v1.51.0

[alextran1502]

v1.51.0

Breaking Changes 🚨

This release added a new container to Immich's ecosystem, namely Typesense. Please update your docker-compose.yml file to include the new container and .env file to have the new environment variable, which is TYPESENSE_API_KEY=some-random-text.

Please make sure to have your server and mobile app on the same version so that the app can work correctly.

Typesense container is required for now, but we plan to make it optional with the fallback of using Postgres full-text search for the searching mechanism.

  typesense:
    container_name: immich_typesense
    image: typesense/typesense:0.24.0
    environment:
      - TYPESENSE_API_KEY=${TYPESENSE_API_KEY}
      - TYPESENSE_DATA_DIR=/data
    logging:
      driver: none
    volumes:
      - tsdata:/data

[...]

volumes:
  pgdata:
  model-cache:
  tsdata:

Highlighted

This is one of the more extended release breaks (a total of 53 pull requests were merged) that we've had for a while due to some large items that we've been working on, such as:

Integrating Typesense and CLIP for a better search system

As mentioned in the previous releases, we want a better search system. As Immich aims to be a fully featured Google Photos alternative, we want to provide you with the best search experience possible. We've been working on integrating Typesense and CLIP into Immich to provide you with a top-notch searching experience for your larger libraries.

We chose Typesense for its open-source nature, attractive features, and self-hosting capability.

We chose CLIP for its ability to search for images based on their content rather than their metadata. So you can do something like this:

The CLIP embedding data is a new column in the database. To get that info for your previously uploaded assets, please consider running the ENCODE CLIP job in the admin’s job panel

We will continue to iterate and improve the search feature in the subsequent releases. We would appreciate any feedback about the initial implementation - what works well and what areas could be improved.

Note: The mobile app does not incorporate this feature yet. It is only applicable to the web in this release.

Allow the mobile app to work offline without an internet connection or when the server is down.

Previously, when the server was down, or your phone didn't have an internet connection, tapping on anything in the mobile app would send you back to the login screen. This is because the mobile app couldn’t retrieve any data from the server. We've added a new feature that allows the mobile app to work offline, so you can still view your photos and videos cached or stored on your device even when the server is down or you don't have an internet connection.

In addition, when viewing photos and videos residing on your device, these are no longer loaded from the server but directly from the device. This reduces your mobile data usage and increases the loading speed of thumbnails while scrolling the timeline.

This has been a more significant feature that required a lot of refactoring from @fyfrey over many months-long processes. We're pleased to be able to release this feature to you finally.

As in other releases, we've also added many smaller features and bug fixes, so please check out the full changelog below.

Thank you

I want to thank everyone who contributed to this release, especially @fyfrey for the huge amount of work he put into the offline mode feature, @jrasm91 and @zackpollard for the refactoring of the server and the integration of the new search system, @martyfuhry for improving the UI/UX of the mobile app, @michelheusschen for improving the web code base and consistency, @bo0tzz for the never-ending stream of CI/CD improvement and answering questions on Discord, and all the new contributors who spared some of their free time to contribute to the project.

This project is for the community, and from the community, I am honored to be a part of it.

---

As always, please consider supporting the project.

🎉 Cheer! 🎉

Support

If you find the project helpful and it helps you in some ways, you can support the project one time or monthly from GitHub Sponsors

It is a great way to let me know that you want me to continue developing and working on this project for years to come.

