Completed Show Intro Skip Option

[crusher11 851]

42 minutes ago, rbjtech said:

The source is largely irrelevant.  DVD's from the same season may have different Intro fingerprints - we have tested this ourselves by cut and pasting an Intro from a same season episode.  The 'new' episode, then picks up the Intro when the 'Authored' Intro did not - to you and I, it's the same.

As we have said before, there is a fine balance on the accuracy of picking up the Intro's - there may be some shows where the accuracy is lower than average - we have seen a few of those (for no 'Audible' reason what-so-ever) but as long as the majority are good (~90%) - then for a first release of emby doing this - you will have to deal with the manual correction of those items.

The source does matter if yours work and Cheesegeezer's don't, though, because there must therefore be a difference between them. 

[TXK57 32]

+1 for this, skip intro would be fantastic to have!

[chef 3745]

46 minutes ago, takeittothemax said:

+1 for this, skip intro would be fantastic to have!

Give the plugin a go. Let us know about your experience with it.

We're very interested in the result people are having locating intros in their media.

[kaj 241]

On 24/09/2021 at 02:51, Cheesegeezer said:

lo siento amigo, mi espanol no es muy bueno.

Pero, Si... debería estar allí.

Basically can anyone help him out. He has installed the plugin but he can't see the route on Synology DSM7.  Any of you guys have any experience with this.

Thanks

The paths changed quite a bit on DSM 7. You may have to get @cayars to chime in on this one.

[horstepipe 356]

10 hours ago, chef said:

Give the plugin a go. Let us know about your experience with it.

We're very interested in the result people are having locating intros in their media.

Is this already ready for ppl having media in the cloud - so for people who have to take care about quota usages?

So is there some kind of daily limit which we can adjust yet?

[Cheesegeezer 3086]

7 minutes ago, horstepipe said:

Is this already ready for ppl having media in the cloud - so for people who have to take care about quota usages?

We did have a user previously that was accessing their media via VPN and some virtual server, his internet was 60MB/s (not Mb/s) and he experienced long processing times for each episode (which is to be expected) There is a lot of Disk I/O and memory I/O, so you will be limited to the speed of your network.

If you can access you media directly to run the processes this is advised.

Quote

So is there some kind of daily limit which we can adjust yet?

You could set this yourself in the Schedule Tasks to run for 1hr/2hr, what ever you are comfortable with for Fingerprint & Detection

The biggest hit will be your initial scan, then after that it will be very low usage.  to give you an idea, a person with 30000 episodes will take approx 2 days to get the chromaprint. then another 11hrs to complete the detection. 

My advise for you is to wait until the plugin is deemed for official release before going for it on your main system.

Edited September 26, 2021 by Cheesegeezer

[horstepipe 356]

But the „initial scan“ can also be limited to e.g. two hours a day, so it won’t hit all episodes in a very short time frame?

[Cheesegeezer 3086]

1 minute ago, horstepipe said:

But the „initial scan“ can also be limited to e.g. two hours a day,

There are 2 tasks that need to complete.  

1. Chromaprint (Audio Fingerprint) - This extracts the audio for the episode and stores it in a database (ready for when the detection task needs it)
2. Detection - This will then compare audio from episode to episode until it finds a match (Intro) and set the Title sequence duration.

So you would want to complete the chromaprint first, delete any task triggers for Detection and chapters. Set the chromaprint to 1-2hrs a day. starting at 1am.

In the log it will tell you how many episodes need to be processed by the chromaprint, when the task is initiated  (however we may look to add this to the UI for some feedback to the user)

1 minute ago, horstepipe said:

 

so it won’t hit all episodes in a very short time frame?

I'm not sure I follow this.  it will take the time it takes to extract the audio (this is Disk I/O heavy) 

The guy that did this quite a few revisions ago, was experiencing 9 secs for extraction locally, but when doing it through VPN and virtual server... it was 200sec per episode.

[horstepipe 356]

Ah ok I will just try it out, I can still abort it if it drives crazy. 
Thanks for the explanations!

[chef 3745]

That's an interesting idea, to have limit.

Limit the over all task, but attempt detection on all new media.

It would take some doing to get right.

Currently we take media from the beginning of a series, and move through the seasons. But in order to properly limit the task, you'd want to take any seasons that are currently airing (have upcoming episodes), and do those first. That way you'd have access to intro data for episodes that are being currently watched, and deal with older season episodes later. You could even handle running the task when the server was in an idle state (no sessions streaming ), for short bursts of time.

It's definitely an interesting idea.

The plugin would need a completely overhauled, but there are some interesting concepts there.

 

 

 

Edited September 26, 2021 by chef

[chef 3745]

I suppose, you could add all your series to the exclude list, except 1 of them. 

That would limit your task as well.

[neik 837]

5 hours ago, horstepipe said:

Is this already ready for ppl having media in the cloud - so for people who have to take care about quota usages?

