I started watching Bubblegum Crisis: Tokyo 2040. It has robots and androids, collectively known as "boomers". "Boomer" already has a meaning, so funny double meaning scenes were unavoidable. So, I decided to take a screenshot every time someone says "boomer".
For this task, I first extracted the subs in VOBSub format with MKVtoolNix, then converted them to SRT (text) with FFMES. I then made a Python script that uses the srt library to look for occurences of a word and return the time codes when it appears. Use it like
./rip.py subs3.srt boomer| tail -n+2 | cut -d " " -f
7 > k3.txt
I was going to use FFMPEG to extract the screenshots, but turns out, it's easier with mpv Here's line I used:
mpv --quiet --no-audio --vo=image --start=0:03:02.309000 --frames=1 --vf=sub 01\ Can\'t\ Buy\ a\ Thrill.mkv
I made another script that gets the shots at the specified times.
No comments:
Post a Comment