Download refern.

Installing refern

1. v1.4.1LatestJuly 21, 2026

   Video Playback, Thumbnails & Network Drives

   • Added a 'Viewer Backdrop' setting so the area around an image or video can be a solid background, translucent, or black. It now defaults to a solid background: the old frosted glass tinted whatever you were looking at with the page behind it
   • Hold Alt (Option on macOS) and scroll over a video to scrub the timeline, one second per notch. Scrolling on its own still zooms. Both are listed in the ? shortcuts sheet
   • Copying a file now behaves the way your file manager does: paste it into another app as the file itself, or into an image editor as the picture. Videos can be copied too
   • refern now recognizes 11 more video file types, so clips it used to skip are indexed with everything else

   + 12 more changes in this release
2. v1.4.0July 19, 2026

   Native Video, Hidden Folders & a Unified Viewer

   • Added native video support: import and organize videos alongside your images, with a player built for reference work. Step frame by frame, scrub a storyboard preview straight from the timeline, read the waveform and filmstrip, set speed and volume, loop, resume where you left off, and pop the video out into picture-in-picture. Videos your system can't play are converted to a playable copy automatically
   • Mark an A and a B point on a video to define a section, then loop it, export it as an .mp4 clip, cut it losslessly with no re-encode, save it as an animated .webp loop (up to 30 seconds), or trim the original (undoable)
   • Add labelled markers anywhere on a video, save any marker's frame as an image, and export your markers and section to .otio and .edl timeline files for your editing app
   • Copy or save the current frame, pull a color palette from it (and search your library by those colors), set a custom thumbnail or display image, and build a canvas board from a video or from its markers

   + 37 more changes in this release
3. v1.3.1July 6, 2026

   Multi-Root Workspaces & a Smarter Auto Tagger

   • A workspace can now hold several separate content folders at once instead of exactly one. Add, rename, re-point, or remove each root from workspace settings, and browse, search, and import across all of them together. Useful when your references live in more than one place on disk
   • Upgraded the Auto Tagger's models: general tags, art tags, and 'Search by meaning' all got a big quality boost, text recognition now reads longer lines more accurately, and noisy filler tags (like '1girl' or 'no-humans') are filtered out automatically. Run a fresh pass from the Pipeline tab to re-tag and re-index with the new models
   • Added a Configure tab to the Auto Tagger: browse every tag the models can produce, then block the ones you don't want or rename them to fit your library, so you can shape suggestions instead of accepting raw output
   • You can now re-tag items that were already tagged (a Pipeline option), and wipe all suggested tags from a single item or a whole selection in the metadata sidebar, so refreshing or clearing tagging data no longer means starting over

   + 26 more changes in this release

Support