Stop. Changing. My. Posters

General Discussions
148

Thanks so much for your help, you put me on the right path.

Since python was installed using Homebrew, I actually had to create a virtual environment using “venv” to be able to install plexapi. Otherwise, I was always getting an error that said something like “This environment is externally managed”.

Just in case other people are in my situation, here’s what I had to do to make it work:

  1. Using a newly open Terminal window, python3 -m venv .venv to create the virtual environment.
  2. source .venv/bin/activate to activate the virtual environment in the current Terminal window.
  3. python3 -m pip install plexapi to install plexapi in the virtual environment.

After all that the script began working fine (after modifying it where indicated to specify my currently used Plex token, of course).

Just a note that to be able to properly run the script again after having closed the previously used Terminal window, source .venv/bin/activate needs to be run again in a new Terminal window to be able the make the script work again (to reactivate the virtual environment, basically).

Thanks again for all your help!

149

Where are we with the option of locking posters that we have curated for well over a decade, forever?

Why is locking a poster, forever, until user action such a foreign concept to plex,inc? Why are we “blessed” with posters that magically change to posters that are worse than netflix’s autogenerated posters?

150

It’s a little weird. Was just talking to a friend and fellow Plex Pass lifer who is reporting the poster changes routinely happen to him, and – as I mentioned upthread – they always get changed to bizarre and completely incompetent posters no one would use. I get that staff are hoping for some silver bullet replication but it seems like it’s a thread with 142 posts and a really consistent refrain from dozens of users explaining it happens to them and all the employees are just replying “Hmm, doesn’t seem like this is replicable”. I can pretty much assure you people aren’t making up freaky AI misfire posters that they don’t want in their library.

I was driven to reply again to this thread because I saw another thread in the Metadata section where someone was pointing out how low quality the metadata in the current plex scraper is at the same time that plex popped up a thing telling me to upgrade my libraries to the new scraper, which would almost certainly expose me to both the issue in that other thread and the issue in this thread. What is going on? Is there a business case for badly running a metadata service instead of using existing and working data sources?

2 Likes
151

If only some dev at Plex would change all covers of a test library to images just showing numbers and ask for people’s settings and then wait a month or two (doing daily checks and log saving). In a library consisting of a couple of thousand movies or TV shows, they should be able to replicate the problem in order to hunt down the reasons.

Obviously, it’s not that kind of problem that you find while looking at your code - nor by letting people describe their problems.

Of course, this would be time-consuming and maybe other problems are easier targets.

152

A silver bullet would be nice, but literally any clue would be useful. Nothing reported so far has led to us finding anything.

I’ve been testing this myself on my own server and all my posters I set manually for this test (starting probably 2 years ago) have not changed from the ones I’ve set. I’ve also not received any internal reports of this happening to anyone on our team.

My hunch is that 99% of reports are due to the people just not being able to tell whether a poster is locked or not, and we have an internal issue assigned for somebody to implement this in a future update.

Then there is the 1% I’m leaving open to the possibility there is some obscure bug that is really difficult and rare to reproduce.

Once everyone can see if their artwork is locked or not it will be much easier to sift through the reports for any legitimate issue.

2 Likes
153

i have had several posters that i specifically set myself change to another poster some time down the line. i have had this happen on a long standing plex server and on a brand new plex server.

victim blaming much?
surely a user manually selecting a poster would “lock” it regardless if you can see if it’s locked or not.

it’s not user reproduceable, hence why “rare” to reproduce.
also, it’s likely the user won’t notice the change until a few days later, by which time any logs (if there were any in the first place), will be hard to find (even if you knew what to find in logs).

154

My personal hunch is they were posters not changed. I spent weeks modifying my large library only to find hundreds changed months later. That said, I know I probably only manually set 30-40% of them, becasue if I like the poster selected I skipped it.

That or a specific release unlocked this field at some point and that is why people who set their library months or years ago all of a sudden had this start happening to them.

155

There is definitely a difference in ‘how’ you manually select a poster that affects whether or not Plex decides to randomly replace them.
If you select a poster from posters that plex offers you from a selection imported with metadata… plex seems to feel it has free reign to replace it.
If you select a poster by either importing from a .jpg or by stating a URL… they seem safe.

If that’s what is meant by locked vs unlocked… then sure, only unlocked posters are being replaced… but isn’t the act of selecting it supposed to lock it, regardless of where the poster is being selected from? Otherwise… what’s the point of selecting it in the first place?

156

Yes, if you change it from the default it will lock. If it’s not locked then a poster can change during a metadata refresh if the poster changes upstream.

157

clearly that “lock” isn’t working then.

yet for the ones i have had changed the original poster was still there in the selection, so the poster hasn’t been changed or removed.

158

If you can demonstrate this behaviour then I can look into any issue.

