Chapter images not displayed in TV series.

[ryzen5000 12]

Hi, I am trying to figure out which setting is responsible for chapter images on a TV series, is it possible that they never got extracted? Or is this just the way that EMBY

is? There are no actual images on  display along the bottom of my TV screen. How can I make the little grey boxes display real chapter scenes from the show?

 

 

https://officerat.store/

Edited November 17, 2022 by ryzen5000

[Happy2Play 8296]

What client is that?  As this is what I see on web client and Roku.

But what is your per library "Video Preview Thumbnails" setting?

[ryzen5000 12]

I'm using the Roku client. My setting for "Video Preview Thumbnails" was set to never. I changed it to scheduled and when media is added. Will report back if this solves the problem after its done scanning.

 

 

https://officerat.store/

Edited November 17, 2022 by ryzen5000

[Happy2Play 8296]

@ryzen5000 You have to decide if you just want Chapter images or interval images. 

But yes this will fix your issue.

[ryzen5000 12]

How do I select between the two? What are interval images? I think I'd like all the images. The server unexpectedly crashed when I went to scheduled tasts and clicked to scan thumbnails. This is kinda odd it was only CPU %20 during the scan, I am rebooting now.

Edited November 15, 2022 by ryzen5000

[Happy2Play 8296]

9 minutes ago, ryzen5000 said:

How do I select between the two? What are interval images?

When you enabled "Generate video preview thumbnails:" it gave you two options via "Thumbnail images interval:" (Chapter marker or 10 second interval images ie bif files)

Chapter marker will only extract images at set chapter markers.  So all seek between markers will show same image.

Interval will populate your chapter marker but also give you scroll/seek images at 10 second intervals.

[ryzen5000 12]

I only have three options on mine.

Video Preview Thumbnails

Generate video preview thumbnails:

Never

As a scheduled task

As a scheduled tast and when media is added

and then a box "Save video preview thumbnails into media folders"

[Luke 37112]

You have to click the Advanced toggle at the top of the dialog to see those. Most users will not need to.

[ryzen5000 12]

Then I clicked Thumbnail image extraction here and its taking a very long time despite my server having one of the most powerful CPU. It doesn't seem to be using all the resources.

[Luke 37112]

It's normal that it takes a long time. The help text underneath the option informs you of this when you enable it.

[ryzen5000 12]

Yeah got it now I see it, I never noticed the advanced toggle, thank you. It was set to 10 second intervals and I think thats fine for me. so now I just need to let it run for a couple days? I only have 40 TB of movies and TV series. Its there any way to crank it up and prioritize the scan? I am scanning the correct way? I don't mind if it uses more CPU and finishes faster.

[Happy2Play 8296]

Power really does not matter Emby as has to probe every item from start to finish.  So depending on your library size it can take days.

This is why the scheduled task has a time limit.  But if you manually trigger the task the limit does not apply.

[ryzen5000 12]

If I pin more CPU cores to Emby will it complete faster?

[ryzen5000 12]

Ah, thank you so much for helping me. There are more features I am sure yet that I haven't discovered. This will make the viewing experience much nicer.

[Happy2Play 8296]

1 minute ago, ryzen5000 said:

If I pin more CPU cores to Emby will it complete faster?

Not that I am aware of.

[Luke 37112]

I'm not sure as the answer is specific to your environment, but you can certainly try. This process has to step through and decode entire video files, which means disk performance could be just as much of a limiting factor as cpu.

[ryzen5000 12]

I am good on disk performance as I am a really big fan of Seagate Exos 18TB X18 Enterprise and they can read and write faster than 280 MB/s. The shares that I don't have cached to my 3500MB/s SSD's still write directly at around 284MB/s they could be faster I don't know because 2.5G is the max for the network I have set up. Not to mention I don't waste my time with parity drives to slow down my system write speed because every component and disk inside my machine is new I think there is less than a .00001 percent chance of failure on these disks they are soo good.

Edited November 16, 2022 by ryzen5000

[ryzen5000 12]

I think the speed of scanning has something to do with the speed of the database. Is there any way to switch Emby to a faster database, like an external one in a docker? Postgresql14 mabye? I read I can uninstall plugins I don't need to speed things up, I will remove the backup plugin and some others  and see if speed improves.

