Yes. Luminary works against folders on your machine and keeps its working metadata inside the selected media root.
No. There is no cloud sync or hosted service built into the extension.
Inside the root folder you choose, in a .luminary directory:
scan-cache.jsontags.jsonpeople.json
Mostly no.
- scanning and browsing are read-only
- tagging and people data are stored in
.luminary - export creates copies
- duplicate review can move selected files to the OS trash, but only after confirmation
- Facebook import copies files into a target folder and writes recovered dates to the copied files
No. Face recognition is optional.
The core browsing, tagging, duplicate review, import, and export workflows do not depend on it.
Luminary starts a separate Node.js worker for face recognition to avoid the Electron and native module mismatch inside the VS Code extension host.
For the best experience:
- install Node.js
18+ - make sure
nodeis available onPATH
Luminary will try GPU TensorFlow first, then Node TensorFlow, then the plain JavaScript backend.
Yes, for timeline scanning, duplicate detection, and export workflows. Image-specific features such as face recognition only apply to supported image files.
It scans a Facebook export posts folder, recovers dates from the export HTML, copies the media into a target folder, and organizes it into year/month directories. JPEG copies get EXIF dates written where possible. Other copied files get their file modification time updated.
Close the extension host, then remove the .luminary folder inside the media root you configured. Luminary will rebuild its state from the collection on the next scan.
Not yet. It is usable, but still early. The goal of this repo is to make the project understandable, configurable, and safe for other people to try and improve.