Some feedback on the new TV Agent (build plexmediaserver-1.21.2.3925-f189f701c.x86_64.rpm):
For context my TV library size is 366 shows with 37k episodes. Creation of library and scan took approx 80 minutes to complete, downloading all metadata. Second scan (refresh) completed in approx 40 minutes. This is slower than previous builds I tested but still far faster than the existing agent. I guess the additional scanning for local files accounts for the additional time taken. But again, its super fast so don’t want to focus on that.
Almost everything was matched correct with a couple of exceptions:
The Haunting where I ended up with duplicates due to having The Haunting of Hill House and The Haunting of Bly Manor separated into different directories. This won’t effect most folks.
Voltron: Defender of the Universe and Voltron: Legendary Defender ended up merged together and I had to split.
Andromeda S01 (themoviedb.org/tv/2985-andromeda) was not picked up and I needed to fix match. The Andromeda Strain was detected incorrectly. What is odd is that S02-S05 was found correctly.
Ronja The Robbers Daughter (themoviedb.org/tv/62495) was not picked up and I needed to fix match. Ronja Robbersdaughter (1984) was detected incorrectly.
3% (themoviedb.org/tv/68467-3) was not picked up and I needed to fix match. Teen Bata Teen (1995) was detected incorrectly.
Some shows seem to use the wrong title. Star Trek: Enterprise (themoviedb.org/tv/314-star-trek-enterprise) is now called Enterprise as one example. I guess its using the original title but I have this not selected in the advanced settings on the library. Other examples, Stargate Universe is now SGU Stargate Universe. Stephen Kings It is now It. Terry Prachett’s mini series have Terry Prachett removed. Your Lie in April now Shigatsu wa kimi no uso. The promised neverland is now Yakusoku no Neverland. Terror in Resonance is now Zankyou No Terror.
Matching on TVDB id is not working when doing a fix match. Found this issue when attempting to fix match Voltron: Defender of the Universe with a search for 78646.
Directed by metadata at the episodes level has been downloaded fixing this very long running bug, very happy to see this.
Season Titles appears to work well for all that I checked. Babylon 5, 24, Legend of Korra, Mr Robot, Torchwood, American Horror Story to name but a few I checked, no issues I could see so far.
Season level summaries also now downloads and again looks good from what I can see.
I have local show, season posters for all my content, all was detected correctly.
For the shows I have external and internal subtitles they were detected correctly.
All my posters and subtitles were successfully added on plex, the only local media assets that the scanner ignores are the extras.
I also saw that sometimes the cast (from theMovieDb) is completed with info from IMDb (I think). Those actors sometimes already exist on theMovieDb but haven’t been credited to that show, would it be possible to get those images too? An example is Shane Berengue he has been added to WandaVision on IMDb, but not on theMovieDb and the IMDb id on theMovieDb has also been filled in so I think it could be possible to implement this.
If you can get information from IMDb could it also be possible to get the directors and writers from IMDb (& maybe also the guest & co stars)?
I have 2 shows that I have in absolute order and named like “show name - E001 - optional stuff.mkv” and they are not picked up at all. I tried putting them in a season folder since I didn’t had it before but that didn’t help at all. The shows are firefly and Star Wars clone wars.
Thanks! I’ll go through this today and try to repro the mismatches.
Matching with an id in Fix Match isn’t yet implemented for the new agent.
The weird naming is strange I have SGU in my library and it’s just named “Stargate Universe”, what country and language settings do you have selected again?
Language would be set to english with certification country set to Ireland using TMDB as ratings source. I attached the logs also in case that might help with any of my feedback.
It’s on the list of known issues and missing features with the preview.
For things that are not matching at all could you give some examples and exactly how the files are name on disk. Or if you could just provide your logs after doing a metadata refresh on the library.
Hmm, mine is working fine. Are you using the bundled web app?
For that specific example I don’t see a German version of the poster so it’s just using the default IMDB poster - if you have some more examples check if the same with those otherwise let me know.
Thanks, I see that’s what IMDB says the title is for British English (and a few others). You could work around this by editing the show and setting the metadata language to English (US) for those specific shows (or use original titles)
Next test I’ll do will likely just set to English (United States) and see what the results are like. If its better than English (United Kingdom) I’ll just go with that. At least until English (Ireland) is implemented
Actually ignore my previous comment. There is a bug somewhere with using the tvdb ids in the filename. I just happened to be working on a branch that has that resilience I was talking about and it did fail but then self corrected afterwards without me seeing.
Okay, so I was mistaken before. We do support tmdb/tvdb ids but there is a bug that was discovered now where we were only looking for the prefix thetvdb and not tvdb which was causing matches to fail - the reason why most of your stuff wasn’t matching.
It’s getting fixed now and will let you know when that is deployed.