Edited November 16, 2022 by ryzen5000

[Luke 37112]

15 hours ago, ryzen5000 said:

I think the speed of scanning has something to do with the speed of the database. Is there any way to switch Emby to a faster database, like an external one in a docker? Postgresql14 mabye? I read I can uninstall plugins I don't need to speed things up, I will remove the backup plugin and some others  and see if speed improves.

I can promise that's not the limiting factor in this. 

[ryzen5000 12]

I know my machine is capable of more speed, although its much faster than Amazon Prime already. The only upgrade I could possibly do to my system is use a Gen 4 NVME for the cache drive instead of Gen 3 2TB. But I am not sure that I would notice any difference. Other than that I look in any direction and I can't find any parts online that would be an upgrade. I simply bought the most expensive parts I could find and all the highest specs that were available at the time I built it.

I wonder if there would be performance gains by another CPU.

I could upgrade my CPU from the Ryzen 7 5800X 28124 to the The AMD Ryzen 9 5950X 45828

 

[Luke 37112]

@softworkz may have some tips.

[softworkz 3340]

1 hour ago, ryzen5000 said:

I know my machine is capable of more speed, although its much faster than Amazon Prime already. The only upgrade I could possibly do to my system is use a Gen 4 NVME for the cache drive instead of Gen 3 2TB. But I am not sure that I would notice any difference. Other than that I look in any direction and I can't find any parts online that would be an upgrade. I simply bought the most expensive parts I could find and all the highest specs that were available at the time I built it.

I wonder if there would be performance gains by another CPU.

I could upgrade my CPU from the Ryzen 7 5800X 28124 to the The AMD Ryzen 9 5950X 45828

Before reading on, you should read this post first: https://emby.media/community/index.php?/topic/78173-multiple-thumbnail-extract-processes/&do=findComment&comment=907826

in general, the only effective way to accelerate library scanning is parallelization, but that's not as easy as it might seem to do for us and then there are still hard limitations (see below), but there is ONE thing that can actually be done in parallel: library scanning and image extraction.

But when you choose to perform image extraction as part of the library scanning - you achieve the opposite. It's a situation like when you'd be a business owner with two pretty fast working employees. But then you tell your employees that each one needs to stop and stand still when the other one is working.
Hence, my #1 tip is to decouple library scanning from thumbnail generation, which means to choose not to perform image extraction as part of library scanning.
I understand that some users don't like the fact that extraction happens just once at night and consider that as being "too late", but well: you should be able to add an interval trigger to the extraction task and let it run every 15 minutes. This allows some real parallelism. Another advantage of the decoupling is that you might get a different view on which part of the operations takes how much time.

One other point is about what you already mentioned: plugins. But you don't need to blindly uninstall plugins - just focus on metadata plugins and which are enabled for your individual libraries and which ones are needed and which might not be needed.

These are the two things you can do, but there's not much beyond, and surely not about hardware. You could buy some Million dollar hardware and it won't get better.
I'll explain by an example - by coincidence, I'm sitting right in front of it.

For a new feature, I made a test where I imported 10k music videos - but remote items, not local items.

• It's just a list. No local library, no media files - just a list.
• The import happened withing a fraction of a second and the 10k items were immediately visible in Emby.
  As Luke already said: this has nothing to do with database performance, not even remotely.
• For remote items, Emby doesn't do image extraction (neither single preview nor chapter/thumbnail images)
• Also, it doesn't probe remote items (only when you play for the first time)
• There's no disk I/O, no mem usage, no CPU usage, no high network usage

And still, it's running for more than two hours already - how can that be?

[softworkz 3340]

Just now, softworkz said:

...how can that be?

An important part of library scanning is metadata retrieval:

• Identify items
• Retrieve data
• Query for available images
• etc.

This procedure requires API calls to be made to various data providers. And most of them have rate limits - either technically implemented or from agreements we made with some.
That means that when we have made an API call to a provider, we need to wait a certain time before we can make the next request to their API.
If we wouldn't adhere to that - or we would allow you to configure/remove it - they would block you (some sooner, some later) - which would be the better outcome. The worst would be when all Emby users would get blocked (that's why there is no such configuration for this  )

To conclude: this is a hard limit that you cannot overcome with more/better hardware. You can only try to rule out all those things that you can change, like the two things I mentioned above.

[ryzen5000 12]

 

[ryzen5000 12]

"Hence, my #1 tip is to decouple library scanning from thumbnail generation, which means to choose not to perform image extraction as part of library scanning." - I will change this setting.

"metadata plugins" - Got it. I can probably disable some of them