movie scraping, issues with tmdb
#1
Had to format my drive this weekend, so today scraping my movie drive with kodi 14.2.

Going through the poster art I noticed that if artwork on tmdb website is listed as 'primary' then the scraper defaults to the 2nd highest voted poster art. If nothing is listed on tmdb site as primary, then it scrapes the highest voted poster artwork like it always has.

This a site? Or plugin issue?

Example
https://www.themoviedb.org/movie/77866-c...nd/posters

When scraped, it defaults to the 2nd poster listed.

Not a huge deal, but often I find the 2nd rated posters are usually bad, or just not 2:3 ratio.
System 1: HK1 RBox - Android 11, S905X4 Quad-Core, 4GB DDR3 - CoreElec KODI 20.3
System 2: Beelink GT1 Ultimate - Android 7.1, S912 Octa-Core 1.5 GHz, 3GB DDR4 - CoreElec KODI 18
Storage: WD MyCloud EX2 Ultra NAS - 2x4TB WD Blue (WD40EZRZ)
Display: 55" TCL 55R615-CA 4K TV
Reply
#2
What does "Primary" mean in this context? Does that mean someone from TMDB has specifically chosen a poster that has its own preference beyond just using votes?
Reply
#3
Must be something new on the site, never noticed it before, but if you go to the link in my OP, you'll see that the #1 poster says primary and has a green check mark. Then go look at other movies, some say primary, some don't.
System 1: HK1 RBox - Android 11, S905X4 Quad-Core, 4GB DDR3 - CoreElec KODI 20.3
System 2: Beelink GT1 Ultimate - Android 7.1, S912 Octa-Core 1.5 GHz, 3GB DDR4 - CoreElec KODI 18
Storage: WD MyCloud EX2 Ultra NAS - 2x4TB WD Blue (WD40EZRZ)
Display: 55" TCL 55R615-CA 4K TV
Reply
#4
So, anyone else have this issue?

I was clicking through my collection updating manually, until I realized how many posters were the wrong ratio/size and how long it was taking. Normally if I have to rescrape the collection (1,100 files more or less) I have about 15 to change... way way more than that this time.
System 1: HK1 RBox - Android 11, S905X4 Quad-Core, 4GB DDR3 - CoreElec KODI 20.3
System 2: Beelink GT1 Ultimate - Android 7.1, S912 Octa-Core 1.5 GHz, 3GB DDR4 - CoreElec KODI 18
Storage: WD MyCloud EX2 Ultra NAS - 2x4TB WD Blue (WD40EZRZ)
Display: 55" TCL 55R615-CA 4K TV
Reply
#5
Not sure. However, I'll move this to the General OS forum to get more attention.
Reply
#6
Interesting...never saw that 'primary' thing on that site and noticed, that it is not there for any language for this movie. In german there is no 'primary' note
Reply
#7
I wiped out KODI and installed XBMC 13.2 Gotham, did a scrape again and same deal... movies that have that PRIMARY flag for the main poster art, that poster doesn't get scraped.

It shows as the 2nd poster in the list when you manually change the artwork though, so not skipped entirely... just not scraped as the default.

So, problem isn't XBMC/KODI version specific... I guess scraper/plugin issue?



Image
System 1: HK1 RBox - Android 11, S905X4 Quad-Core, 4GB DDR3 - CoreElec KODI 20.3
System 2: Beelink GT1 Ultimate - Android 7.1, S912 Octa-Core 1.5 GHz, 3GB DDR4 - CoreElec KODI 18
Storage: WD MyCloud EX2 Ultra NAS - 2x4TB WD Blue (WD40EZRZ)
Display: 55" TCL 55R615-CA 4K TV
Reply
#8
I think this is a new feature of this site and maybe not documented in their API documentation. So I think, this won't ever be fixed for Kodi Helix, because there have to be done some changes to the scraper. Maybe it will be fixed in Isengard...

What does the devs say?

