I decided to flush my metadata to make a clean switch to TMDB.
I went from 650 MB to 300 MB… but as this was the first complete rescan in a very long time it was also the first time I was using TMDB a source fro all movies…
I got several (~10) movies showing with wrong metadata or missing. It turned out to be a mismatch of the year, IMDB and TMDB list different year some these movies, and my files were set to use IMDB year.
This revealed a quit bad surprise, as Infuse couldn’t find a clean match of movie and year, it just took some other random movie in some occasion, and for others the movies were placed under “others” with no metadata.
Wouldn’t it be better if you you create a “section” unrecognized files with a warning to notify users to take action? And if you get a match on the name but the year is off by one, couldn’t you go suggest or auto select that metadata?
But isn’t Other used for non Movies/TV shows, i.e. home/family videos of sort?
I would like to separate files failing in the scrape process vs. files that should not scrape…
I had the movie “LUV.2012.mp4” showing up as completely different movie, then I think it should be better to guess for LUV from 2013… but best would be not to guess and place it in a folder, and make a warning to show look at the folder?
Another example was “Begin Again.2013.mkv” ending up in Other…
Perhaps a warning triangle next to the rescan button? that act as a shortcut to the folder with files failing scraping (note that Other is not very descriptive name).
Yes, this is something I discovered recently that was more extensive (on my library of nearly 3000 movies) than I understood at first; because if you keep your video files in folders with local metadata in the form of .nfo / .xml and cover and fanart .jpgs, movies that don’t immediately fail (if you have mismatched years, for example) and wind up in OTHERS don’t announce they failed and grabbed the data for the wrong movie — how would Infuse know? (… without comparing to ID tags that might already exist in local .nfo or .xml — which Infuse does not seem to do currently.)
What happens in that case is your movie library appears to be correct on import because the movie artwork (pulled from your local source) and title (pulled from your local .nfo) and perhaps even plot (pulled from the same) appear to describe the movie you have exactly …
… except you can’t find it in your collections, because collections are built not from your local metadata but the metadata retrieved from TMDB when the movie was imported — and when the original scan results in a mismatch, that metadata is from a different movie that won’t be part of the collection you are looking for. Same goes with mismatched cast and crew info and thus not being able to find all your expected files when searching your library for films be a specific actor/actress, creative, producer or crew person.
What I have since wound up doing is [a] bulk-renaming all my local .nfo & .jpg files to .nxx and .jxx, [b] removing all library metadata from Infuse (by unchecking all favorites and then deleting metadata cache) and [c] rebuilding the library (re-importing my collection) without my local artwork or metadata getting in the way. That made it easier (though still very tedious!) to identify movie/metadata mismatches by [d] looking for unfamiliar movie titles and/or poster artwork I knew I didn’t have in my collection, and [e] looking for duplicated movie titles and/or poster artwork that would reveal (at least two) different movies that Infuse imported under the actually correct name of (at most) one of them.
Most of the time the issue stemmed from the wrong year issue (per TMDB reckoning) you noted. Next most often was my discovering that Infuse doesn’t approve of middots as seperators. I also found TMDB and maybe also Infuse can’t seem to handle "The"s and "A"s (and their foreign equivalents) out of order … as you might rename a file so it sorts alphabetically by the first significant word instead of its leading common article (but I’ve since found this only applies to filenames — parent folder names are ignored completely, so you can still name folders sensibly as long as all filenames match and are written out in with leading articles only in front). Sometimes movies just show up with different titles in different places and you just need to use the name preferred by the moderators at TMDB.
Colony ,The (2021) ‘‘Tides’’
Formula 51 (2001) ‘‘The 51st State’’
The most troublesome mismatches have been (usually) due to movies with very short titles, most often just one or two words, whose titles and date match exactly with the intended title on TMDB, but are still hijacked by more popular movies whose titles including those words — such as the 2017 movie “Wonder” being hijacked by the 2017 movie “Wonder Woman”, 2002’s “Spider” being hijacked by “Spider-Man”, 2021’s “Together” being hijacked by 2021s “Together Together” …
… or even (prior to the 7.3 update at least) the 2016 movie “Lion” being hijacked by 1994’s “The Lion King” for some bonkers reason — which I’m told is “The Lion King”'s having been premiered on Armenian TV in 2016.
Finally, sometimes movies in my collection happen to fail to other movies on TMDB which are exact matches in both title and year (which really stinks).
I’ve found a workaround today for troublemaking titles in these latter few scenarios.
In such failure to scrape correctly cases I begin testing (via themoviedb.org website) any listed "alternative title"s (not including foreign titles that include non-standard characters) to see if they result in a correct on first-try match; and if so, renaming my files to take advantage.
For sanity sake, my parent folder names keep the common MOVIE TITLE (YEAR) format but I add (in double apostrophe’s after the year) the alternative title of the movie that results in the correct match. This lets me use my batch-renaming scripts to swap out the common movie title in the video and metadata filenames with the alternative title which results in matched searches. The .nfo files (built with Kodi), as well as the scraped titles acquired by Infuse and displayed in lieu of being overridden by local .nfo or .xml, seem to always show the common movie name by default, even if you used an alternative name as the file name (from which the search query was generated).
Some currently working examples of these include:
#Alive (2020) ''Saraitda'' / Saraitda (2020).mkv
Big Blue ,The (1988) ''Le Grand Bleu'' / Le Grand Bleu (1988).mkv
Escape Room (2017) ''60 Minutos Para Morrer'' / 60 Minutos Para Morrer (2017).mkv
Hero (2002) ''Jet Li's Hero'' DC / Jet Li's Hero (2002) DC.mkv
Invasion (2020) ''Vtorzhenie'' / Vtorzhenie (2020).mkv
Love (2011) ''Angels & Airwaves Love'' / Angels & Airwaves Love (2011).mkv
Platform ,The (2019) ''El hoyo'' / El hoyo (2019).mkv
Spider (2002) ''Desafie Sua Mente'' / Desafie Sua Mente (2002).mkv
Wonder (2017) ''Cudo'' / Cudo (2017).mkv
The previously mentioned 2021 movie “Together” unfortunately didn’t have a working alternative title.
So I created one. I called it “Together (vi)” matching IMDB and listed the type as “disambiguation”. It immediately worked and solved the problem. Which makes me 95% certain a TMDB moderator will soon delete it. ¯_(ツ)_/¯
Just wanted you to know how much I’m appreciating your very thorough posts.
I’d normally switch off when posts are that long, but yours are very well detailed and very informative. I’m having the same/similar issues and I’m left with a handful of files that link to the wrong movie - despite being a very accurate title and year - a movie with the same year and usage of that one word ends up stealing the match. It’s pretty annoying.