Have a look at VideoSubFinder, it does text recognition on video files and outputs clean images of the text - you can then use an optical character recognition program to convert those into text files

I mostly use mine for functional stuff, odd shaped brackets, pipe adaptors, electronics enclosures, so not very photogenic, but my two favourite things were a replacement door latch for the microwave, saved the cost of buying a new one, and a very nerdy bit of wall art 🖖