~/tools/photoprism
PhotoPrism
tool

PhotoPrism

PhotoPrism is an open source, self-hosted photo and video app — a Google Photos alternative that indexes the folders you already have on disk and helps you rediscover them with AI tagging, face recognition, search, and maps, all on your own server.

What is PhotoPrism?

PhotoPrism is an open source, self-hosted application for browsing, organizing, and searching photos and videos. Written in Go with a TensorFlow-powered indexer, it points at the folders you already keep on disk, reads their metadata, and adds automatic tags, face recognition, and maps — leaving your original files exactly where they are.

What is PhotoPrism best for?

PhotoPrism is best for people who already have a large, organized photo library on a NAS or server and want to search and rediscover it without moving anything. Its read-only indexing and WebDAV access make it a strong fit for photographers and archivists who care about keeping their own folder structure and file names intact.

What can PhotoPrism do?

  • Index existing folders on disk without changing or moving your files
  • Auto-tag photos by content and recognize faces with on-device machine learning
  • Search by label, location, color, camera, resolution, and quality
  • Place trips on high-resolution world maps using privacy-preserving geocoding
  • Handle RAW files, Live Photos, and video, with metadata from Exif, XMP, and Google Photos takeouts
  • Serve your library over WebDAV so other apps can read and sync it
  • Run anywhere Docker does — Linux, a Synology NAS, or a Raspberry Pi

Where does PhotoPrism fall short?

  • No official mobile app. PhotoPrism ships as a progressive web app only. There’s no first-party phone app that automatically backs up your camera roll — you rely on third-party tools like PhotoSync or a WebDAV client, which is a real gap if automatic phone backup is your main goal.
  • Single-library, single-user roots. PhotoPrism was built around one shared library rather than separate per-user accounts; multi-user roles and a user-management UI are newer and partly gated behind paid tiers, so it’s a weaker fit for households wanting truly separate photo spaces.
  • Some polish sits behind membership. The core app is free and complete, but extras like 3D vector and satellite maps and enhanced geocoding require a paid membership or commercial license.

Is PhotoPrism free?

Yes — the Community Edition is free and open source under the AGPL-3.0 license, and it includes the features most people want, including AI tagging, search, and face recognition. Optional paid memberships and commercial licenses add extras like advanced maps, additional user roles, and priority support, and a managed cloud option is available through PikaPods from about $6.50/month. Self-hosting is comfortable on 4 GB of RAM.

What does PhotoPrism replace?

PhotoPrism is a self-hosted stand-in for Google Photos, iCloud Photos, and Amazon Photos. It gives you the same tagging, search, face recognition, and shared browsing, but your library stays on your own storage instead of a subscription cloud tied to Google, Apple, or Amazon.

FAQ

Is PhotoPrism open source? Yes. PhotoPrism is licensed under the AGPL-3.0, an OSI-approved open source license, and the full source is on GitHub. Paid memberships fund development and unlock a few extras but the core app stays open.

Can I self-host PhotoPrism for free? Yes. The Community Edition is free to run on your own hardware — you only pay for the server. It’s happy on around 4 GB of RAM, which is lighter than many machine-learning photo apps.

Is PhotoPrism a good Google Photos alternative? For organizing and searching an existing library, yes — the AI search and face recognition are excellent. If your priority is automatic phone backup, Immich is the better-fitting Google Photos alternative because it ships official mobile backup apps.

What do I need to run PhotoPrism? A machine with Docker and roughly 4 GB of RAM, plus storage for your photos. It runs well on a Linux server, a Synology or other NAS, or a Raspberry Pi.