[Help] Frame-accurate subtitles

[theprion]

Greetings, all!

I've been venturing myself into resyncing subs, and most of what I do is move the lines around on the audio track waveform, using freewares like Aegisub and Subtitle Edit.

I'm very meticulous about having it all playing in sync but I do it all by the naked eye (and ear). Although I've had good results with playback, I wonder if I'm not overlooking something or if my subtitles would hold up to professional standards that I'm ignorant of. 

I've seen experienced users mentioning "frame-accurate" subtitles and I would like to be advised on how to make sure, from a technical standpoint, that the subs are properly synced to video.

Thanks!

[Beber]

That sounds like me. When I said frame-accurate subtitles, I mean subtitles with time-codes that correspond to those of a frame, so that I don't have time-codes with a value that comes between frames, if that makes sense to you. The idea behind my reasoning is to have subtitles that appear and disappear at the exact same moment as the burnt-in subtitles when there are some. 'Cause there's nothing I hate more than having 2 sets of subs that are not in perfect sync with each others, as if it wasn't already a pain in the ass to have those 2 sets of subs which make it up to 4 lines of subs on screen. The rest of the time, you won't notice any difference if it's close enough.

The way I do it with free softwares (Visual Sub Sync) is to work with videos at 25 fps. Why? Because the free softwares use millisecond as smallest measurement: it goes hour, minutes, seconds, milliseconds. And as it turns out, at 25 fps, a frame is 40 ms long, so that's a nice round number to achieve. Doing it at 24 fps or 23.976 fps is a pain in the ass since a frame's worth is not a nice round number anymore.

Or you can go the professional route and use a pro software that actually works with frame as the smallest measurement and goes: hour, minutes, seconds, frames.

[theprion]

Yes @Beber, it was you indeed! Thanks for responding.

So can I assume that "frame-accuracy" in subs is more relevant to subbing burnt-in text and on-screen titles rather than captioning audio?

I mean, I can get it "virtually" synced already by just using the audio waveform as reference, and I don't even get to worry about frames that way. But is that good enough? Or should I get my head around the numbers?

[Beber]

That will be good enough. If your eyes don't tell you there's something off, you'll be fine. It will show when failing to have 2 sets of text perfectly in sync. That's when it's annoying.