PhotoPrism

This release introduces new detection and embedding models for face recognition: more faces are found, including small ones in group pictures, and clustering now keeps different people apart instead of grouping them as one. Existing libraries keep their current model until you run photoprism faces migrate, which re-embeds every face and keeps the names you have already assigned.

People whose names should remain private can now be flagged as Private or Hidden to hide their names and face regions from non-privileged accounts, while their pictures remain visible.

In addition, 360° photos, videos, and fisheye DNG files from Insta360 cameras can be viewed directly, maps have moved to MapLibre GL JS v6, and the default size limit has been raised to 5 GB so that larger videos are indexed without changing the configuration.

A special thank you to all our members and everyone who contributed! 🌈✨

What’s new?

• Viewer: Added support for 360° Insta360 photos, videos, and fisheye DNG by @omerdduran
• AI: Improved facial recognition with new detection and embedding models by @omerdduran
• AI: Added a Normalize option to keep compound label names whole
• AI: Upgraded ONNX Runtime from v1.26.0 to v1.29.0
• People: Improved clustering to separate people whose faces were grouped as one
• People: Added a confirmation before a typed name creates a new person
• People: Added Private and Hidden options to hide names and face regions
• People: Fixed two people being merged when reassigning a face
• People: Improved error handling when updating counts by @keif888
• People: Improved memory usage and performance when naming clusters
• People: Improved the speed of face updates in very large libraries
• Places: Upgraded MapLibre GL JS from v5 to v6 (requires WebGL 2)
• Index: Added support for Cineon, ORI, and camcorder video files
• Index: Improved support for large files by raising the default size limit to 5 GB
• Thumbs: Improved oversized preview requests to return the largest available size
• Thumbs: Improved error messages to be logged as a single readable line
• Covers: Improved album, label, and folder covers to load from the thumbs cache
• Metadata: Improved the docs to include complete XMP and DCMI tags
• Config: Added a thumb-size-face option that limits the sources rendered for face crops
• Config: Added a face-cluster-core-retry option to control the second clustering pass
• CLI: Added faces migrate to switch a library to a new face model
• CLI: Added faces reset --all to start over without detecting faces again
• CLI: Added faces status to help with configuration and troubleshooting
• CLI: Added faces subjects, faces ls, and faces markers to inspect face data
• CLI: Added a Detail column showing how well a face crop was resolved
• Helm: Added a values schema that validates the chart configuration
• Tests: Improved backend and acceptance test coverage by @keif888
• Security: Upgraded libheif from v1.23.1 to v1.23.4 (8 CVE fixes)
• Security: Upgraded Go from v1.26.5 to v1.27.1
• Translations: Improved French by @jean-louis67
• Translations: Improved Hebrew by @avma

Translations

Missing user interface translations have been generated with the help of DeepL and Google Translate. Native speakers are welcome to help us improve them where needed. Learn more ›

Installation Packages

The packages attached to this release are intended for experienced users and maintainers of third-party integrations only, as they require manual configuration and do not include tested system dependencies. Since we are unable to provide support for custom installations, we recommend using one of our Docker images to run PhotoPrism on a private server or NAS device. Learn more ›