Accurately ripping chapters/celltimes

[DoctorM]

I'm working on an atypical PAL2NTSC project.
As part of it I'm shifting the chapter stops to match the NTSC DVD. (Don't ask why, it's crazy and it's unimportant with regard to my problem.)

I've run into a problem after ripping the celltimes from the NTSC disc.
I have tried PGCDemux, IFOEdit, ChapterXtractor and DVD Decrypter. They seem to agree in their numbers but the actual chapters towards the end get further off from where they should be. Towards the end it's more than a second.
(Ok, not completely true, DVD Decrypter gets different 'CCE Cell Information' from the rest (and further wrong), but it's Chapter Information times match).

What I'm doing is using the above software to get the celltimes. I then use the NTSC video loaded into VirtualDubMod to find specific frames (yes, I've removed pulldown flags).

I made a list of converted celltimes divided by 1.25 to account for the pulldown. I then jump to these frames in VDubMod.
The time indicator in VDubMod agrees it's the correct location (the times that IFOEdit, ChapterXtractor, DVD Decrypter list). Unfortunately, this is NOT where the chapters actually are when the DVD is played back.

Like I said, it gets further and further off the mark and towards the end I start getting a second or more of the previous scene.
It's sloppy and I don't like it.

Any suggestions? What would make these programs come up with the wrong celltimes/chapters points?

[mpucoder]

It's neither sloppy or incorrect. NTSC DVD timecodes are in non-drop form. Simply put this is like having the video at 30fps, and makes it easy to get the exact frame number. But NTSC video actually runs at 29.97fps, which is where your drift is coming from. Try multiplying by 1001/1000 to get the actual time from a timecode.

[Video Dude]

I had to do this a few times.

There are a number of freeware timecode calculator / converter programs that can automate what mpucoder suggested above and do the whole pal to ntsc timecode conversion in one step.

[DoctorM]

Any links Video Dude?

[Video Dude]

I believe these are the two I had the most luck with. Both are freeware.


http://www.mediachance.com/dvdlab/freetime.htm

Direct Link:
http://www.mediachance.com/files/WFTimeSetup.exe



http://www.paul-pelletier.com/TimeCodeConverter/TimeCodeConverterPage.htm



Edit: I had found an Excel spreadsheet that could batch, but I can't find the link right now.

[DoctorM]

Hmm. I'm still having trouble arriving at what should be the right number.
For example the chapter that I'm told is at 241101 should be around 192906 (at 23.976fps).
No matter how I shuffle the numbers I come up further off the other direction.

241101*1001/1000/1.25=193074 (Originally I was getting 192881.)

I really seem to be off by a factor of 1.0001-ish instead of 1.001. Thoughts on how I can arrive at that?

Edit: Thanks for the links, I'll give it a go.

Edit 2:
No luck with the calculators.

It's neither sloppy or incorrect.
My comment wasn't regarding the programs. My finished project was sloppy and incorrect.

[nixo]

I rarely do NTSC stuff but ran into the same problem with ChapterXtractor.
What seemed to get me correct chapters was choosing a frame-based chapter format in ChapterXtractor - Ogg-style chapters won't work - and change the fps to 24, not 23.976.

[DoctorM]

Yeah, I've seen that with ChapterXtractor as well.

Soooo. It looks like my best answer is to take the celltimes divide by 1.25 and multiply by 1.0001. Not that the number makes sense, but it does put me within a frame or two of where I really want to be.

Frankly, the 1.001 at least had sound footing, 1.0001 is just odd.

[DoctorM]

Mystery solved.
I loaded the movie in DVDRemake Pro and cell by cell manually located the first frame and compared it to my AVS script in VDub.

For more than half-way through the frames are accurate 70-80% of the time, give or take 1 frame. That's easily rounding error.

Here's the kicker, just past half way there is a scene that's a bit screwed up. The frame rate shifts to 29.97 interlaced for about 2 seconds. Just enough to bump everything by 18 frames.
This isn't some animated or special effect scene, it's just that somebody screwed up in the transfer.

It happens a second time a few chapters from the end, pushing the chapters off by 27 frames.

So while hardware/software players (and DVDRemake Pro!) have no problems. Once you frameserve it, all bets are off.

Which makes sense when you're talking variable framerates.
So what do you do? Just what I did. Locate the chapters manually.

Hope this info helps others.