v2.2.0 #23377
Replies: 26 comments 43 replies
-
|
Awesome! Thank you all for the great update! |
Beta Was this translation helpful? Give feedback.
-
|
Incredible Work! Soo much Features ❤❤❤ |
Beta Was this translation helpful? Give feedback.
-
|
The chat feature is amazing ❤️ |
Beta Was this translation helpful? Give feedback.
-
|
Yeah there is a bug that hide the comment/activity button when clicking on the image from the activity. It is fixed here #23366 |
Beta Was this translation helpful? Give feedback.
-
|
Now if only unregistered guests to shared albums could chat and like photos 😊 |
Beta Was this translation helpful? Give feedback.
-
|
and if only we could have push notifications enabled for those new comments! |
Beta Was this translation helpful? Give feedback.
-
|
and also tag people hahaha, well there are too many features that would be nice 😄 |
Beta Was this translation helpful? Give feedback.
-
|
at that point just make it Matrix-compatible... |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for this release. I would love to see the OCR working, but it does not work for my setup. |
Beta Was this translation helpful? Give feedback.
-
|
Downloading the file outside from docker and mounting it into the container works. |
Beta Was this translation helpful? Give feedback.
-
|
This is how I fixed it for the time being. Three files in three different locations restart your docker container to have it create the folder structure for each of the files |
Beta Was this translation helpful? Give feedback.
-
|
Changelog says |
Beta Was this translation helpful? Give feedback.
-
|
Typo, fixed! Thank you |
Beta Was this translation helpful? Give feedback.
-
|
Whats the exact use-case for this? |
Beta Was this translation helpful? Give feedback.
-
|
Does OCR support characters other than Latin? |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your hard work, now I will wait for the entire library to be scanned) |
Beta Was this translation helpful? Give feedback.
-
|
Thank you so much for this release and your hard work! I have updated the server and client to 2.2.0 and the client is now crashing all the time. Happy to share any logs if helpful. Client: Pixel 9 Pro XL / Immich 2.2.0 |
Beta Was this translation helpful? Give feedback.
-
|
Please help us open a GitHub issue. If it is crashing previously as well, you can track it in this issue #23088 |
Beta Was this translation helpful? Give feedback.
-
|
Thanks Alex! I read through the issue and it exactly describes what I observe. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks! Immich seems to have lost track of the majority of my photos synced app to server and is having a heck of a time trying to back them up (again). It's odd because they are backed up--I can see them accessing Immich via a web page. |
Beta Was this translation helpful? Give feedback.
-
|
There is some context missing here. Did the local database get removed, or was the app reinstalled? The app won't backup already backup assets again unless
|
Beta Was this translation helpful? Give feedback.
-
|
The App was updated after the server was updated, both from 2.1.0 to 2.2.0. I'll check logs I have access to and browse current issues/create a new one when I have time (likely not today). Right now I just have backup disabled haha, might be a tomorrow issue. |
Beta Was this translation helpful? Give feedback.
-
|
Sounds good, please open a GitHub issue so we can easily converse and collect logs there |
Beta Was this translation helpful? Give feedback.
-
|
I deleted the app's SQLite database and re-logged in, which resolved the sync issue. |
Beta Was this translation helpful? Give feedback.
-
|
Same issue for me on the IOS app. I resolved it by reinstalling the app on the phone. |
Beta Was this translation helpful? Give feedback.
-
|
Video seek bar seems not to work here... I can view videos but when I seek the playback stops and I only can play from begin. Android App I tried with file on the server and with file on phone... |
Beta Was this translation helpful? Give feedback.
-
|
Confirmed, on Android, we are looking into it |
Beta Was this translation helpful? Give feedback.
-
|
Does this fix the cuda issues from the last release? I didn't see it mentioned in the change log. Thank you! |
Beta Was this translation helpful? Give feedback.
-
|
Looks like it's failing for both. Main part that sticks out to me: Entire output of one failure. |
Beta Was this translation helpful? Give feedback.
-
|
You are probably using rootless setup or running container under another user. I believe it the same issue as #23393 |
Beta Was this translation helpful? Give feedback.
-
|
Yup... |
Beta Was this translation helpful? Give feedback.
-
|
Try to mount your own volume to |
Beta Was this translation helpful? Give feedback.
-
|
Same issue here, guessing same approach should work for me as well? I am running imagegenius's docker container with CUDA tag. `[Nest] 245 - 10/31/2025, 5:37:25 PM LOG [Microservices:MachineLearningRepository] Machine learning server became healthy (http://127.0.0.1:3003). [Nest] 245 - 10/31/2025, 5:38:46 PM WARN [Microservices:MachineLearningRepository] Machine learning request to "http://127.0.0.1:3003" failed with status 500: Internal Server Error |
Beta Was this translation helpful? Give feedback.
-
|
Wondering if we can use other models for specific languages like smart search models. |
Beta Was this translation helpful? Give feedback.
