No Actor images in new library creation

[rdf52 5]

I am taking a different approach from my last attempt at getting Emby up and running to replace Plex.  My last problems dealt with actor images disappeared from the detail page when I would refresh metadata on a particular movie.  I think the general consensus was that the embedded actor metadata was causing the issue.  Here's that link: 

So this time, I took 50 movies out of my Plex library and using ffmpeg, removed ALL metadata from the mp4 videos.

I downloaded tiny media manager and created nfo files for all 50 movies.

Created a brand new installation of Emby and created a new movie library with ALL options in the library settings OFF except for nfo reader (ON) and download images in advance (ON).

After I created the library, took about 30 seconds, all movies were in the library but none of the movies showed the image of the actor, directors, etc.  A placeholder was shown and the names of the actors is shown.  If I click on the image placeholder, the correct image comes up immediately.  If I then return to the movie detail page and refresh the browser, the image of that actor is shown.

My NFO files all contain this type of information on each actor in each movie:

 <actor>
    <name>Kevin Costner</name>
    <role>Agent Ethan Renner</role>
    <thumb>https://image.tmdb.org/t/p/h632/bykmxJHLfbFM3NT05RZXhx8YTzF.jpg</thumb>
    <profile>https://www.themoviedb.org/person/1269</profile>
    <tmdbid>1269</tmdbid>
  </actor>

My server text file is attached. 

 

embyserver(2).txt

[Luke 42077]

HI, it's due to disabling internet metadata. Currently the actor image urls are not imported from nfo files.

[rdf52 5]

This is broken!  If I click on the actor image placeholder I get an image from the nfo file.  

What is the use of having all the metadata reader selectors and being able to arrange them in order of preference if they are not honored during an import.

[Luke 42077]

Quote

If I click on the actor image placeholder I get an image from the nfo file.  

No, that's not what happened. When you click on an actor, if there is no existing image, it will try and get one on demand.

[Luke 42077]

Quote

What is the use of having all the metadata reader selectors and being able to arrange them in order of preference if they are not honored during an import.

I think we did this a long time ago because when users have bad urls in their nfo files, then they just end up coming to us as if it's a problem with the software. So I think at some point we just said enough was enough and stopped importing it.

[rdf52 5]

6 minutes ago, Luke said:

No, that's not what happened. When you click on an actor, if there is no existing image, it will try and get one on demand.

It may be getting the image from the internet, but the name to search for came from the nfo file as I highlighted in my first post.  If during the import it got the name of the actor from the nfo why couldn't it download the image at that time as it has the URL of the actor's image in the nfo.  

[Luke 42077]

19 minutes ago, rdf52 said:

It may be getting the image from the internet, but the name to search for came from the nfo file as I highlighted in my first post.  If during the import it got the name of the actor from the nfo why couldn't it download the image at that time as it has the URL of the actor's image in the nfo.  

That would slow the library scan to a crawl, and if the images all come from the same domain, such as Tmdb, then we could end up hammering them with image requests. (Which they have asked us not to do).

So it happens later, on demand, when the image is actually needed. Of course by then, the image url may no longer be valid, or maybe it was never valid to begin with, and we end up having to deal with the mess in the form of troubleshooting.

[rdf52 5]

8 minutes ago, Luke said:

I think we did this a long time ago because when users have bad urls in their nfo files, then they just end up coming to us as if it's a problem with the software. So I think at some point we just said enough was enough and stopped importing it.

That may be so, but today there seems to be no set procedure or priority for importing the detailed information of a video, or at least it is not documented.  If you had in the library settings these metadata sources, for example:

• NFO
• Local/Embedded
• TheMovieDB
• TVDB
• IMDB
• Etc

With the ability to enable/disable each one and drag them in order of preference, then what ever information was not located in the first source could be searched in the second then third source as long as each source was enabled.  Yes, eventually, for the sake of time, you would have to abandon the search after perhaps the third  source.

Today, if you have embedded data, I was told that took priority, but if you do a metadata refresh, then embedded actor or genre metadata must not have priority or my first attempt at using Emby to replace Plex would not have been an issue.  In that first try, all my videos imported with full details showing but as soon as you refreshed metadata for a video, the actor images disappeared.  The video still had the actor data embedded.  What changed that made it decide that was not the highest priority.

I am sorry to be such a pain in the a** but as a retired engineer who started writing code back in the Intel 8080 days, I understand procedures and priorities and how codes is structured.  I really want Emby to be my next media server platform but until I understand the procedures and priorities of how data is imported and refreshed I am not ready to go all in with Emby.  I have spent many hours working on my library details and would like to keep those details going forward with Emby.  

[rdf52 5]

4 minutes ago, Luke said:

That would slow the library scan to a crawl, and if the images all come from the same domain, such as Tmdb, then we could end up hammering them with image requests. (Which they have asked us not to do).

