Posts: 24
Joined: Feb 2009
Reputation:
0
I see that in XBMC 9.04 the new default scraper is themoviedb.org. Anyone already using 9.04 from SVN that would like to share their experience of using themoviedb.org?
I'm still scraping using IMDB, and the data retrieved is mostly OK except for the movie synopsis, which can vary greatly in quality (basically depending on the grasp of the English language of the user posting their synopsis on imdb.com). Sometimes it's laughably bad.
For this reason I'm really looking forward to switching to themovidedb.org, so I'd like to know how the switch will go when it happens. Would I just rescan my entire library? Movie posters and fan art still working OK? Happy with it?
Posts: 3,660
Joined: Feb 2008
Reputation:
93
Jeroen
Team-Kodi Member
Posts: 3,660
Image quality is much much better in general on tmdb. Plot descriptions are much more to the point and not enormous elaborate narrations like with iMDB.
Occassionally there's no movie information available for a certain movie, but that's no wonder being relatively new and all. I do find that for me tmdb is more accurate in picking the right movie. Unless you have a very particular preference for movies you should be fine.
And when you do come across a video not being available in the database take a couple of minutes to add it yourself.
One thing that bugs me about tmdb though is the sometimes ridiculously long genre information. IMO a movies should be tagged with three genres max, but some go way over that.
Some kind of hybrid scraper would be awesome though.
Posts: 24
Joined: Feb 2009
Reputation:
0
Jeroen, many thanks for the info.
So, I installed 9.04 from an SVN build today and rescanned my library using themoviedb.org. Overall, I'm quite happy though in my experience tmdb has been LESS accurate in picking the movie. For example, it somehow managed to choose The Dark Knight from a folder (and movie file) titled 'Batman Begins (2005)'. There were numerous other examples also (though only a small percentage of the total) ... definitely more mismatches than the IMDB scan in any event.
Another gripe regards the tmdb movie user rating. The ratings on tmdb are obviously not as mature as on IMDB (not many people having voted). As both you and migueld have said, a solution to this would be some kind of a hybrid scanner that will allow me to pluck the rating from IMDB.
Nevertheless, I'd say overall, given the higher quality artwork and plot synopsis, this is a great improvement over IMDB.
PS. I've registered on themoviedb.org and have already started adding missing data.
Posts: 26,215
Joined: Oct 2003
Reputation:
187
If you do a manual refresh on the movie that came back as the wrong one, is the correct title available for choice?
If so, please identify the exact filename/foldername used for the lookup, as the fuzzy matching I added should be taking care of that for you.
Cheers,
Jonathan
Posts: 24
Joined: Feb 2009
Reputation:
0
2009-03-26, 09:45
(This post was last modified: 2009-03-26, 09:56 by beforeseven.)
Jonathan, thanks for taking the time to look into this. Each time I do a manual refresh, the correct title is indeed available. Where the scraper has chosen the wrong title, I think the correct title has almost always been the second choice on the list. These are the results from a few I manually refreshed just now:
Filename: Batman Begins (2005)/Batman Begins (2005).avi
Results:
1. The Dark Knight
2. Batman Begins
Filename: Before Sunset (2004)/Before Sunset (2004).avi
Results:
1. Sunset Blvd.
2. Before Sunset
3. etc. [...]
Filename: Catch Me If You Can (2002)/Catch Me If You Can (2002).avi
Results:
1. Catch 'em if you can
2. Catch Me If You Can
Then there were a few cases when the scraper chose a sequel, instead of the original:
Filename: Die Hard (1988)/Die Hard (1988).avi
Results:
1. Live Free or Die Hard
2. Die Hard
3. etc. [...]
Filename: Back to the Future (1985)/Back to the Future (1985).avi
Results:
1. Back to the Future Part II
2. Back to the Future
3. etc. [...]
Hope this helps,
Rob
Posts: 26,215
Joined: Oct 2003
Reputation:
187
2009-03-26, 10:32
(This post was last modified: 2009-03-26, 10:52 by jmarshall.)
And "from an SVN build" means what exactly? As that's the exact problem I fixed about a month ago. A debug log would tell you for sure.
EDIT: Grrr - found the problem - sorting by relevance was commented out. No doubt by me whilst testing :p
Fixed in r18946.
Cheers,
Jonathan
Posts: 12,706
Joined: Nov 2003
Reputation:
129
spiff
Team-Kodi Member
Posts: 12,706
we use the title returned by the api. there is no issue. if you want the long titles, ask the tmdb guys
Posts: 24
Joined: Feb 2009
Reputation:
0
2009-03-26, 11:45
(This post was last modified: 2009-03-26, 11:47 by beforeseven.)
I meant "issue" only in the sense of what I had just mentioned; not in the bug-track sense. However, as a software developer of sorts myself (web apps), I would probably would have taken more of an interest had some alerted me to the fact that there were problems in the data I'm pulling in.
I've signed up for an API key for TMDb to check it out.
Posts: 26,215
Joined: Oct 2003
Reputation:
187
Let us know if there's a better way to do things. We have an "alternate title" field to fill in I think already (not sure if it's in-use though).