-
|
It seems like PaddleOCR (which Immich uses) has multilingual finetuning, but you have to use proper models. |
Beta Was this translation helpful? Give feedback.
-
Nice. I wonder if it's easy to replace or if I should I wait for proper management in the UI. |
Beta Was this translation helpful? Give feedback.
-
|
What 'proper management' in the UI are you waiting for? The setting to select the model is at https://my.immich.app/admin/system-settings?isOpen=machine-learning+smart-search. |
Beta Was this translation helpful? Give feedback.
-
Hey, I currently haven't updated to v2.2.0 as it's not pushed yet on truenas community apps, but I assume you mean I can change the OCR model name (like on Smart Search CLIP Model textbox) and it'll grab the specified model and deploy it? That's what I mean. |
Beta Was this translation helpful? Give feedback.
-
Smart Search model and OCR model are different. Where you can type which search model you want to use, available OCR models is only PP-OCRv5_mobile or _server right now. And only default ones, since OCR works with Latin languages for now. |
Beta Was this translation helpful? Give feedback.
-
|
In android app, text does not show in the Search box when you type |
Beta Was this translation helpful? Give feedback.
-
|
Can you record a screencapture and open a github issue so we can track and fix it? |
Beta Was this translation helpful? Give feedback.
-
|
i forced close the app and now it works fine! :) |
Beta Was this translation helpful? Give feedback.
-
|
Will the ability to add an asset to an album via the info sheet on the mobile app be fixed/re-instated in the future? As described in #21935. Adding assets to albums as you view them is a bit of a hassle without it! |
Beta Was this translation helpful? Give feedback.
-
|
What OCR engine is being used here? More details about that? |
Beta Was this translation helpful? Give feedback.
-
|
This one I think https://www.modelscope.cn/models/RapidAI/RapidOCR/ |
Beta Was this translation helpful? Give feedback.
-
|
Love it! I just set my server up a few days ago and already it's getting better! Love it so much that I ordered the retro disc with the server key, the retro immich magnet, and sticker pack ❤️ |
Beta Was this translation helpful? Give feedback.
-
|
Does this fix the crash of the immich during batch uploading (pi5, 4GB ram). The previous versions worked perfectly but with v1.129.0 the system got overloaded and crashed. I tried to limit RAM use, and Concurency but in vain. I found thread on the internet that more users have the similar problem. I read the notes about the release but did not find the exact comment about this fix. Thanks. for reply |
Beta Was this translation helpful? Give feedback.
-
|
Android mobile: seeking videos hangs. After clicking ahead on the seekbar, the video pauses and refuses to play. The check play button just overlays on the video |
Beta Was this translation helpful? Give feedback.
-
|
Updated to immich 2.20 and immich web is now really, really slow. |
Beta Was this translation helpful? Give feedback.
-
|
Do you have the OCR tasks running currently? If so then more than likely your server is just overwhelmed. Check your CPU and memory usage. |
Beta Was this translation helpful? Give feedback.
-
|
OCR-Job is 0 Might be the image in the yaml the problem: |
Beta Was this translation helpful? Give feedback.
-
|
Now I got: |
Beta Was this translation helpful? Give feedback.
-
|
After restarting my TrueNAS it is working again... |
Beta Was this translation helpful? Give feedback.
-
|
2.2.0 upgrade broke my instance (unraid, openvino), issue with symptoms like #22742, mobile app constantly trying to reupload all assets |
Beta Was this translation helpful? Give feedback.
-
|
I got a strange problem with OCR: The same issue with the mobile model. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your work. I have to say that duplicate detection was a promising feature that for my use case I was taking a note of but could not use it (yet) and was waiting for some development such as the direction shown with this release. For context, I have a large external library, which is collected through the years from many sources. It includes, for example, travel albums (= file folders and subfolders) with my OH, so her photos she arranged and my photos I arranged. But there are also quite a few albums (=file folders and subfolders) where she stored my photos or she stored mine. So - duplicates. I also have a number of cases where I organised the file folders so it has the originals in a subfolder (sometimes also RAW images alongside the converted JPEG), and some of them exist in other folders as edited (color correct etc) versions. Again - duplicates. The issue I have with the current duplicate detection is that I can't reliably select which one to delete, because I'd like some consistency: I would like to get to a point where a) I can decide which ones to still keep and stack and b.) if I delete one as obsolete, this will be on a basis of current folder/path, so I am not ending up with half a folder deleted here and half there. At minimum and probably easy to implement as an extension of the additions in this release is to add the path as another row in the table shown (or at least do so when the "Folders" option is switched on for the user, as is for me), as well as to what the source of the asset is (a specific external library or uploaded through Immich directly). Ideally, there could be an advanced version of duplicate detection that provides a folder-level grouping of duplicates (folder A and folder B both have a number of corresponding duplicates, choose your action on a folder level). |
Beta Was this translation helpful? Give feedback.
