protect the workspace, but include both the media root and any configured data location.

refern also includes periodic automatic backups with one-click restore built into the app. If something goes wrong with the SQLite index (such as a crash during a write), refern can restore from a recent backup in one click from the Settings panel. This is separate from your personal backup strategy but adds a layer of protection for the metadata even if you have not configured an external backup yet.

When you point refern at an existing folder, it indexes the files in place without touching them. If you later decide to use a different tool or stop using refern entirely, your original files are exactly where they were, completely unchanged, and your folder organization is preserved on disk. There is no export step and no lock-in. Day to day, you can drag files straight out of the grid into any other app or into your file manager and they arrive as real files, whether you drag one, a multi-selection, a folder, a canvas, or a group.

For artists currently using Eagle, refern includes an importer with full parity across its supported transfer surface. It copies the first media placement into the refern workspace and represents later Eagle folder placements as linked database records. Password locks, saved smart-folder rules, arbitrary plugin-owned data, and unsupported media are outside that surface. For more on what the canvas and graph features add on top of the library, see the refern vs Eagle comparison and the best Eagle alternatives guide.

Common problems and fixes

"My backup drive ran out of space and stopped backing up months ago." Check your backup tool's status screen. Prune old backup snapshots, archive completed project folders to cold storage (like Backblaze Vault or Amazon Glacier), or add a larger drive. Set a calendar reminder to check backup storage quarterly.

"I lost a PureRef board and the .pur.old file is from three weeks ago." If you do not have a more recent copy from an external backup, the board content from the last three weeks is likely unrecoverable. Going forward, save PureRef boards to a folder that Dropbox or another sync service watches, so each save produces a versioned copy in the cloud.

"My Eagle library folder is larger than I expected." Eagle's standard import copies media into its managed library. If you retained the source originals, their media bytes also exist outside the library; generated data and later file handling affect the final total. Your options are to increase backup capacity, move the .library folder to a drive with more space, or migrate. refern's Eagle importer creates a refern-side copy for the first media placement and linked database placements thereafter, so budget for that destination copy during migration.

"My library lives in a synced folder and my tags stopped saving." This is the failure mode to watch for if you put a workspace inside OneDrive, Google Drive, iCloud Drive, or Dropbox. When the cloud account runs out of quota, the sync client can mark files in that folder read-only, and that includes the SQLite index your tags and ratings live in. Reads keep working, so the library looks completely normal, but every write is rejected. refern now detects a read-only database when it opens the workspace and tells you, instead of accepting edits that quietly go nowhere. If you hit it, free up space in the cloud account, clear the read-only attribute on the folder, or move the workspace to a normal local folder and back that folder up instead of syncing it. A synced folder is a convenience, not a backup, and it is worth checking your quota on the same quarterly reminder you use for backup drive space.

"I reorganized my refern workspace folder externally and now some paths look wrong." Run the reconcile function (Dashboard, "Sync with disk") to let refern diff its index against your current folder state and update paths for moved files. Back up before running a large reconcile to have a clean fallback.

Next steps

A backup strategy is the foundation, but organization makes a library useful even after you have protected it. Once your backup is running:

• Learn how refern's tags and smart folders can surface the right image at the right moment without manual browsing.
• If you are migrating from PureRef and want to understand the organizational differences, the refern vs PureRef comparison covers them in detail.
• If you are coming from Eagle, the best Eagle alternatives guide explains what to look for in a replacement.