What's Changed

Server

• feat(server): added additional storage template preset by @Chipwingg in feat(server): added additional storage template preset #1903
• chore(server): bump API version post release by @zackpollard in chore(server): bump API version post release #1909
• fix(server): storage template unit test by @michelheusschen in fix(server): storage template unit test #1906
• feat(server): Machine learning's image optimisations by @ollywelch in feat(server): Machine learning's image optimisations #1908
• fix(server): exif extraction swapped params by @michelheusschen in fix(server): exif extraction swapped params #1914
• feat(server)!: search via TypeSense by @jrasm91 in feat(server)!: search via typesense #1778
• chore(server): disable TypeSense logging by @alextran1502 in chore(server): disable TypeSense logging #1925
• fix(server): duration string parsing by @jrasm91 in fix(server): duration string parsing #1923
• feat(server): improve thumbnail relation and updating by @michelheusschen in feat(server): improve thumbnail relation and updating #1897
• feat(web,server): explore by @jrasm91 in feat(web,server): explore #1926
• fix(server): Increase authentication cookie max-age by @bo0tzz in fix(server): Increase authentication cookie max-age #1971
• fix(server): update album updatedAt on assets/users removed/added by @fyfrey in fix(server): update album updatedAt on assets/users removed/added #1977
• fix(server): remove encoded video file on asset delete by @raisinbear in fix(server): remove encoded video file on asset delete #1980
• fix(server): timezone mismatch in server tests by @brighteyed in Fix timezone mismatch in server tests #1918
• feat(server): CLIP search integration by @alextran1502 in feat(server): CLIP search integration #1939
• feat(server): Fallback to text search if machine-learning is disabled by @bo0tzz in feat(server): Fallback to text search if machine-learning is disabled #2015
• fix(server): CLIP search return empty result by @alextran1502 in fix(server): CLIP search return empty result #2018
• feat(sever): machine-learning image optimisations by @ollywelch in Feat/ml image optimisations #1916

Mobile App

• feat(mobile): Enter server first for login by @martyfuhry in feat(mobile): Enter server first for login #1952
• feature(mobile): sync assets, albums & users to local database on device by @fyfrey in feature(mobile): sync assets, albums & users to local database on device #1759
• fix(mobile): Fixes hero animation on main timeline by @martyfuhry in fix(mobile): Fixes hero animation on main timeline #1924
• fix(mobile): iOS backup debug info tile only shown in iOS by @martyfuhry in fix(mobile): iOS backup debug info tile only shown in iOS #1951
• feat(mobile): Transparent bottom Android navigation bar by @martyfuhry in feat(mobile): Transparent bottom Android navigation bar #1953
• fix(mobile): do not crash on malformed asset duration by @fyfrey in fix(mobile): do not crash on malformed asset duration #1921
• fix(mobile): Gallery viewer fullscreen edge case by @martyfuhry in fix(mobile): Gallery viewer fullscreen edge case #1959
• fix(mobile): Sorted shared album and added share user doesn't reflect change in album view by @alextran1502 in fix(mobile): Sorted shared album and added share user doesn't reflect change in album view #1955
• fix(mobile): add isar source code as a git submodule for F-Droid build by @fyfrey in fix(mobile): add isar source code as a git submodule for F-Droid build #1985
• fix(mobile): Fix iOS login layout issue by @martyfuhry in fix(mobile): Fix iOS login layout issue #2004
• feature(mobile): allow app to be used offline by @fyfrey in feature(mobile): allow app to be used offline #1932
• chore(mobile): remove integration test temporarily by @alextran1502 in chore(mobile): remove integration test temporarily #2008
• feature(mobile): no longer wait for background backup in settings by @fyfrey in feature(mobile): no longer wait for background backup in settings #1984
• feat(mobile): Share album name and adaptive shared album display by @martyfuhry in feat(mobile): Share album name and adaptive shared album display #2017
• feat(mobile): persist album sort order by @truedem0n in feat(mobile): persist album sort order #1997
• fix(mobile): proper syncing with Recents album on iOS

