Photonix
photonix.org
- Free plan
- Web
- iPhone and iPad
- Android
- Self-hosted
What Photonix is
Photonix is an open-source photo manager for organizing, presenting, searching, and rediscovering collections through a web browser. Its classifiers identify faces, objects, styles, colors, and locations for filtering. Face recognition groups faces across images, and people can be labeled without training a model. Photonix says object detection runs within the software without sharing images externally. It compares metadata to connect related versions, including RAW files, camera JPEGs, and edits, and watches selected folders for changes from Lightroom, Darktable, or Photoshop without plugins. The maker says it reads RAW formats including CR2, NEF, DNG, ARW, and RWL. Photonix is self-hosted with Docker Compose examples for x86/amd64, ARM32/ARM64, and Raspberry Pi. Android and iPhone apps require a Photonix server. Analysis runs on that server, while account and metadata data stay on installation machines and chosen backups. The free plan is free.
Who it suits
Photonix suits photographers and professionals who want self-hosted photo organization within existing editing workflows. It also fits users seeking local analysis and open-source software.
Good to know, and what to watch
- Classifies faces, objects, colors, styles, and locations
- Supports major camera RAW formats
- Tracks related RAW, JPEG, and edited versions
- Works with Lightroom, Darktable, and Photoshop folders
- Runs analysis on the installed server
- Requires a self-hosted server
- Mobile apps have no useful function without a server
- Photonix alone is not a photo backup
- Video support was not yet available
Plans
Every plan and pricePhotonix
Open source
The verdict
Photonix offers broad search and organization tools while keeping processing on a self-hosted installation. Its server requirement and lack of built-in backup or available video support matter when planning a photo library.
What the maker publishes
- Audience
- The maker describes Photonix as photo management software for professionals and says it is intended to fit into photographers’ existing editing workflows. source
- Editing integrations
- It watches user-selected folders for edits and says no plugins are needed to work with tools such as Lightroom, Darktable or Photoshop. source
- Face recognition
- Face recognition groups detected faces across photos, and users can label people without training a model. source
- Local processing and data
- The privacy policy says photo analysis runs on the installed server and account, metadata and app data stay on the installation machines and any chosen backups. source
- Mobile apps
- Android and iPhone apps connect to a Photonix server and do not provide useful functionality without one. source
- No backup guarantee
- The maker says Photonix alone cannot provide a backup when photos remain on the same hard drive, and recommends maintaining separate copies. source
- Object detection privacy
- The maker says object detection runs within Photonix and does not share images externally. source
- Photo versions
- Photonix compares photo metadata to identify related versions, including RAW files, camera JPEGs and edits. source
- Purpose
- Photonix organizes photo collections for storing, presenting, searching and rediscovering photos through a web browser. source
- RAW support
- The maker says Photonix reads major camera RAW formats including CR2, NEF, DNG, ARW and RWL. source
- Self-hosted deployment
- The installation guide uses Docker Compose and provides an example for x86/amd64 and ARM32/ARM64, including Raspberry Pi. source
- Smart tagging
- Its classifiers detect faces, objects, image styles, colors and locations for photo filtering. source
- Support and community
- The maker describes Photonix as an open-source community project and points users to GitHub issues, chat discussions and documentation. source
- Video limitation
- The FAQ says video support was not yet available and describes limited video support as planned. source
Checked against the maker's pages on 9 Oct 2026.