Maybe it would be better to open a ticket for that. It's not a bug if TMDB changes something which Team Kodi can't know off...but I guess it will get much more attention if it would be reported at trac
Reply
#9
@THRobinson - just to confirm your findings - I am experiencing the same behaviour. Strange thing testing with a relatively small sample, this is the only example I can find with that annotation, which causes this behaviour. WIll do some more testing.

If you decide to take David1977's advice and report to TRAC as a bug, post the link here.

EDIT - of the small sample I was able to identify that had the 'primary' notation - in addition to Contraband (Schindler's List and all Star Wars movies) only Contraband showed this behaviour, all others scraped correctly, using the voting system as expected.
Reply
#10
Well, I know there are a few movies that have it and scraped incorrectly, like Avatar and Boogie Nights... top of my head. I'll have to click through the library later and find some more. It seems random... not based on year/popularity of the movie for what has that 'primary' designation.
System 1: HK1 RBox - Android 11, S905X4 Quad-Core, 4GB DDR3 - CoreElec KODI 20.3
System 2: Beelink GT1 Ultimate - Android 7.1, S912 Octa-Core 1.5 GHz, 3GB DDR4 - CoreElec KODI 18
Storage: WD MyCloud EX2 Ultra NAS - 2x4TB WD Blue (WD40EZRZ)
Display: 55" TCL 55R615-CA 4K TV
Reply
#11
Yip it did it for Avatar as well - don't have Boogie Nights so can't check that one - seems really random.
Reply
#12
Ya it's weird... I do a lot of adding artwork and voting on that TMDB site, so, when I see a poster in my collection the wrong shape or just crappy, I always login and vote... that's the only reason I noticed it. Power outtage on the weekend messed up my system so had to backup files and format C:\ and start over. Figured rescrape vs restore from backup, that's when I saw it... may have been this way for months... no idea.
System 1: HK1 RBox - Android 11, S905X4 Quad-Core, 4GB DDR3 - CoreElec KODI 20.3
System 2: Beelink GT1 Ultimate - Android 7.1, S912 Octa-Core 1.5 GHz, 3GB DDR4 - CoreElec KODI 18
Storage: WD MyCloud EX2 Ultra NAS - 2x4TB WD Blue (WD40EZRZ)
Display: 55" TCL 55R615-CA 4K TV
Reply
#13
Hey guys,

Thought I could add some insight Big Grin

The "primary" image is a new feature on the website but it really has no affect on the API. As long as you're grabbing the 'poster_path' field from the movie response, it will select the image marked as "primary". If one isn't marked as primary, we default to the previous behavior of highest rated image per language is what we do.

If I was to guess the issue, Kodi is not using the 'poster_path' field for the default selected image but rather is calling the full images method and sorting by rating and simply grabbing the highest rated.

Obviously Kodi can keep calling highest rated but this feature was introduced so we could more reliably serve consistent images. The highest rated isn't always the best.
Reply
#14
Thanks Travis - hoped you'd see this and respond.

One question though - who decides what the primary image is? wasn't that the whole idea of having a ratings system.
Reply
#15
I always wanted to see something like this happen. Some older art is the incorrect 2:3 ratio and looks bad when flipping through the movie list, but, image so old its been voted on so many times and no way to knock it down the list.

Figured best way would be to make the highest rated 2:3 image the primary... May not be the best but at least the image is the right dimensions.
System 1: HK1 RBox - Android 11, S905X4 Quad-Core, 4GB DDR3 - CoreElec KODI 20.3
System 2: Beelink GT1 Ultimate - Android 7.1, S912 Octa-Core 1.5 GHz, 3GB DDR4 - CoreElec KODI 18
Storage: WD MyCloud EX2 Ultra NAS - 2x4TB WD Blue (WD40EZRZ)
Display: 55" TCL 55R615-CA 4K TV
Reply

Logout Mark Read Team Forum Stats Members Help
movie scraping, issues with tmdb0