PureRef Lost Your References? Why It Happens and a Safer Setup (2026)

By refern | Last updated: June 2026

PureRef lost your references because every image you add is embedded inside a single .pur binary file. When a save is interrupted by a power cut, a full disk, or a crash mid-write, that file can corrupt. At least three independent users reported losing months of work to this exact scenario on the official PureRef forum. [pureref.com/forum/read.php?5,1367] This article explains the recovery path, shows you how to set up a safer backup routine within PureRef, and covers what a fundamentally different architecture looks like if you want to remove the single-file risk entirely.

Why PureRef files go missing or corrupt

PureRef is a canvas-first tool built around a proprietary .pur file. That file is a binary that embeds your images directly: there are no separate image files on disk once a board is saved. The design keeps things portable (one file to share or move), but it creates a single point of failure.

When PureRef saves, it writes new data to the .pur file. If that write is interrupted, the output can be a partial or malformed binary that PureRef cannot read. The triggers reported on the official forum include:

• Power loss or sudden shutdown while PureRef is saving a large board
• Disk full condition that truncates the write mid-stream
• App crash during a manual save or an autosave cycle

The forum thread where users reported this ([pureref.com/forum/read.php?5,1367]) includes descriptions like "months worth of references" lost and users who discovered the damage only when they reopened the file the next day. The .pur file showed no obvious sign of trouble until PureRef tried to parse it.

This is not a bug unique to PureRef. Any tool that writes a large binary in a single operation carries this risk. PureRef mitigates it with a rolling backup, but that backup is easy to miss.

How to recover a corrupt or missing PureRef file

PureRef writes a .pur.old file alongside your main .pur every time it saves. This one-generation backup is your only built-in recovery path. It is not prominently documented in the UI.

Step 1: Look for the .pur.old file

Navigate to the folder that contains your .pur file. You should see two files:

MyReferences.pur
MyReferences.pur.old

The .pur.old file is the state of your board from the previous successful save.

Step 2: Replace the corrupt file

If MyReferences.pur will not open or opens blank, do the following:

1. Rename MyReferences.pur to MyReferences.pur.broken (keep it as a safety copy).
2. Rename MyReferences.pur.old to MyReferences.pur.
3. Open the renamed file in PureRef.

You will lose work done between the last successful save and the corruption event. How much depends on how often PureRef was saving.

Step 3: If the backup is also corrupt or missing

If MyReferences.pur.old is also unreadable, your options are limited:

• Check cloud backup services (Dropbox, OneDrive, Google Drive) for a versioned copy of the .pur file.
• Check your operating system's shadow copies or Time Machine snapshots.
• If neither is available, the images are not recoverable from PureRef itself. The originals are embedded in the binary; there is no image folder to fall back on.

This is the core limitation: PureRef keeps no separate copy of your source images. If both the .pur and the .pur.old are gone or corrupt, and you have no external backup, the references are gone.

How to back up PureRef references properly going forward

PureRef does not offer a built-in multi-generation backup or cloud sync. A manual backup routine is the only protection beyond the single .pur.old rolling file.

Minimum viable backup

Copy your .pur file (and the .pur.old alongside it) to a second location after each work session. An external drive, a cloud folder (Dropbox, Google Drive, iCloud, OneDrive), or a NAS all work. Automate this with your OS backup tool or a sync utility so it happens without thinking.

Key points:

• Back up both the .pur and the .pur.old in the same copy operation. If the main file corrupts silently, you want the .pur.old from before the damage happened.
• Version your backups. A single-destination sync that always overwrites will not help if corruption went unnoticed for a few sessions. Keep at least 3 to 5 dated copies, or use a cloud service with file version history (Dropbox version history is 180 days on the free plan; Google Drive keeps up to 100 versions).
• Large boards need more attention. PureRef loads all images into memory uncompressed [pureref.com/forum/read.php?2,1947], so large boards are both slower to save and more exposed if a save is interrupted. Boards with hundreds of images in high resolution are worth backing up more frequently.

