Info Wanted: Frame differences in different projects?

There seem to be a handful of experts here on the frame differences between, say, the PAL GOUT and the NTSC GOUT, or various Laserdisc releases (see this thread).

What I’d like to know is if there is any way, other than manually comparing a couple hundred thousand still images against each other, to do this frame comparison for any two given releases?

Specifically, I’m interested in which frames are missing/added in our film-based preservations, compared to the NTSC GOUT: Puggo’s releases and the new Grindhouse ESB release from -1, and ideally also a quick way to find differences in future -1 releases.  If anyone has any ideas, let me know.

This is for subtitles, so I can handle being off by a few frames, as long as it’s reasonably close.  For the Grindhouse release, I tested a bulk-conversion script with some quick-and-dirty approximations, but I wouldn’t mind learning how to get more accurate values for the future.

CatBus said:

What I'd like to know is if there is any way, other than manually comparing a couple hundred thousand still images against each other, to do this frame comparison for any two given releases?

Diffing the begin-shot-change numbers is one (pretty good; not perfect) way to tell if two sources are in sync. (Empty diff => all shots in both sources begin at the same frame number.)

If the two sources are sufficiently similar, then it's a 99% guarantee.

I can give further instructions if you're interested.

Yeah, I discovered this while syncing Puggo's projects, but never reported back, like a bad, bad thread starter.

Since I'm dealing with subtitles, I really can afford to be off by quite a few frames.  Jumping from shot to shot and verifying sync in each shot is perfectly adequate for my purposes.

Please take a look at this ASAP: https://github.com/tp7/Sushi

Scroll down to read the description/README.

That's really cool.  I can imagine a couple scenarios that might throw it off (syncing my subs designed for the stereo mixes with Puggo's mono mixes), but under most circumstances I imagine that's not really an issue.

Might be worth a shot the next time I need to do a resync.  Ultimately I'd use to just produce conversion data for my scripts rather than using Sushi directly, because my scripts convert graphical subs as well as text subs, Sushi seems text-only.