Plugin: EmbyCredits, detect end credits and add auto skip.

[yocker 1677]

New version "2.5.7.0" in the catalog.

Replaced the black frame detection method with a new much faster method.
Settings now auto save except the rules system which still needs the save button.

Please notes:
Any set rules MIGHT have to be updated as i had to change some things in the rules, very sorry about that!

[yocker 1677]

New version "v2.5.7.1" in the catalog.

Quick fix for hostnames not working in the OCR server settings.

[yocker 1677]

New version 2.5.7.5 available in the catalog.

Added new visual aid to fine tune time stamps with, this is available when pressing the "play at time stamp butto.
Also small bug fixes.

[Babatom 16]

@yockerI've been trying to get the plugin working on Unraid with Emby running in Docker but keep getting "Video validation timed out after 10 seconds" for every episode.

Emby itself works fine, transcoding and everything. Im on Unraid with the linuxserver/emby Docker image. Did I miss something in the config or is this a known issue with Docker setups? Tried disabling hardware aceleration already but same result.

credits-detection-debug-2026-05-17T20-01-58.log

[yocker 1677]

20 minutes ago, Babatom said:

@yockerI've been trying to get the plugin working on Unraid with Emby running in Docker but keep getting "Video validation timed out after 10 seconds" for every episode.

Emby itself works fine, transcoding and everything. Im on Unraid with the linuxserver/emby Docker image. Did I miss something in the config or is this a known issue with Docker setups? Tried disabling hardware aceleration already but same result.

credits-detection-debug-2026-05-17T20-01-58.log 10.58 kB · 0 downloads

Thank you for reporting it.  

I removed an argument from FFprobe that might cause it to take a while some times and causing the time out on some systems.
It's in the catalog now.

If this doesn't fix it then it could be a number of things wrong and i would need a log from Emby with debugging enabled as well to fix it.

Forgot to say:
You can also try and disable video validation.
It checks for if there is actually a video stream before running detection, without it and if a file doesn't have a video stream the detection will just fail on the file and move on.

Edited May 17 by yocker

[Babatom 16]

@yocker

It seems to be working now—thank you very much! Now I'll just run a full scan.

[yocker 1677]

New version up in the catalog (2.6.2.0).

New OCR method that improves the detection speed.


Smart Frame Skipping very simply dynamically skips frames and thus skips frames not needed for detection and in that way improves the detection speed.

Remember to report any bugs you might find!
Enjoy!

Edited June 9 by yocker

[sydlexius 322]

@yockerSo much for summer holiday?

[yocker 1677]

47 minutes ago, sydlexius said:

@yockerSo much for summer holiday?

Not me that added this function so my summer is still on.  

Edited June 10 by yocker

[malghana 0]

Hi,

 

Thanks for taking the time to develop and maintain this amazing plugin,

 

I'm running everything on Unraid, how can i install this and the OCR? i didn't find them or any docker compose

 

Thanks!

[yocker 1677]

1 hour ago, malghana said:

Hi,

 

Thanks for taking the time to develop and maintain this amazing plugin,

 

I'm running everything on Unraid, how can i install this and the OCR? i didn't find them or any docker compose

 

Thanks!

Tesseract and PaddleOCR does note have an Unraid template so you need to manually add them in Unraid's docker section.

@Neminem Made a great guide for PaddleOCR. Please note: It takes up A LOT of space, 25GB+ of docker space.
unRaid PaddleOCR setup guide. · yocksers/EmbyCredits · Discussion #5

There is also a guide for installing them in the plugin it self, there you can see how to install Tesseract or PaddleOCR.


Tesseract is a lot smaller but a little less accurate and slower but nothing major.

[yocker 1677]

Just made this in time for me to leave for the next summer event.
New beta up in the catalog (v2.6.4.0)

Added ability for the plugin to read and import metadata time chapters from MKV files.
If a credit (or some close variant) is detected it will be converted and imported into Emby as CreditStart.



The setting can be found in the Performance section.

[sydlexius 322]

14 hours ago, yocker said:

Just made this in time for me to leave for the next summer event.
New beta up in the catalog (v2.6.4.0)

Added ability for the plugin to read and import metadata time chapters from MKV files.
If a credit (or some close variant) is detected it will be converted and imported into Emby as CreditStart.



The setting can be found in the Performance section.

Any way to trigger this on its own?

