Frequently Asked Questions¶
What are sidecar files and where do I find them?¶
A sidecar is a file which sits alongside your main photo or video files, typically using the same name, and a different extension like
New sidecar files will be created in the storage folder by default so that the originals folder can be mounted read-only.
PhotoPrism will always look out for existing sidecar files and use them for indexing,
PHOTOPRISM_DISABLE_BACKUPS are set to
Three types of metadata sidecar files are supported currently:
If not disabled via
--disable-exiftool, Exiftool is used to
automatically create a JSON sidecar for each media file.
This way, embedded XMP and video metadata can be indexed as well.
Native metadata extraction is limited to common Exif headers.
Note that this causes moderate overhead when indexing for the first time.
JSON files may also be useful for debugging as they contain the complete metadata, and can be processed using common development tools and text editors.
PhotoPrism can also read JSON files exported by Google Photos. Support for additional schemas may be added over time.
If not disabled via
--disable-backups, PhotoPrism will automatically create / update
YAML sidecar files while indexing and after manually editing fields like title, date, or location.
They serve as a backup in case the database (index) gets lost, or when folders are synced with a remote
Like JSON, YAML files can be opened using common development tools and text editors. Changes won't be synced back to the original index though as this might overwrite existing data.
XMP (Extensible Metadata Platform) is an XML-based metadata container format invented by Adobe. It offers much more fields (as part of embedded models like Dublin Core) than Exif. That also makes it difficult - if not impossible - to provide complete support. Reading Title, Copyright, Artist, and Description from XMP sidecar files is implemented as a proof-of-concept, contributions welcome. Indexing embedded XMP is only possible via Exiftool, see above.
Which folder will be indexed?¶
If you skip configuration and don't use one of our Docker images, PhotoPrism will try to find
a photo library by going through a list of common
~/Photos. It will also search for other resources
like external applications, classification models, and frontend assets.
Your library will be mounted from
~/Pictures by default when using our
example docker-compose.yml file, where
~ is a placeholder for your home directory.
You may mount any folder accessible from your computer, including network drives. Note that PhotoPrism won't be able to see folders that have not been mounted unless you install it locally without Docker (developers only).
Multiple folders can be indexed by mounting them as sub-folders of
volumes: - "~/Family:/photoprism/originals/Family" - "~/Friends:/photoprism/originals/Friends"
Which file types are supported?¶
PhotoPrism's primary image file format is JPEG. While indexing, a JPEG sidecar file may automatically be created for RAW, HEIF, TIFF, PNG, BMP, and GIF files. It is required for classification and resampling.
Only MPEG-4 AVC video files are fully supported for now. Transcoding of other codecs is planned for a later release.
PHOTOPRISM_DISABLE_EXIFTOOL must be
"false" or unset in order to
extract and index metadata like location and duration from video files.
You're welcome to open an issue if you experience issues with a specific file format.
Why is my storage folder so large? What is in it?¶
The storage folder contains sidecar, thumbnail, and configuration files. It may also contain index database files if you're using SQLite. Most space is consumed by thumbnails: These are high-quality resampled, smaller versions of your originals.
Thumbnails are required because Web browsers do a pretty bad job at resampling large images so that they fit your screen. Using originals for slideshows and search result previews would consume much more browser memory, and reduce overall performance, as well.
If you're happy with lower quality thumbnails, you can reduce their JPEG quality and/or set a size limit. Note that existing thumbnail files won't be replaced automatically after changing config values.
You may also choose to render thumbnails on-demand if you have a fast CPU and enough memory. However, storage is typically affordable enough for most users to go for better quality and performance instead.
Can I skip creating thumbnails completely?¶
The smallest configurable limit is 720px for consumption by the indexer and TensorFlow during image classification. Recreating and keeping them in memory is too demanding, even for the most powerful servers.
I'm having issues understanding the difference between the import and originals folders?¶
Import is a temporary folder from which you can move or copy files to originals in a structured way that avoids duplicates. Most users with existing collections will want to index their originals folder without import, so that existing file and directory names don't change. On the other hand, importing may be more efficient when adding files as you don't need to re-index originals.
What exactly does the read-only mode?¶
There are users who don't want us to modify their original files and folders in any way, so we've added a configuration option for this use case. It will disable uploads, import and future features that might rename, update or delete files in the originals folder.
