@hellowlol a question for you.

[KBlixt]

Hi, I was unsure of how to contact you with a question since github removed their messenge feature.

I've been considering making a script that removes the green "approved for certain audiences" that appear ahead of trailers. Although I don't know where to start.

Considering that your project has to do with analyzing a fairly repetetive pattern from video file. I was wondering if you know of a way to do this.

I don't need a detailed explanation just a few key components that you know of that I could google and use in my script.

Sorry if this is the wrong place for this question.

Thank you.

[Hellowlol]

Hi! You need to identify some features that you can detect. The green stuff that you was talking about is that alsways the same?

If so i would probably use opencv template matching. See https://www.pyimagesearch.com/2015/01/26/multi-scale-template-matching-using-python-opencv/ for a good guide about that.

[KBlixt]

Yes, the screen is always the same kind of green (or very close) and the only other thing on the screen is some white text in the middle but at least a fifth of the screen from the edge is always the same green. So I recon that the easiest and most reliable way to detect the pattern would be to look for a green frame.

Normally this static picture lasts 5-6 seconds if it's there. It's also completely silent. Here is a few examples: https://www.google.se/search?q=green+approved+screen+trailer&client=ms-unknown&prmd=isvn&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjNveXFyMvbAhVOSZoKHTzBBGUQ_AUIESgB&biw=412&bih=652&dpr=2.63

Great thank you for the pointer. Although this seems to mainly be comparing edges. I'd like to compare pixel color in a given area. And then if the frame match the green during the first 8 seconds of the trailer it'll get trimmed. And obviously trim the audio stream with the same amount of time.

[Hellowlol]

Can you post a example picture?

[Hellowlol]

This is a example to detect color. https://www.pyimagesearch.com/2014/08/04/opencv-python-color-detection/ have you tried to just look for the text like the word preview? There is two functions in the credits.py file you can use

[KBlixt]

I linked a Google search. Pretty much all of them are what I'm talking about.

And while all of them are very similar the text does varies and I think the colour filtering that you linked to would give a more reliable result.

Thank you so much, now I know where to start.

[Hellowlol]

All of them had the word preview and audiences. I think that would be a lot simpler solution, slower but simpler it’s like 4 lines of code. Good luck 👍

[KBlixt]

hm, yeah. that's true. Thank you a bunch for pointing out a direction though.