immich-app/immich
1
0
Fork 0
mirror of https://github.com/immich-app/immich.git synced 2026-07-25 14:00:45 +03:00
Code Issues 783 Packages Projects Releases 22 Wiki Activity

[PR #7513] feat(server): optimize person thumbnail generation #11665

New Issue
Closed
opened 2026-02-05 14:43:44 +03:00 by OVERLORD · 0 comments
OVERLORD commented 2026-02-05 14:43:44 +03:00
Owner
Copy Link

Original Pull Request: https://github.com/immich-app/immich/pull/7513

State: closed
Merged: Yes

Description

Person thumbnails are currently generated by cropping to a buffer and passing this buffer to sharp to resize. This is inefficient:

  1. It ends up generating two thumbnails instead of one (the buffer is implicitly a new JPEG file)
  2. libvips is optimized for longer image processing pipelines, so separating the workload here is worse for performance
  3. It lowers quality as person thumbnails are compressed at least three times: once due to the use of the compressed preview image, a second time with the buffer, and a third time with the resize

This PR combines cropping and resizing to the same pipeline for better performance and less compressed thumbnails. Additionally, it uses the original image to generate the thumbnail to further reduce compression artifacts, scaling the bounding box dimensions as needed.

How Has This Been Tested?

I ran facial recognition on all assets and confirmed the person thumbnails are accurately cropped and look normal.

Before:

person-thumbnail-before

After:

person-thumbnail-after

**Original Pull Request:** https://github.com/immich-app/immich/pull/7513 **State:** closed **Merged:** Yes --- ## Description Person thumbnails are currently generated by cropping to a buffer and passing this buffer to sharp to resize. This is inefficient: 1. It ends up generating two thumbnails instead of one (the buffer is implicitly a new JPEG file) 2. libvips is optimized for longer image processing pipelines, so separating the workload here is worse for performance 3. It lowers quality as person thumbnails are compressed at least three times: once due to the use of the compressed preview image, a second time with the buffer, and a third time with the resize This PR combines cropping and resizing to the same pipeline for better performance and less compressed thumbnails. Additionally, it uses the original image to generate the thumbnail to further reduce compression artifacts, scaling the bounding box dimensions as needed. ### How Has This Been Tested? I ran facial recognition on all assets and confirmed the person thumbnails are accurately cropped and look normal. Before: ![person-thumbnail-before](https://github.com/immich-app/immich/assets/101130780/2e9f4bc3-89f2-499c-aeea-bfe2887b5937) After: ![person-thumbnail-after](https://github.com/immich-app/immich/assets/101130780/3d536801-5b12-4038-abb2-9641a31c5fd8)
OVERLORD added the pull-request label 2026-02-05 14:43:44 +03:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
accessibility
changelog:enhancement
changelog:security
changelog:skip
changelog:translation
cli
date-time
dependencies
documentation
external-library
format
good first issue
mobile-beta
mobile-beta
mobile-beta
needs-answer
nice to have
pull-request
Mirrored from GitHub Pull Request
 sharing
tech-debt
📱mobile
🖥️web
🗄️server
🧠machine-learning
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
OVERLORD
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: immich-app/immich#11665
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?