159

how can i demonstrate something that isn’t user reproduceable?

however see my post from 15 days ago that shows to instances where the poster was changed, and the previous poster that was selected is still on the list of available posters.

plex had replaced the poster at an unknown date, i went back in and re-selected the poster that i had used previously. that example was on a brand new plex server where i had specifically set the posters on both those shows, only to find that some point later plex had decided to change them to another poster (which i assume was a new addition to the list of available posters).

160

What I would recommend is that download a copy of your database and store it somewhere to act a snapshot.

Then use your server normally and if you come across the problem again, immediately grab the server logs and a fresh copy of the database. Then send me both databases and the logs and I can see if I can find something.

I cannot reliably fix an issue that nobody can give me any data supporting that an issue exists. As soon as I have some evidence pointing to an issue I can investigate further.

161

It sounds like you are not believing that there’s a problem at all.

It is obvious that in any large real-life library, people don’t take notes when changing posters and HOW they changed posters (taking another existing one or uploading a new one).

This way, you will probably never get a detailed “proof”.

Here’s how you can still test this without leaning back blaming users of not providing enough evidence:

a) take one small sample movie in a folder. Create one black with a white dot in centre named poster.jpg (make sure it’s something easily identifiable).
b) Create two lists of - let’s say IMDB’s Top 250 movies (using the correct Plex naming) in a text file - one in English one in another language Plex supports (German, for example).
c) create a five-line script to copy that file/folder 250-times and rename both copied file and folder using the English list. Move those 250 movies into “library base folder 1”.
d) do c) with the German list, moving the 250 movies into “library base folder 2”.
e) change both scripts to add the same poster.jpg to all newly copied/renamed folders. and execute both scripts again, this time using “library base folder 3” and “library base folder 4”.
f) Copy “library base folder 1” and “… 2” to "library base folders 5 and 6…

Now you should hav 1500 movies in six base folders (6 times 250), half of them with English names, half of them with “Language X” titles - one third of them using a local cover art asset, two thirds of them not.

f) Create a new PMS with standard settings,
g) create six new movie libraries using standard settings, three of them using English language, three of them using that foreign language you chose. Make sure that you prefer local assets and add one of the four library base folders to each of the four libraries.
h) Make sure that Plex gets the identification right in all six libraries and check if all local cover art assets are being correctly used.

Now, in the first two libraries that do not have local cover art already applied, add the same Cover file via each movie’s Edit Metadata UI.

At the end of this process, all movies should have the same image as Cover art in four libraries - applied differently.

Now go to libraries number five and six - again make sure that all movies are correctly identified by Plex… and change the poster for all 500 movies to something which already exists - let’s say the third available image. Take comparision screen shots for all two libraries.

Now, save the LOGS…

Every morning, quickly check the first four libraries for changed posters and take logs.
Every week, check library five and six for changes using the comparison screenshots.
Identify the problem of the next couple of months…

You could check for

  • larger numbers of test movies
  • different library languages
  • different applications and sources of posters

Want to make it more complex (to find patterns)? Then do not use the same cover, but covers having only numbers and perhaps different file formats.

Too much effort for you?
Anybody else volunteering?
Somebody else from Plex staff? Users should only be asked to take a more “USER” approach of helping you finding the problem.

But of course, this can only be done if somebody would really be interested to find the real cause.

162

It has nothing to do with believing, I just haven’t received anything but anecdotal evidence yet. I believe there is a real possibility of a bug, but without being able to see it happening I don’t know where to start looking.

Thanks for taking the time providing this, I can ask one of our test engineers to set up a test library for this.

164

i guess we are all imagining it then?
typical plex “it’s all ok here, bruv” response.

i’m sure the business priority remains on building out the plex social network than maintaining the core product. but i guess that’s what happens when you take investment from data mining companies.

1 Like
165

I’ve let you know what you can do to help me, I unfortunately cannot do more than that for now.

I have to prioritise my time and chasing down rare, elusive bugs with no way to reproduce them is extremely time consuming.

166

this would only be of any benefit it you noticed the change immediately, and that would only be case the if you are sitting watching the plex library 24/7 for any changes that might happen at a random interval.

in my case i only noticed it whilst i was browsing thru the library, so the change could have happens many days prior.

yes it would expect it is time consuming for you, but it would be equally time consuming for the user - which do you think it is more reasonable to ask to do that? it’s not the user!

I would suggest you do what the previous poster suggested and set up an insanely overkill situation and then sit at it for hours on end so you can see it happen in real time.

167

Well… even without being able to manually recreate the issue… you can still investigate how the lock works.
What we are seeing is that if we select posters that Plex has downloaded and offers to us to select… selecting those is not counting as locked.
If we select an image from our HD or from a URL… they are locked.

Shouldn’t be that hard to examine how the lock is applied in these two scenarios… and thus spot the difference in how one scenario works and the other does not.

168

i assume you’re directing that at the plex employee at not me.
how i can i investigate how the lock works when any such lock is not visible to the user.

i never use that method so would have no idea how it works.