So it happens later, on demand, when the image is actually needed. Of course by then, the image url may no longer be valid, or maybe it was never valid to begin with, and we end up having to deal with the mess in the form of troubleshooting.

Not hammering the source make sense and I can certainly understand that.  But aren't you hammering the source if I import 2000 videos into Emby and when finished, ALL the metdata images are showing?

[Luke 42077]

4 minutes ago, rdf52 said:

That may be so, but today there seems to be no set procedure or priority for importing the detailed information of a video, or at least it is not documented.  If you had in the library settings these metadata sources, for example:

• NFO
• Local/Embedded
• TheMovieDB
• TVDB
• IMDB
• Etc

With the ability to enable/disable each one and drag them in order of preference, then what ever information was not located in the first source could be searched in the second then third source as long as each source was enabled.  Yes, eventually, for the sake of time, you would have to abandon the search after perhaps the third  source.

Today, if you have embedded data, I was told that took priority, but if you do a metadata refresh, then embedded actor or genre metadata must not have priority or my first attempt at using Emby to replace Plex would not have been an issue.  In that first try, all my videos imported with full details showing but as soon as you refreshed metadata for a video, the actor images disappeared.  The video still had the actor data embedded.  What changed that made it decide that was not the highest priority.

I am sorry to be such a pain in the a** but as a retired engineer who started writing code back in the Intel 8080 days, I understand procedures and priorities and how codes is structured.  I really want Emby to be my next media server platform but until I understand the procedures and priorities of how data is imported and refreshed I am not ready to go all in with Emby.  I have spent many hours working on my library details and would like to keep those details going forward with Emby.  

We do have that already. The only difference is that NFO is always at the top if it is enabled.

[Luke 42077]

1 minute ago, rdf52 said:

Not hammering the source make sense and I can certainly understand that.  But aren't you hammering the source if I import 2000 videos into Emby and when finished, ALL the metdata images are showing?

Relatively, no. 2000 movies is not that big of a deal. What could be a big deal though is when each of them have anywhere from 10-30 actors and we're getting all of those images one right after the other.

[rdf52 5]

9 minutes ago, Luke said:

Relatively, no. 2000 movies is not that big of a deal. What could be a big deal though is when each of them have anywhere from 10-30 actors and we're getting all of those images one right after the other.

Well obviously the code already does exactly that.  In one of my test-beds I imported 2014 videos and 25,000+ episodes of tv shows.  When the import procedure was completed all images, including all actors, directors, writers, etc, were showing up the first time I viewed the results.  If things stayed that way we wouldn't be having these discussions.  Refreshing metadata made the actor, writer, director images disappear. 

[Luke 42077]

3 minutes ago, rdf52 said:

Well obviously the code already does exactly that.  In one of my test-beds I imported 2014 videos and 25,000+ episodes of tv shows.  When the import procedure was completed all images, including all actors, directors, writers, etc, were showing up the first time I viewed the results.  If things stayed that way we wouldn't be having these discussions.  Refreshing metadata made the actor, writer, director images disappear. 

OK then please open a topic about that and provide a specific example. Thanks.

[rdf52 5]

2 minutes ago, Luke said:

OK then please open a topic about that and provide a specific example. Thanks.

That was my first attempt at using Emby and was documented in this case: 

The title stated duplicate movies.  We figured that out pretty quick but then we got into why were my actor images disappearing.  That's when I was told it was because I had embedded data for actors.

That test-bed has been deleted to try this latest attempt but if need be, I can try all that again with my original video files from my plex media library.

[Luke 42077]

OK then let's focus on that topic. 

To close out this issue, I think we might be able to start importing the image url again during the nfo import. The image will be downloaded later on demand, but I think the server will recover better from bad image urls then it might have a few years ago. I think you would see the broken image icon once, and then the placeholder after that.

[rdf52 5]

37 minutes ago, Luke said:

I think we might be able to start importing the image url again during the nfo import

That would be ideal and would really solve the issue for me with the nfo only workflow. 

I was so sure that the nfo only workflow was going to work that I purchased 2 additional 8 TB USB drives.  I copied my entire Plex movie library onto the first new 8 TB drive.  I have had a script running for about 24 hours now that is using ffprobe to read all the current metadata from the plex video and write the data into a JSON output file on the 2nd new drive.  Then ffmeg is stripping all metadata from the video and doing a -codec copy to the 2nd new drive.  At the end of the process I was going to use tiny media manager make the nfo file have the same information as the original embedded plex video.

By the way, while doing some testing on the nfo only workflow, I stripped a couple of tv show with 4 seasons each, created nfo files and imported them into an Emby TV show library.  They acted the same as a movie.  All data was good but just had placeholders for the actor images.