Imposing a strict folder structure on the user is in my mind just plain insanity.
As a feature addition sure, it’s cool to have that kind of metadata organization but it shouldn’t be imposed to the point where I simply cannot access media if it doesn’t fit some kind of arbitrary naming convention.
Just allow us to turn it on or off, or at least allow the user to collate miscellaneous media with simple filenames and a placeholder image.
It’s a machine (program) it can’t recognize the files you add unless there is some kind of structure to it, even face recognition program do require some point of comparison to a face to start with and this is not AI just dumb ifs or elses.
Adding to Ismelda’s comment: you can already do that by using the Other video library type and the personal media agent.
Plex needs the structure to match your media – if you can live without the matching / automatic metadata… or if you’re ok with your own embedded metadata, use the other video type.
Other than that there’s already a number of existing suggestions asking for more flexibility in particular with tv-show content – down to specifying your own regular expressions in order to determine the different information required to match your content.
While those requests pop up occasionally, none of them has gained much attention/support.
@Ismelda, it doesn’t need any structure that is client-facing which should affect the organization of files. What you’re describing is the job of inodes inside a file system.
Facial recog programs like OpenCV use ready made algorithms, some of which require tagged data. Other, more modern, ML programs/algos use unsupervised learning which do not require any kind of pre-defined pattern or tagging.
Ok awesome thanks @tom80H, “other video” option seems to work just fine I’ll use that from now on. Should really be highlighted in the docs as a first option in big lights to avoid this issue coming up again and again, I’ve been using plex for a while now and only just discovered this option yet as you say there are a number of threads. Clearly the docs need to highlight this feature.
Really tho, this shouldn’t be an issue at all. I’m a senior software engineer so I don’t want to hear anymore nonsense about needing to pre-organize to some kind of special folder structure to organize the data. Recursively scanning for metadata isn’t rocket science nor is it ML and shouldn’t need some special folder labelling, certainly not one with spaces and parentheses. At the worst case, I would say each actual media file needs it’s own folder but really, even that would be a very very basic implementation.
There’s actually a feature suggestion to allow some kind of fingerprint / soundprint / … media recognition. The key challenge with that is there’s no reference for those. Some argue that Plex is in a particular good position to use their existing matches to establish such a reference… I can only guess this could be a thing in the future. Compared to audio/music recognition there’s just a whole load of additional complexity (i.e. different translations, cuts…) that need to be covered; so while you can recognize a music track based on a few seconds and a handful of media characteristics, doing the same for a movie will need significantly more power – and I’m not sure users will appreciate if Plex starts scanning/processing Terabytes of their data to get an accurate match.
@medicineman2500 , I am with you. The way this works is really not how it should work in 2021.
When users set up a new library, PMS asks the users to “please help us” by following our naming and organizing guide.
No. PMS should help the user to make this step quick and easy.
I am still in the process of making a large video collection Plex-compatible.
It is laborious, which is a big downside in its own right and a high entry level to Plex.
The method is so error-prone (both technical and user), that the forum is full of support requests.
And the result is not even great, because even a “perfectly” organized media collection – by Plex’ definition – still has significant shortcomings.
On the one hand, the file management must be followed meticulously.
On the other hand, that file layout is often not helpful for use outside Plex, e. g. via DLNA.
The list could go on.
As for myself, I would happily donate fingerprints (see here) from my own collection. And then everybody else who has the same movie with the same fingerprint, will enjoy magic recognition – no matter where in the filesystem it sits and no matter how it is named.
If enough people do the same – and I expect there’s a large part of similarly cooperative people in the community –, then building a reference database, as @tom80H mentioned, will be quick and painless.
A reference database for metadata and subtitle collection would be fantastic.
I’ve never worked with this kind of search algorithm before but my first guess would be a simple binary search like algorithm that takes samples to store at normal intervals (each boundary) and then when matching data, runs against samples of library content at known timestamps until it finds the sample as a common binary search would.
This would become a trade off between store size and accuracy of course. Say we take 8 samples of 8 bytes per piece of media, that’s already 64 bytes each. I’m sure there are engineers with greater experience in sampling and collating huge media datasets, all I know is that this is a well-studied problem. Whole industries have formed around the solutions offered.
Apart from that, I’ll just second what has already been said. All organizational complexity should of course be abstracted away into the database.
There are other considerations like seeding torrents, I don’t want to organize my data coz I use magnet links. I simply want to read from a dump, which is arguably terrible practice but I honestly cbf.
2021 clean-up:
Other video library type option is available