Completed Show Intro Skip Option

[ZSZQ 10]

3 minutes ago, rbjtech said:

Try setting the parallelism in the plugin to 1 (no parallelism..) - that should limit CPU and disk resources to the minimum.

The Credit detection uses ffmpeg Black Frame Detection however - and that will use as many threads and CPU power as it can find  - we have no way of easily limiting that unfortunately, it's an ffmpeg limitation.

Alternatively, the Introskip Core may be a better option for you - as that currently just does Intro Detection and is, as I understand it, lighter on the CPU and disk as it is doing less work  - but it won't have all the features in the Introskip Plugin - such as editing and credit detection, nor can it detect Intro's over 10 minutes into the episode.

 

I have tried to change many parameter combinations before.
These settings will only affect the "Episode Audio Fingerprinting" task.
It is no effect on the task of "Episode Title Sequence Detection".

Whenever the "Episode Title Sequence Detection" task is run, the core occupation is 100%.
I don't know where there is a problem.
So in the end I can only reduce the core assigned to Emby.

In addition, all functions are running very well.

[chef 3745]

15 minutes ago, ZSZQ said:

 

In the past, plug -in scanning occupied very high resources.

The host of my host collapsed. I have to restrict the core of EMBY alone.
Is it improved now or in the future?
The core scan will not be so high.

 

 

Just tested, the scan occupation is still high.

Credit detection will be a very taxing process. It is cpu bound with no limits. It will take all the resources available.

However, it will also release resources if another application wants them.

 

Side note:

There just so happens to be another way to detect credit squences, without using black frame detection. Surprise!

But that is a topic for another day.

 

[rbjtech 4254]

3 minutes ago, chef said:

Side note:

There just so happens to be another way to detect credit squences, without using black frame detection. Surprise!

But that is a topic for another day.

 

Hmm sound intriguing - Is this the OCR that Cheese was experimenting with or something new and wizzy ? 

[ZSZQ 10]

6 minutes ago, chef said:

Credit detection will be a very taxing process. It is cpu bound with no limits. It will take all the resources available.

However, it will also release resources if another application wants them.

 

Side note:

There just so happens to be another way to detect credit squences, without using black frame detection. Surprise!

But that is a topic for another day.

 

I'm just curious why the core will not have such a high CPU occupation.

But at least the plugin function is still the most complete. This is why I can't do without it.

If the core of the future can achieve the same function of plug -ins, I may not have to choose the core.

Edited May 24, 2022 by ZSZQ

[rbjtech 4254]

1 minute ago, ZSZQ said:

I'm just curious why the core will not have such a high CPU occupation.

But at least the plugin function is still the most complete. This is why I can't do without it.

If the core of the future can achieve the same function of plug -ins, I may not have to choose the core.

The 'Detect' phase uses the ffmpeg black frame detection for cedits - I'm pretty sure it's that which is causing your CPU spikes.

Both the Intro detect and the credit initial detect should not be causing you any issues - the Core uses the same technology, although implemented slightly differently.

It's only when it tries to accurately detect the Credit sequence to the nearest frame, does it jump in CPU resources.  It needs to do this because credits very often do not have repeat Audio that Intro's do - and thus, we cannot use the same method to detect them. 

If set to 1 for FP and 1 for Detect though, I am very surprised if that is giving your system issues for the short period it takes to do the credit detection.

 

[chef 3745]

10 minutes ago, rbjtech said:

Hmm sound intriguing - Is this the OCR that Cheese was experimenting with or something new and wizzy ? 

Hehe, something else.

@samuelqwe is a coding, guru, ninja. 

[ZSZQ 10]

9 minutes ago, rbjtech said:

The 'Detect' phase uses the ffmpeg black frame detection for cedits - I'm pretty sure it's that which is causing your CPU spikes.

Both the Intro detect and the credit initial detect should not be causing you any issues - the Core uses the same technology, although implemented slightly differently.

It's only when it tries to accurately detect the Credit sequence to the nearest frame, does it jump in CPU resources.  It needs to do this because credits very often do not have repeat Audio that Intro's do - and thus, we cannot use the same method to detect them. 

If set to 1 for FP and 1 for Detect though, I am very surprised if that is giving your system issues for the short period it takes to do the credit detection.

 

all right.
Then don't pay attention to this.
It can be run normally.

