Emby Keeps Rewriting Wrong Movie Title

[rod626 3]

I have a bunch of movies with filenames like 0001.mp4-1200.mp4.  When I set up the original collection, I forgot to disable the metadata lookup, and Emby "discovered" hundreds of incorrect movie titles.  For example, 0004.mp4 was detected as "Fantastic Four", 0002.mp4 was detected as "Terminator 2", and so on.  

I have disabled all metadata lookup on these collections, deleted the offending *.NFO files, and refreshed all metadata.  However, whenever I refresh, Emby just creates a new *.NFO file and dumps the information from the Emby database back into the NFO file.  If I disable the *.NFO writer, Emby does not dump the incorrect information back to the NFO file, but it keeps the wrong title in the Emby database.  It appears to be impossible to force Emby to forget the incorrect Movie titles.

Is there a "right way" to clear these?  I am considering two workarounds:

A. Write a program that scans for any movie with a discovered IMDB ID.  Rename it to something, refresh the library, then rename it back and refresh the library.  This is a pain, because I will lose all my favorites and history.
B. Write a program that creates a dummy *.NFO with fake titles and then refresh metadata.  I have verified that Emby will copy the information from my NFO into Emmy's database.  This is a pain, because Emby stores a bunch of other stuff like "original title", "sort title", etc. that are totally wrong and useless.  I have no way to know what things I need to replace to ensure Emby's database is clean.

[Happy2Play 9783]

13 minutes ago, rod626 said:

Is there a "right way" to clear these?

Once in the database all you can do is remove them from the library, do library scan then readd them.

But may want to support this request.

https://emby.media/community/index.php?/topic/75868-a-plex-like-unmatch-button/

But yes with items like those you need to have nfo files with media as auto detection will always be wrong.

How do you know what they are with naming scheme like that?

[rod626 3]

18 minutes ago, Happy2Play said:

But yes with items like those you need to have nfo files with media as auto detection will always be wrong.

How do you know what they are with naming scheme like that?

Thanks, I have also verified that the *.NFO can be used to fix the titles, but cannot replace the tagline, community ratings, or IMDB/TMDB IDs.

I don't actually care what the titles are (except of course, it shouldn't be called "Iron Man" or whatever), because I organize everything by collections, playlists, favorites, and tags.  I have some automated processes to tag movies like "outdoors" or "Boston" and so on.  I like the way that Emby keeps the metadata and BIF files in the same folder as the media, so it feels safe to put in the effort to curate the metadata.  However, it's frustrating that things like favorites and playlists are only available in the Emby database.  When doing work to organize the collections, it's not clear which work is backed up with the files and which work is subject to the whims of a spurious movie detection.
 

Edited September 25, 2022 by rod626
Clarification

[Happy2Play 9783]

5 minutes ago, rod626 said:

However, it's frustrating that things like favorites and playlists are only available in the Emby database.

Yes as userdata is tracked by providerid and when no providerid it is tracked by custom databaseid that is lost when items move.

As for Playlists they have their own xml with path that has to be updated if you move media.

Playlist – Manual Migration : (emby.media)

 

[rod626 3]

29 minutes ago, Happy2Play said:

As for Playlists they have their own xml with path that has to be updated if you move media.

Oh, that's wonderful!  Thank you!

[rod626 3]

I managed to fix these by copying the media files and deleting the old copy.  Renaming doesn't work, because Emby is smart enough to detect file renames and rename the file in the database.  Thanks for all the help!

[Luke 42083]

 An unmatch feature is coming in Emby Server 4.8:

 

[rbjtech 5284]

14 hours ago, Luke said:

 An unmatch feature is coming in Emby Server 4.8:

 

Which in it's current guise - will not solve the OP's request.  As when they refresh, all the incorrect Identifications will simply come back ...

[Luke 42083]

5 hours ago, rbjtech said:

Which in it's current guise - will not solve the OP's request.  As when they refresh, all the incorrect Identifications will simply come back ...

There’s no guarantee of that, but if it does then using identify would be a better option.