To add to this: Is the kind of precision that a 50ms step allows for even realistically necessary for most subtitle use cases? At least I don’t care if a subtitle appears 1/20th of a second earlier/later than the dialogue starts, and then it depends on my reading speed anyway. Plus, well, I’ve never encountered even professional subtitles (consistently) this precise anywhere in the wild.
For my experience, asynchronous subs start to become a problem only beyond a second or two out of sync, so if the UI does not easily allow for more complicated inputs, at least increase the steps by a factor of 5 or 10.