[yocker 1677]

16 hours ago, sydlexius said:

Any way to trigger this on its own?

No but, if those chapters are detected then they will be used and detection skipped.
TimeMarkEdit can import them without detection running though.

[yocker 1677]

New version up in the catalog (v2.6.5.0)

Added:
1) The plugin can now detect, use and import chapter marks imbedded in MKV files.
If enabled the plugin will start by searching for embedded chapters and if there will use these instead of running detection.

[Rowan_89 0]

On 29/01/2026 at 09:22, yocker said:

New version (1.7.1) up on yocksers/EmbyCredits Github.

Added:

1. Setting to enable thumbnails being shown with a snapshot of the video at the detection point.
2. Play video at time stamp button.

Changed:

1. Moved OCR enhancement settings to their own tab as they are advanced settings that can change the detections a lot or even make them not work.
2. Moved Notifications settings to their own tab.
3. Settings are now in collapsible sections to make the settings page less cluttered and easier to navigate.

Fixed:

1. Scheduled task will no longer look for old chapter types and will now correctly skip videos with time stamps when the setting is enabled.
   Note: Emby seems to have changed chapter types in the last update so old detections had their chapter types changed and no longer valid, schedule task would wrongly pick those up before as having the correct time stamps.

Known bugs:
Intel QSV does not work, it will require an update of Embys FFMpeg to work, developers have noted that it's on it's way.
Please use VAAPI instead.

Thank you to @Neminemfor help with testing and bug hunting.

If you use and like the plugin please consider donating a coffee at:https://buymeacoffee.com/yockser
This is by no means necessary and i'm happy as long as people like the plugin.  

Just wondering if you managed to fix the Intel QSV with OCR? I have played around a bit and cant seem to get it to work. Im using unraid and forwarded the device to tesseract with it still not working. (I do have QSV working with emby without any issues). 

[yocker 1677]

3 hours ago, Rowan_89 said:

Just wondering if you managed to fix the Intel QSV with OCR? I have played around a bit and cant seem to get it to work. Im using unraid and forwarded the device to tesseract with it still not working. (I do have QSV working with emby without any issues). 

No, decided to not touch it before a new FFmpeg is added to Emby.
Vaapi should work and is just as fast as QSV for detection as all that is done is extract images for Tesseract/PaddleOCR to handle.

Quick question as you say QSV works for you in Emby, are you running the official Emby container or Linuxserver?
Asking because i can't get QSV to work on my server and makes testing a QSV fix very hard.

All that said, QSV *SHOULD* work as is when/if Emby has a working QSV since it uses the same FFmpeg as Emby.

[Babatom 16]

5 hours ago, yocker said:

are you running the official Emby container or Linuxserver?

It works for me with linuxserver. There was an ffmpeg update in the new beta.

[yocker 1677]

Added a new beta to the catalog (v2.6.6.0)
Changed a few FFMpeg arguments plus added "-hwaccel qsv" as fallback.
This *SHOULD* work and the fallback makes ffmpeg automatically try it self if everything else fails.

Just so people don't think it's a major problem (should ofc. be fixed anyway!!)
Vaapi in this instance will work and is just as fast as the hardware acceleration is only for extracting images and nothing else.

[sydlexius 322]

@yockerI noticed some behavior in this that you might want to put some guard rails on.  Specifically, I had an episode of a show whose credit detected before the intro.  Since I can't imagine a circumstance when this is ever a legitimate issue (I don't think old TV shows worked like old movies where the credit reel ran first), you might want to prevent the credit segment from being written if the IntroStart/IntroEnd tag timestamp comes after.  Perhaps it could even be used to weigh confidence score checks?

[yocker 1677]

30 minutes ago, sydlexius said:

@yockerI noticed some behavior in this that you might want to put some guard rails on.  Specifically, I had an episode of a show whose credit detected before the intro.  Since I can't imagine a circumstance when this is ever a legitimate issue (I don't think old TV shows worked like old movies where the credit reel ran first), you might want to prevent the credit segment from being written if the IntroStart/IntroEnd tag timestamp comes after.  Perhaps it could even be used to weigh confidence score checks?

Never thought about that, guess i shouldn't be surprised by now of how many different kinds scenarios there are.  
I will look into it asap.

Many thanks!

Edited 1 hour ago by yocker

[sydlexius 322]

The gratitude is all mine!  This does a seriously great job at covering a gap in my household's use of Emby!