Is there a way to rip hardcoded subs from a video into a .srt file?
Have tried subrip and avisubdetector, without any success
Sure, I hope you know image processing, how to program and have a lot of free time.
>>54498492
Hardcoded...You mean rendered into the video? Listen, just cave and write it up from scratch yourself cause you won't have any success doing what you're asking.
That's like asking if there's any way to rip layers from a .jpg.
>>54498581
>>54498547
I know it sounds absurd, but there are threads mentioning how subrip has been successful in doing so, but I cant get it to work. I just need a simple yes or no, so I can be content with my failure
>>54498610
Yes, it is possible. But not for you since you have to ask. It would not be worth your time and effort. There is no existing program that will do what you want, but it is still possible if you had the technological prowess, which you don't.
Most subs are easily available online. Why would you want to do this?
Just get a better source for the movie/video whatever.
>>54498649
now I'm curious
>>54498684
no english subs for this film ;_;
>>54498492
Yes, there is a way.
1.) Isolate the subtitles. You can limit the search in the lower third of the image or so. When subtitles appear you'll have a characteristic sudden change in color composition (many FFF-pixels or whatever the color of the subtitles is). You can detect that. If it happens, save the image somewhere with two time stamps: When it first appeared and when it disappeared. Maybe you can exploit correlation between successive frames to isolate the subtitles more. Or just average all the frames with one string shown and save that one. I don't know.
2.) Feed those images to an text recognition framework. There are some free/open ones. If you did (1) well, this should work. You use the time stamps to generate a simple *.srt file. That should be very simple.
The hardest part is probably to get familiar with the text recognition framework. When that is done, it's not difficult. Whether it will be successful largely depends on the quality of the source.
>>54498492
The easiest way is this.
1. Use a screen capturing program to take a snapshot of the movie once every second of the film. Make sure the snapshots include timestamps.
2. Use photo editing software to batch crop the snapshots to only the lower third of the screen.
3. Use text recognition software to batch analyze the snapshots and create a text file for you based on what they see.
4. Create a .srt file by hand using a translation program or source, and the timestamps attached to each snapshot.
There is no easy way to do this.
>>54498739
What's the name of the film?
>>54498610
>>54498688
OCR, you stupid idiot.
>>54499054
Звёздный мaльчик (1958)
>>54499181
Upload the movie to youtube and allow the auto-generated caption to run, and then use automatic translation.
It will come out as mostly gibberish, but maybe you can make educated guesses from that point.
>>54499557
Well, it's based off of a story which I've already read. But thanks
>>54499682
>random unrelated fact
>"but thanks" micro aggression
https://www.youtube.com/watch?v=ml1VJs5MmoY
>>54499780
>Japanese trip
>Probably some 23 year old white guy living with his mom
Why aren't you posting this in /sqt/ baka