[crusher11 851]

There's no way to bulk-add an intro, is there? It failed to find one for The Alfred Hitchcock Hour, they're all 16 seconds long and start at 0:00...that's a lot of effort that ought to be able to be automated somehow.

[Cheesegeezer 3086]

23 minutes ago, crusher11 said:

There's no way to bulk-add an intro, is there? It failed to find one for The Alfred Hitchcock Hour, they're all 16 seconds long and start at 0:00...that's a lot of effort that ought to be able to be automated somehow.

Negative and not an option we are looking at currently. But might be available in the future

[crusher11 851]

2 minutes ago, Cheesegeezer said:

Negative and not an option we are looking at currently. But might be available in the future

Le sigh...

[Cheesegeezer 3086]

8 minutes ago, crusher11 said:

Le sigh...

I mean what’s available right now it bloody impressive.  Its way more than you get with any other media platform. 
 

we literally cant pander to everyone’s specific requirements. 
 

the tools are there to edit manually if a user has that OCD in them. 
 

This release… is a conversion release to aid in testing the core features and eventually we will move our intro detection out of the plugin. 
but remember they are 2 very different features at the minute.

@rbjtech said what this release is best, but the words escape me 

[rbjtech 4254]

6 minutes ago, Cheesegeezer said:

@rbjtech said what this release is best, but the words escape me 

Maintenance and Integration release ? I think .. 

[crusher11 851]

26 minutes ago, Cheesegeezer said:

I mean what’s available right now it bloody impressive.  Its way more than you get with any other media platform.

I'm just sighing at the impending tedium of manually adding all those intros, rather than at anything to do with your work.

[Banquero 30]

Cuando estará este plugin en el catálogo de Emby? No me apetece instalarlo por SSH, y además espero que los desarrolladores de Emby, implementen un "boton" para elegir si saltar la introducción o no..Es esto posible?..Entiendo que sería correcto pagar por este pedazo de plugin..Muchas gracias...

[samuelqwe 313]

19 minutes ago, Banquero said:

Cuando estará este plugin en el catálogo de Emby? No me apetece instalarlo por SSH, y además espero que los desarrolladores de Emby, implementen un "boton" para elegir si saltar la introducción o no..Es esto posible?..Entiendo que sería correcto pagar por este pedazo de plugin..Muchas gracias...

Already available in the catalogue, and if you have the chapters enabled in the plugin, you should see the intro skip button.

[Cheesegeezer 3086]

11 minutes ago, samuelqwe said:

Already available in the catalogue, and if you have the chapters enabled in the plugin, you should see the intro skip button.

Because of the new chapter markers you will need to refresh metadata (not images) on your tv libraries.

the run the chapter task again, if you want to see them in your chapter display you will also need to run image extraction 

[datanet 120]

5 hours ago, Bagul said:

Thank you for your answer it is very clear!  
 

So for now the plugin is way ahead of the Emby core.

 

4 hours ago, chef said:

That might not necessarily be the case.

They are two separate entities with the same objective.

The plugin has an editor, but it works around what the core implements.

@chef is being modest.  Yes, the plugin currently has far more features than the core implementation.  With the exception of the manual skip button that is available only in the core implementation (which actually can be exposed to the plugin now), the core implementation is a small subset of the autoskip plugin feature set. 

I can understand the tricky and somewhat political position chef is currently in since he would not want to deter the adoption of people using the core implementation since the autoskip plugin team is working closely with the core Emby team to get to feature parity in the core implementation.  In addition, changes to the core code are required to further enhance the overall conceptual working capabilities of the autoskip feature.  Luke is going to err on the side of safety when it comes to implementing features since it will be baked in to the core code and those features need to work without error for the masses.  In the plug-in world, things are slightly more forgiving since plug-ins are seen as optional and if it doesn't meet your needs or work as expected, you simply remove it...no harm no foul.  So as well as the plug-in has been tested and actually works quite well, Luke is not going to assume all features work without issues for the masses and his inclusion of features to the core code will be conservative and incremental.  So feature parity in the core will probably take some time before the masses have access to everything that is available today in the plug-in.

As I stated in a previous post, I hope the two can co-exist and the plug-in version becomes more of an extension to what is implemented in the core...sort of, as I suggested previously, an autoskip+ or autoskip enhanced.  Where existing and new features can be added beyond what is available in the core implementation.  