Web

• fix(web): restore album drag and drop upload by @michelheusschen in fix(web): restore album drag and drop upload #1933
• fix(web): nested layout navigation issue by @alextran1502 in fix(web): nested layout navigation issue #1936
• fix(web): admin pages layout issue by @michelheusschen in fix(web): admin pages layout issue #1943
• feat(web): require page load to export title by @michelheusschen in feat(web): require page load to export title #1956
• feat(web): improve search bar + add to search page by @michelheusschen in feat(web): improve search bar + add to search page #1957
• feat(web): Make scaling of albums overview more responsive by @matthinc in feat(web): Make scaling of albums overview more responsive #1981
• fix(web): short layout retention after navigation by @michelheusschen in fix(web): short layout retention after navigation #1994
• feat(web): improve /auth pages by @michelheusschen in feat(web): improve /auth pages #1969
• refactor(web): common layout for user pages by @michelheusschen in refactor(web): common layout for user pages #1995
• feat(web): bundle and 'sveltify' leaflet by @michelheusschen in feat(web): bundle and 'sveltify' leaflet #1998
• fix(web): storage template extension display by @etnoy in Fix storage template extension display #2002

Others

• chore(docs): add missing space in version_announcement_overlay_text_2 for ru-RU locale by @brighteyed in Add missing space in version_announcement_overlay_text_2 for ru-RU locale #1919
• chore(deps): bump docker/setup-buildx-action from 2.4.1 to 2.5.0 by @dependabot in chore(deps): bump docker/setup-buildx-action from 2.4.1 to 2.5.0 #1976
• chore(docs): added note on scope of redis optional parameters in example.env by @dragosrotaru in fix(docs): added note on scope of redis optional parameters in example.env #1974
• chore(docs): Document how photo dates are determined by @twitsforbrains in Document how photo dates are determined #1978
• chore(docs): Document fallback timezone setting (fixes [BUG] Invalid timezone in DSLR photos uploaded via CLI #2000) by @etnoy in Document fallback timezone setting (fixes #2000) #2003
• chore(docs): Document cli path parameter change by @etnoy in Document cli path parameter change #2011
• chore: Move away from docker hub where possible by @bo0tzz in chore: Move away from docker hub where possible #2006

New Contributors

• @brighteyed made their first contribution in Add missing space in version_announcement_overlay_text_2 for ru-RU locale #1919
• @dragosrotaru made their first contribution in fix(docs): added note on scope of redis optional parameters in example.env #1974
• @raisinbear made their first contribution in fix(server): remove encoded video file on asset delete #1980
• @twitsforbrains made their first contribution in Document how photo dates are determined #1978
• @etnoy made their first contribution in Fix storage template extension display #2002
• @truedem0n made their first contribution in feat(mobile): persist album sort order #1997

Full Changelog: v1.50.1...a

---

This discussion was created from the release v1.51.0.

[jackbrown1993]

Great release, good job to all involved. I think we should update the release notes to clarify the some-random-text API key for Typesense.

[Stoobiedoo]

I'm getting this on start up:
WARNING: The my-random-text variable is not set. Defaulting to a blank string.

And this error: "connect ECONNREFUSED" on the webpage

I've added the strings to the docker-compose.yml as
- TYPESENSE_API_KEY=${my-random-text}
And in the .env I added this to the end
TYPESENSE_API_KEY=my-random-text

Thanks for any help understanding how I messed it up.

[jrasm91]

${my-random-text} means it is looking for a env variable named my-random-text. You probably want TYPESENSE_API_KEY=${TYPESENSE_API_KEY}, which is how it is setup in the default docker-compose file.

[Stoobiedoo]

Thank you so much for the clarification. All fixed up now!

[elgatho]

Hello, when new version will be in App store? Just can't wait:)

[alextran1502]

Hello, when new version will be in App store? Just can't wait:)

They are being reviewed by the app stores now

[MozPri]

I'm getting error 500 after the update.

500 getaddrinfo ENOTFOUND immich-server

I'm using Portainer with the latest compose and .env files.
Also, the immich_typesense container does not start. State == exited.

Any suggestions?

[alextran1502]

Can you provide the server log?

[MozPri]

_immich_server_logs.txt

[alextran1502]

You might need to open an issue and provide your whole setup info so we can help. Make sure you bring all containers down then up again for a proper restart

[uninvitedguest]

@MozPri I'm running in portainer and received the same web error, but my issue was that while the typesense container would run, the server container would start and then immediately quit.