So is there some kind of daily limit which we can adjust yet?

I'm right in the middle of fingerprinting stuff and haven't had a quota issue so far and I let it run 24hours/day. 

So, from my experience quota isn't an issue, at least for fingerprinting.

[crusher11 851]

Another question re: Alfred Hitchcock Hour...how does it go through the episodes to determine matching? The intro for episode one is a false detection, so if it starts at episode one and moves through...is it possible that skipping over episode one would yield better results?

[chef 3745]

39 minutes ago, crusher11 said:

Another question re: Alfred Hitchcock Hour...how does it go through the episodes to determine matching? The intro for episode one is a false detection, so if it starts at episode one and moves through...is it possible that skipping over episode one would yield better results?

It will scan:

Skipped 1 and 1 (can't compare the same episode)

1 and 2

1 and 3

1 and 4 etc.

 

Then:

2 and 1 (yes we do it again if a intro wasn't found, in the future we may stop this from happening)

Skipped: 2 and 2 (can't compare the same episode)

2 and 3

2 and 4 etc.

 

 

 

We skip episodes if they both have intro data.

Edited September 26, 2021 by chef

[crusher11 851]

So it would compare one to two, fail, compare one to three, match that, then use that for everything thereafter? As such, if it started with episode two I might get a better result, right? Is there a way to achieve that without completely screwing my metadata/filenames/server?

[horstepipe 356]

Is there a UI implementation / button for that feature yet? I guess that would have built into the core / into the apps code? 
So if the plug-in has finished its work, how can it be used for now, just via the chapter marks I guess?

[TXK57 32]

I may have missed it but I'm not seeing where to grab this plugin. I took a look at your github and did not see it there. Could you provide the link so I could test it out?

[seanbuff 840]

4 hours ago, takeittothemax said:

I may have missed it but I'm not seeing where to grab this plugin. I took a look at your github and did not see it there. Could you provide the link so I could test it out?

Try this: https://emby.media/community/index.php?/topic/101687-introskip-instructions-beta-releases/

 

[crusher11 851]

20 hours ago, crusher11 said:

So it would compare one to two, fail, compare one to three, match that, then use that for everything thereafter? As such, if it started with episode two I might get a better result, right? Is there a way to achieve that without completely screwing my metadata/filenames/server?

@chef If I remove episode one, scan it, then add episode one back in, what will happen? It'll just compare episode one to the existing, hopefully correct, fingerprint right? 

[TXK57 32]

10 hours ago, seanbuff said:

Try this: https://emby.media/community/index.php?/topic/101687-introskip-instructions-beta-releases/

 

Thank you, that is what I needed!

[chef 3745]

20 hours ago, crusher11 said:

So it would compare one to two, fail, compare one to three, match that, then use that for everything thereafter? As such, if it started with episode two I might get a better result, right? Is there a way to achieve that without completely screwing my metadata/filenames/server?

If we match 1 & 3, we try 1& 4 etc.

 

Just because 1 & 3 matched doesn't nesessarily mean 3 will match with other episodes, so we try them all, skipping over episodes where both have intro data.

 

Most of the time, I have noticed that I get results within the first couple episodes.

But, I see your point. You might get a better detection between two other episodes, and we are taking the first positive detection to save time.

 

Perhaps, in the future, we could scale a detection percentage of episodes (between 0 and 1). If an episode detection scaled higher we could take that data instead.

However, the run time of the task would grow for the initial task.

It's definitely worth keeping in mind.

[chef 3745]

36 minutes ago, crusher11 said:

@chef If I remove episode one, scan it, then add episode one back in, what will happen? It'll just compare episode one to the existing, hopefully correct, fingerprint right? 

If you remove one episode in a completed season, the season won't have the fingerprint data available to scan the one removed episode. 

 

If the season is on going (still has episodes which haven't aired yet) then you could remove the episode from the data base and rescan just one. 

But you can't do that in the UI.

[TXK57 32]

I grabbed the plugin and tossed it in there but I need to change the path that it is looking for the files. The fingerprint task is failing since it is looking for C:\Users\***username***\AppData\Roaming\Emby-Server\programdata\data, and I'm using the portable version and running as a service. My data is not under a user appdata.

Is there a config file or something I can change this in? I would think that it should auto recognize the path based on Emby but I guess not.

[crusher11 851]

56 minutes ago, chef said:

If you remove one episode in a completed season, the season won't have the fingerprint data available to scan the one removed episode. 

 

If the season is on going (still has episodes which haven't aired yet) then you could remove the episode from the data base and rescan just one. 

But you can't do that in the UI.

Right, I'd have to delete the data for the entire season, but that's a given because it's all garbage anyway.

[rbjtech 4266]

it's all under ..\programdata\Plugins\.. the root path doesn't (or shouldn't) matter.