I could not find a documentation of config parameters?¶
You may run
photoprism help in a terminal to see all options and commands.
We also maintain a complete list of config options in these docs.
Our Docker Compose examples are continuously maintained and inline documentation has been added to simplify installation.
Can I deploy PhotoPrism on a shared domain like
No. This is not recommended, and currently not supported for the following reasons:
- It's less effort, less complexity, and requires less testing on our side. We prefer to invest our time in feature requests that provide more value to our users.
- Browser connections, routing, and related resources don't need to be shared with other apps, so it's faster and easier to configure for you.
- Cloud photo apps typically use multiple sub domains for faster loading - reducing the number of hosts and putting everything on a shared host is the exact opposite of what you should do.
- It's easier for you to clear the browser cache, or change PhotoPrism specific settings for ad blockers.
- The same origin policy generally works on the (sub-)domain level, so it may be insecure to host multiple apps on the same domain - at least without significant effort in additional security measures.
- Share URLs are shorter and easier to remember.
Why is PhotoPrism getting stuck in a restart loop?¶
These restarts are triggered by Docker (based on your configuration) when PhotoPrism is unable to start. They are typically caused by one or more of the following reasons:
- Your (virtual) server disk may be full.
- The storage folder may not be writable.
- Your database server may be unavailable.
- There are connection issues caused by a proxy or firewall.
- Kernel security modules such as AppArmor or SELinux may be blocking permissions.
Please check the server logs for a detailed error message like "disk full" or "wrong permissions". If you're using Docker Compose, you may enter this command to see the last 20 log entries:
docker-compose logs --tail=20
Linux kernel security may be disabled on private servers, especially if you have no experience
with configuring it properly. Use the commands
chown to fix file system permissions
on Linux and macOS.
Available disk space can be displayed with
df -h. The size of virtual disks and memory can be
increased in Docker settings.
How can I uninstall PhotoPrism?¶
This depends on how you installed it. If you're running PhotoPrism with Docker Compose, this command will stop and remove the Docker container:
docker-compose rm -s -v
Please refer to the official Docker documentation for further details.
How can I mount network shares with Docker?¶
There are multiple ways of using network storage. One of the easiest might be to directly mount NFS shares with Docker.
You can mount any number of NFS shares as folders. Follow this
if you want to mount the originals folder as a share:
services: photoprism: # ... volumes: # Map originals folder to NFS: - "photoprism-originals:/photoprism/originals" volumes: photoprism-originals: driver: local driver_opts: type: nfs # The IP of your NAS: o: "username=user,password=secret,addr=18.104.22.168,soft,rw" # Share path on your NAS: device: ":/mnt/photos"
For Windows / CIFS shares:
volumes: photoprism-originals: driver: local driver_opts: type: cifs o: "username=user,password=secret,rw" device: "//host/folder"
This was tested with TrueNAS and NFS, but other (network) file systems may be mounted with Docker as well.
Mounting the import folder to a share which is also accessible via other ways (e.g. CIFS) is especially handy, as you can dump all data from a SD card / camera directly into that folder and trigger the index in the GUI afterwards. So you can skip the upload dialog in the GUI and it's a little faster.
I'm using an operating system without Docker support. How to install and use PhotoPrism without Docker?¶
In general, you would build / install it like a developer since we don't have packages for specific operating systems yet.
Instead of using Docker, you can manually type the commands listed in our development Dockerfile and replace packages with what is available in your environment. You often don't need to use the exact same versions for dependencies.
If your operating system has Docker support, we recommend learning Docker as it vastly simplifies installing and upgrading.
Do you support Podman?¶
Podman works just fine both in rootless and under root. Mind the SELinux which is enabled on Red Hat compatible systems, you may hit permission error problems.
Do you provide LXC images?¶
Any plans to add support for Active Directory, LDAP or other centralized account management options?¶
There is no Active Directory, LDAP, or Single Sign-On support yet as we didn't consider it essential for a first release. It might be added later, maybe as a premium feature for our sponsors and contributors.
Our development and testing efforts are focused on small servers and home users. Adding functionality that is primarily useful for business environments, or that only benefits few private users with special needs, diverts resources away from features that benefit everyone. Professional users are welcome to reach out to us for a custom solution.