Tips for reducing corruption risk

• Save often. PureRef has a manual Ctrl+S shortcut. Use it after any significant addition or arrangement.
• Keep disk space available. A full disk is one of the documented corruption triggers. Keep at least a few gigabytes free on the drive where your .pur files live.
• Close PureRef before shutting down. A hard power-off while PureRef is mid-save is the highest-risk scenario. A clean application quit ensures the final save completes.
• Split large boards. The official PureRef guidance for RAM issues recommends splitting a large board into multiple smaller .pur files. [pureref.com/forum/read.php?2,1947] Smaller files also mean a corruption event loses less work.

PureRef's genuine strengths (be clear-eyed)

Before discussing alternatives, it is worth being fair about what PureRef does well. It is not the right tool for every workflow, but for what it does, it does it genuinely well.

PureRef is lightweight, starts instantly, and puts a reference board on screen while you work in another app. Its always-on-top overlay is best-in-class, with the ability to pin above a specific application like ZBrush or Photoshop. Its transparent-to-mouse (click-through) mode lets artists eye-drop colors from a reference directly into their painting app. These are purpose-built features that no library-first tool matches for the "session overlay" use case.

PureRef is also effectively free for personal non-commercial use (pay-what-you-want, suggested $7 or $15 as of 2026 [pureref.com/download.php]), which makes it accessible to students and early-career artists. It has been in use since 2013 and is taught in concept art and game art curricula alongside Photoshop and ZBrush. [conceptartempire.com, therookies.co] That trust is earned.

The .pur single-file format is a feature as much as a limitation: one file is easy to share, easy to move between machines, and easy to archive per project. For short-lived, per-assignment moodboards, it works well.

The limitations that lead to lost references are real, but they exist alongside genuine strengths. This article is about helping you avoid data loss in PureRef, not arguing you should leave it.

How refern's architecture avoids the single-file failure mode

refern takes the opposite approach to file storage: it never embeds your images in a proprietary container. A refern workspace is a normal folder on your disk. refern builds an index (SQLite + thumbnails) alongside your originals, but your original files stay where they are, as plain files. A crash, a power cut, or a corrupted index cannot destroy your references because the references are just files in a folder.

Here is a direct comparison of the two approaches:

If the SQLite index in refern becomes corrupted (for example, due to a power cut during a thumbnail write), refern can rebuild it by re-scanning your folder. Your images are unaffected because they were never moved or modified. This is the fundamental difference: refern's crash-resumable pipeline [as built into the workspace lifecycle] picks up where it left off. A corrupt index is an inconvenience, not a loss.

refern is not immune to all data loss. If your drive fails or you delete files from your folder, those originals are gone just as they would be with any other tool. The point is narrower: the specific failure mode of a proprietary save file corrupting your entire collection does not apply when your originals are plain files on disk.

What refern adds beyond safer storage

Beyond the architecture difference, refern is a full reference library tool:

• Full-text search across filenames, descriptions, notes, source URLs, creators, and tags
• Hierarchical tags, tag groups, linked tags, and tag macros
• Color search (find images by hex color or by clicking a color swatch)
• Image-to-image visual similarity search
• Smart folders (saved search queries that update automatically)
• Infinite canvas with layers, groups, text, shapes, freehand drawing, image filters, and non-destructive crop
• Typed entity links (grouped, derived-from, placed-in-canvas, cross-reference) and a navigable relationship graph
• Browser extension for Chrome, Firefox, and Safari with hover-save and batch-save
• Import from Eagle (folders, tags, ratings, sources, notes)
• Reads embedded EXIF, IPTC, and XMP metadata on import
• Scales to large libraries (a user with 27,000 images confirmed smooth performance)

An alpha user put it plainly after the launch: "organization and search like eagle cool, canvas from pureref." If you use PureRef purely as a session overlay and clear it after each project, it may remain the right tool for that role. If your reference collection is growing beyond individual projects and you want to find images later by tag, color, or source, refern is designed for that.

For a full feature-by-feature breakdown, see the refern vs PureRef comparison.