A few XMP usage questions
After reading the documentation and latest release notes (
1.58.0), I have a few questions related to usage of XMP sidecars:
(1) Updating date metadata
If I have an existing photo that has an incorrect created_date exif data. It sounds like I can use xmp files to update the date, either by creating a new xmp file within the existing file system and running the DISCOVER job, or by updating the existing xmp file within the existing file system and running the SYNC job. This will then queue metadata refresh.
- In order to update the file structure (since the created_date was updated), will I need to run the STORAGE MIGRATION job or is that part of the queued metadata refresh?
(2) How are xmp files handled during bulk cli upload for duplicates?
Consider the following use cases:
(2a) I have an existing asset foo.jpg with no existing xmp file in immich. If I upload a duplicate file and as part of upload and there is also a new xmp file (<newXmpFile>), how will the xmp files be resolved?
- will <newXmpFile> be uploaded to the existing asset?
- will these assets not be treated as duplicate?
(2b) I have an existing asset bar.jpg with and existing bar.jpg.xmp file in immich. If I upload a duplicate file and as part of upload and there is also a new xmp file (<newXmpFile>), how will bar.jpg.xmp and <newXmpFile> be resolved?
- will <newXmpFile> overwrite the existing bar.jpg.xmp for the existing asset?
- will these assets not be treated as duplicates?5 Replies
I would add @hackerman to this since he is the author of this feature
1. I believe so. The XMP jobs act the same was as the metadata refresh, just with reading / syncing XMP files. If the metadata refresh didn't move the file previously, the behavior here is the same.
2. Afaik, uploading of files is the same as before, just XMP files are ingested along with the media file IF if is successful. There is no duplicate or update handling during the upload process.
I hope this clears it up 🙂
Thanks for responses 🙂
This definitely clears up my question for (1)
For (2)
I'm not familiar with existing logic for duplicates, but my current understanding is that a byte check between the files is done to determine duplicates. Id identified as a duplicate, that upload is ignored.
From above response, the file byte check would not look at the xmp files. - So if an uploaded file is identified as a duplicate, does that mean that associated metadata (including xmp files) would not be extracted? - ie, any xmp files of duplicate uploads are ignored
From above response, the file byte check would not look at the xmp files. - So if an uploaded file is identified as a duplicate, does that mean that associated metadata (including xmp files) would not be extracted? - ie, any xmp files of duplicate uploads are ignored
That sounds right to me. If the media is deemed duplicate, the XMP would be ignored. In order for that file to get paired with the XMP, it would need to be deleted then re-imported into Immich or manually place the XMP into the immich filesystem
Sounds good, thank you for the clarification!