Edited May 24, 2022 by datanet

[Cheesegeezer 3086]

8 minutes ago, datanet said:

 

@chef is being modest.  Yes, the plugin currently has far more features than the core implementation.  With the exception of the manual skip button that is available only in the core implementation (which actually can be exposed to the plugin now), the core implementation is a small subset of the autoskip plugin feature set. 

I can understand the tricky and somewhat political position chef is currently in since he would not want to deter the adoption of people using the core implementation since the autoskip plugin team is working closely with the core Emby team to get to feature parity in the core implementation.  In addition, changes to the core code are required to further enhance the overall conceptual working capabilities of the autoskip feature.  Luke is going to err on the side of safety when it comes to implementing features since it will be baked in to the core code and those features need to work without error for the masses.  In the plug-in world, things are slightly more forgiving since plug-ins are seen as optional and if it doesn't meet your needs or work as expected, you simply remove it...no harm no foul.  So as well as the plug-in has been tested and actually works quite well, Luke is not going to assume all features work without issues for the masses and his inclusion of features to the core code will be conservative and incremental.  So feature parity in the core will probably take some time before the masses have access to everything that is available today in the plug-in.

As I stated in a previous post, I hope the two can co-exist and the plug-in version becomes more of an extension to what is implemented in the core...sort of, as I suggested previously, an autoskip+ or autoskip enhanced.  Where existing and new features can be added beyond what is available in the core implementation.  

This is the idea of the plugin, we it will be a core IntroSkip++ version

[samuelqwe 313]

11 minutes ago, datanet said:

As I stated in a previous post, I hope the two can co-exist and the plug-in version becomes more of an extension to what is implemented in the core...sort of, as I suggested previously, an autoskip+ or autoskip enhanced.  Where existing and new features can be added beyond what is available in the core implementation.

Exactly, that’s the goal. This latest plugin update is mostly to allow people to "migrate" their data and start to consolidate the two data sources. The plugin needs to augment, not replace.

Edited May 24, 2022 by samuelqwe

[Banquero 30]

25 minutes ago, samuelqwe said:

Ya está disponible en el catálogo, y si tiene los capítulos habilitados en el complemento, debería ver el botón de salto de introducción.

Vaya, no lo había visto...Estoy emocionado..¿Entonces puedo elegir si saltar o no la introducción de una serie? Es decir se ve un "boton" en la pantalla, para decidir saltar o no?. Y otra pregunta...Una vez  que he actualizado los metadatos, que debo resetear para la detección de esta sesión, que aparece en la primera pantalla del plugin, o debo configurar algo más? Muchas gracias...

[Luke 37049]

1 minute ago, Cheesegeezer said:

This is the idea of the plugin, we it will be a core IntroSkip++ version

Right, rather than having to tell users to turn off the core detection, please tell me what api's you need so that you can work with the existing data and perhaps just add to it.

[datanet 120]

1 minute ago, Luke said:

Right, rather than having to tell users to turn off the core detection, please tell me what api's you need so that you can work with the existing data and perhaps just add to it.

This right here!!  Awesome job guys.

[samuelqwe 313]

1 minute ago, Banquero said:

Vaya, no lo había visto...Estoy emocionado..¿Entonces puedo elegir si saltar o no la introducción de una serie? Es decir se ve un "boton" en la pantalla, para decidir saltar o no?. Y otra pregunta...Una vez  que he actualizado los metadatos, que debo resetear para la detección de esta sesión, que aparece en la primera pantalla del plugin, o debo configurar algo más? Muchas gracias...

If you were using the plugin before this version and you had chapters enabled, you would need to refresh the metadata for your shows to reset their chapters, and then run the chapter task from the plugin to add the new intro markers.

Once that’s done (and assuming you have Emby Premiere) you should start to see the button that allows you to skip or not while watching your shows. Of course, also make sure auto-skip is turned off in this case, otherwise you won’t have the option and intros will skip automatically.

[Cheesegeezer 3086]

1 minute ago, Luke said:

Right, rather than having to tell users to turn off the core detection, please tell me what api's you need so that you can work with the existing data and perhaps just add to it.

@rbjtech can you help Luke so he can see whats in our TitleSequence DB.

if you can give us post methods privately or publicly, we can do this in the code behind so there isn’t duplication on acquiring the intro points.

this is gonna move fast buddy and will absolutely kill all the competition

 

[ebr 14910]

Why does the plug-in need any intro detection logic at all anymore?  Can't it just let the core do the detection of those and it only deals with end credits?