Michael Mayer

Apparently `go-xmp` has a method `xmp.ScanPackets(io.Reader)` which we could try, see https://github.com/trimmer-io/go-xmp/issues/1

> @lastzero > > There seems to be a lot of JPEG talk for being an XMP document. The XMP data is just in another segment? What is it that...

This can now be tested with the `photoprism/photoprism:test` image available on Docker Hub and at a later time with the `photoprism/photoprism:preview` image (before we finally release it).

There's also a server side cache.

This should happen when you change the private flag of a related picture. So maybe you can trigger it with that, even when you have manually manipulated the database.

This is the SHA1 checksum for full images. For face crops, it is extended with information about the area: https://github.com/photoprism/photoprism/tree/develop/internal/crop

@heikomat Does the frontend part look good to you now? Looking at the backend code, especially the SQL joins, I'm afraid this may have performance implications. We intentionally put all...

Since we can't easily merge it given the potential issues and questions outlined above, I'm closing this PR for now. However, we plan to add similar functionality once we have...

Thank you so much for your feedback! We'll keep this in mind, but need to finish the multi-user features first. So no real chance to release it before next year.

Safari 17 added [support for **JPEG XL**](https://developer.apple.com/documentation/safari-release-notes/safari-17-release-notes) ❤️