To resolve this I added the env_file: and stack.env lines to the compose, and I also removed any special characters from my API key (kept it at just letters and numbers). If you are using any custom networks, don't forget to add the typesense container to the custom network.

  typesense:
    container_name: immich_typesense
    image: typesense/typesense:0.24.0
    environment:
      - TYPESENSE_API_KEY=${TYPESENSE_API_KEY}
      - TYPESENSE_DATA_DIR=/data
    logging:
      driver: none
    volumes:
      - /mnt/ssd/immich/tsdata:/data
    env_file:
      - stack.env
    networks:
      - immich-internal

Note that I bound /data to a /tsdata volume on a separate drive (not that I know what goes in there). Nothing you need to replicate.

[uninvitedguest]

Thanks for your hard work on this! I'm exciting for the new functionality.

I got lucky that I took a deeper look at the .yaml before updating and caught that there was a new container. The docker compose file still notes Version 3.8, as it has for a long time. Would these changes not have warranted a version update to flag that there were changes to the .yaml?

[jrasm91]

The version key in the yaml file is the docker yaml specification version, not the immich server version.

See https://docs.docker.com/compose/compose-file/04-version-and-name/

[uninvitedguest]

Right, that's what I'm referring to. The .yaml received some updates (new container added, image references from docker hub changed to ghcr.io.)

Changing the yaml version would flag that there are changes to configuration that need to be captured when going through an update.

[alextran1502]

@uninvitedguest that is related to docker and docker-compose so it is not for Immich to use to keep track of the docker-compose.yml file changes

[Pixelartist]

I think this is great! Sadly it killed my setup since Typesense does not currently support an armv7 version:
Failed to deploy a stack: typesense Pulling no matching manifest for linux/arm/v7 in the manifest list entries
From the releases I can see they supported v8 a couple of times. Maybe you can see a quick way of fixing it?

[alextran1502]

The only way to fix this is for you to build the Typesense image on your machine and use that image in the compose file

[Pixelartist]

I looked through their github issues and it seems that currently there is no way to properly build it. The issue is mentioning a request from a user trying to build it on a mac (the new ones with the arm chip) and no one yet has succeeded.
Since I am running it on a NAS, options are rather limited.

[bacmanni]

@Pixelartist Did you find a working solution for this? I am trying to get Immich working on my raspberry pi 4 and I am getting basically the same error with version 1.71.0. I found a one tutorial on how to get immich working on pi but that one uses older version. So I am kind of stuck without the info if immich is supposed to be even working on pi.

[raisinbear]

@bacmanni It works on a RaspberryPi, especially 4. You should, however, have the 64 bit version of raspbian if you want to easily deploy with typesense. They don’t have a 32 bit (armv7) build in their official docker repo afaik.

Edit: you can disable search and machine learning and comment the respective sections in the compose file. I see armv7 containers are still built for 1.71.0, so it Immich might still run on 32 bit. Haven’t checked myself though. typesense was an addition sometime this year I think. That would explain why old tutorials wouldn’t mention this particular problem.

[bacmanni]

@raisinbear Ah, so the problems is the 32bit OS. At the time of installing my Raspberry there were some stability issues with it (or at least that what I read) so I decided to with 32bit. Maybe it is time to give that another change. It might be more future proof solution generally. However, thanks for pointing me to a right direction.

[steve-human]

Shouldn't this also be included in https://github.com/immich-app/immich/releases/latest/download/docker-compose.yml? as well as https://github.com/immich-app/immich/releases/latest/download/example.env

As a new user I was having issues installing because TYPESENSE_API_KEY was required, but the documentation doesn't mention anything about typesense

[alextran1502]

This is release 1.52, we are on release 1.100 now and Typesense is no longer used

[steve-human]

Ugh, sorry about that. I must have a very old image on my system. Docker compose pull fixes the issue