github imagegenius/docker-immich noml-v1.130.0-ig299

pre-release21 hours ago

ImageGenius Changes:

move to use seperate baseimage

immich Changes:

v1.130.0

Highlights

Welcome to release v1.130.0 of Immich. After almost three weeks of brewing, we are happy to bring you the new version, which is packed with features, performance enhancements, and UX improvements to boost the app's experience; we hope you will like it. Please take a look at some of the highlights below.

  • Folder view in the mobile app
  • Better person naming interface on the web
  • External library scan performance
  • Faster performance and lower memory footprint for the timeline on the web
  • Remember the text search mode
  • Machine learning improvements
  • Jump-to-timeline enhancement

Folder view in the mobile app

You can now use the folder view in the mobile app. It can be accessed from the Library page under the quick links section. We hope this helps users with the browsing experience of a large external library.

Better people naming interface

On the web, the naming interface has been redesigned to allow users to quickly add and edit a person’s name without clicking through several buttons and prompts.

image

External library scan performance

The external library scan code has been rewritten from the ground up. It uses every trick in the book to improve performance, including moving logic to SQL, reducing filesystem calls, and batching asset imports. You can expect a 10x to 100x performance gain when scanning external libraries.

Some numbers:

  • An initial library scan with 19k assets: 1m40s down to 9 seconds
  • An initial library scan with 500k assets now takes 1m58s.
  • An initial library scan with >5M assets now takes 6m50s. This library size was virtually impossible before.
  • Idle rescan with 22k assets, i.e., scanning a library that has not changed: 1m22s now down to below 10 seconds

This change also increases the reliability of library scans. Note that this performance gain only relates to the library scan itself — sidecar discovery, metadata extraction, and thumbnail generation still take the same time.

Possible breaking change: If you use creative exclusion patterns for your libraries, please check if these are still respected and report any issues to us.

Web timeline improvements

The timeline has been “virtually” rewritten to be significantly faster while using much less memory. Scrolling, scrubbing, and resizing the window are all snappier than ever. The web experience on a mobile device has also been improved with more responsive scrubbing and smaller thumbnails to fit more assets on the screen.

ROCm and RKNPU acceleration for machine learning

Machine learning can now be configured to use AMD GPUs and Rockchip NPUs, making Smart Search and Face Detection jobs much faster than on the CPU. See the documentation on how to set it up!

Note: the ROCm image is massive at over 30GiB uncompressed, so please make sure you have enough disk space for it.

Remember the text search option.

The search input now remembers and displays the last choice from the three available text search modes: context, file name, and description.

Jump-to-timeline enhancement

We added a button to quickly locate a photo on the timeline from all the detail viewers (not from the main timeline) in the mobile app.

New search models

Smart search can be even smarter now with a series of new models:

  • ViT-B-16-SigLIP2__webli
  • ViT-B-32-SigLIP2-256__webli
  • ViT-L-16-SigLIP2-256__webli
  • ViT-L-16-SigLIP2-384__webli
  • ViT-L-16-SigLIP2-512__webli
  • ViT-SO400M-14-SigLIP2__webli
  • ViT-SO400M-14-SigLIP2-378__webli
  • ViT-SO400M-16-SigLIP2-256__webli
  • ViT-SO400M-16-SigLIP2-384__webli
  • ViT-SO400M-16-SigLIP2-512__webli
  • ViT-gopt-16-SigLIP2-256__webli
  • ViT-gopt-16-SigLIP2-384__webli

Not sure which model is right for you? We added tables showcasing each model’s estimated memory usage, speed, and quality by language so you can make a more informed decision, along with more detailed steps on how to use them. Check it out here!

Support Immich

If you find the project helpful, you can support Immich by purchasing a product key at https://buy.immich.app or purchase our merchandise at https://immich.store

What's Changed

🚀 Features

🌟 Enhancements

🐛 Bug fixes

📚 Documentation

🌐 Translations

New Contributors

Full Changelog: immich-app/immich@v1.129.0...v1.130.0

Don't miss a new docker-immich release

NewReleases is sending notifications on new releases.