-
|
Is it just me or has seeking over network become less usable with this version? I feel like I am seeing less updates when scrubbing. Using an iPhone 16 Pro with effictive network speed of 300-400 mbps (measured with iPerf3). |
Beta Was this translation helpful? Give feedback.
-
Is it just me, or is the copy to clipboard button gone?
|
Beta Was this translation helpful? Give feedback.
-
|
Just you. It's only available in secure contexts. |
Beta Was this translation helpful? Give feedback.
-
edit: woops didn't see the reply before mine. Yes I am using https. |
Beta Was this translation helpful? Give feedback.
-
|
Yep, sorry. Just realized that it was the problem, was setting up a new NAS. |
Beta Was this translation helpful? Give feedback.
-
I believe it's an inherent limitation enforced by the browser, not Immich. |
Beta Was this translation helpful? Give feedback.
-
|
Right, just saw that code-server (vscode) also says clipboard doesn't work for insecure content. Thanks everyone! |
Beta Was this translation helpful? Give feedback.
-
|
Is there an option to use openvino to increase OCR detection speed? |
Beta Was this translation helpful? Give feedback.
-
|
OCR is really cool! I'd like a way to see a list of the detected words in an image, to check why some search results appear and some don't. Being able to manually change them would be top too. |
Beta Was this translation helpful? Give feedback.
-
|
We can add that in the future. This is just the first pass of the feature. |
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
v2.2.0
The month following our stable release 🎉 has flown by! We are blown away by the support from everyone around this milestone, especially for the Retro DVD 📀! If you want to get one for yourself, you still can, but once this batch is gone, we won’t be making any more, EVER! So get one while you can from immich.store!
Highlights
Welcome to version
v2.2.0of Immich. This release comes with the traditional Immich fashion, where the features and enhancements list is longer than the bug fixes list. Let’s dive right into the highlights of this exciting releasev2versionsDocker tag for
v2versionsThe
example.envfile has been updated to useIMMICH_VERSION=v2, which is a mutable Docker tag that points to the latest2.x.xrelease image.Optical Character Recognition (OCR)
OCR search has been one of the most requested features on our users' wishlists. Thanks to the fantastic work of @flipped-1121 and @mertalev, Immich can now recognize texts in images, adding a powerful capability to its already powerful search features. This will help users find lost family recipes or a crypto wallet’s passphrase more easily.
The OCR pipeline is integrated into the existing machine learning flow. For new uploads, it will run automatically with the existing machine learning flow. For existing setups, you can go to
Administration > Job > Click on “All” for OCR jobto extract OCR information for your galleryOCR currently has official support for English, Chinese (Simplified), Chinese (Traditional) and Japanese text. Languages using Latin script generally work as well, but may have lower accuracy or interpret characters without the diacritics.
To search for letters in photos, a new radio button has been added to the search type on the web app, and in the dropdown list on the mobile app.
[Web] Wasm implementation for justified layout calculation 🦀
This release comes with a blazing-fast re-implementation of the justified layout on the web, which makes calculating the size of the thumbnails in each bucket several times faster, significantly improving the rendering performance for a large bucket, i.e., 10s of thousands of assets in a single month
[Web] UI improvement for the review duplicates screen
To help you decide which asset to keep in a group of duplicates, more information has been added to the info section below each asset, including date/time, location, and the number of albums the asset is in. The info section has been redesigned to make scanning the information faster, with different properties getting highlighted. We will keep improving on this interface to make it more intuitive
[Mobile] Show albums in the asset’s detail
You can view and navigate to the albums that the asset belongs to in the info sheet
[Mobile] Show similar photos
The popular feature “Show similar photos” has made its way to the mobile app. You can use the button in the detail sheet to find similar images to the one you are viewing.
[Mobile] Chat-style for album activity view
The album’s activity view now displays the messages and actions in a chat-style view, making it easier to follow
[Mobile] High precision seeking for video
The progress bar of the video player now has higher precision duration handling. This makes seeking much more fluid than before, as well as making position updates in the progress bar smooth, rather than jumping from second to second.
[Mobile] New UI to present server-client version mismatch
As we are now in the stable-era, we want to make the notification of a new version more subtle, less distracting. The update message and icon’s color on the mobile app have been redesigned for that purpose.
When your server version is older than the available one, a message with a clickable link is displayed, directing you to the latest release notes. When your mobile app version is older than the one available on the App Store, a message with a clickable link appears, directing you to the App Store to perform the update.
Support Immich
If you find the project helpful, you can support Immich by purchasing a product key at https://buy.immich.app or our merchandise at https://immich.store
What's Changed
🔒 Security
🚀 Features
🌟 Enhancements
🐛 Bug fixes
📚 Documentation
New Contributors
Full Changelog: v2.1.0...v2.2.0
This discussion was created from the release v2.2.0.
Beta Was this translation helpful? Give feedback.
All reactions