2018-01-06, 07:03
Skin Helper Widgets 1.0.38 is here. Check out the change log
Change Log (1.0.38):
Updated Jan. 10
----
Hey everybody, I wanted to let you know the new version of Skin Helper Service Widgets (1.0.37) is in the beta repo. While Marcel has been busy, he's given me permission to continue development and push updates as needed. Since there are a few new features I've been working on, I wanted to post a change log and ask that anyone using the new beta version of Skin Helper Service Widgets with Titan post their feedback / bug reports here.
Change Log (1.0.37):
These changes have been useful for me, and I've found that similar/recommended movies actually provide better results than what Netflix or TheMovieDB offer. However, my library has a huge number of genres (250+) since I scrape them directly from DVD Netflix (here's my Python scraper if you're interested), so any feedback or suggestions on improving the similarity system for a typical TMDb/TVDB-scraped library is welcome.
Change Log (1.0.38):
- Overhauled Similar Media & Recommended Media to take advantage of the experimental recommendation system (setting must be enabled)
- Similar Media gets a recently watched item, then finds all similar movies AND tv shows. For example, if you watch a movie drama, you might see tv dramas in the list now (although they generally show up lower in the list)
- Recommended Media returns a list that contains unwatched movies and tv shows that are most similar to all recently watched movies and tv shows
- Similar Media gets a recently watched item, then finds all similar movies AND tv shows. For example, if you watch a movie drama, you might see tv dramas in the list now (although they generally show up lower in the list)
- New Playlist widget that lets you pick a playlist, then displays it with items sorted by their recommendation score. Uses the item limit for widgets. Available for movies & tvshows
- New Mixed Playlist widget that lets you pick a movie playlist and a tvshow playlist, then displays them together with items sorted by their recommendation score. Uses the item limit for widgets. Available under Mixed Media
- Tweaked similarity and recommendation score calculations
- General improvements to caching for similar, recommended & playlist widgets
Updated Jan. 10
----
Hey everybody, I wanted to let you know the new version of Skin Helper Service Widgets (1.0.37) is in the beta repo. While Marcel has been busy, he's given me permission to continue development and push updates as needed. Since there are a few new features I've been working on, I wanted to post a change log and ask that anyone using the new beta version of Skin Helper Service Widgets with Titan post their feedback / bug reports here.
Change Log (1.0.37):
- Similar widgets on homescreen will only choose from the last x recently watched titles, rather than any watched title. Configurable in settings (default=8). Movies/TV shows widget only
- Added a setting for hiding watched items in similar widgets on homescreen (default=enabled). Movies/TV shows widget only
- Completely changed the way similar movies are calculated. Every movie now gets a "similarity score" (0-100%), with contributions from genres, directors, writers, cast, rating, year, MPAA rating, and movie set
- Also changes the way similar tv shows are calculated. Contributions are from genres, studio, cast, rating, year, and TV rating.
- Adds new behavior for Recommended Movies/Tv Shows that can be turned on in settings. It finds all unwatched titles that are most similar to the last x recently watched titles (uses same setting as above). May run slower, so it's off by default
- Fixes issue where similar movies/shows showed result for wrong title
These changes have been useful for me, and I've found that similar/recommended movies actually provide better results than what Netflix or TheMovieDB offer. However, my library has a huge number of genres (250+) since I scrape them directly from DVD Netflix (here's my Python scraper if you're interested), so any feedback or suggestions on improving the similarity system for a typical TMDb/TVDB-scraped library is welcome.