[HELP] Reverse-Telecine 29.97 vob file

[Miranda]

What is the best setting to de-interlace and de-telecine to 23.976?

[Selur]

There is no 'best' method to do deinterlacing or ivtc every source.

To convert properly 29.97 (30000/1001) telecined content to 23.976 (24000/1001) progressive frames any filter that does IVTC should work fine.
Problem is one often encounters content that isn't properly telecined, you often encounter mixed content, content with pattern shifting etc.

In general I would recommand to look at the preview and try:
1. TIVTC
2. VIVTC
3. QTGMC in bob-mode + sRestore
if none of these work directly you can try to tweak these methods and see if one of them helps or bob your content and live witht he dublicates, or use another tool.
There are a few tools designed for 'funky' ivtc decimation like Wobbly (https://github.com/dubhater/Wobbly), Yatta (https://github.com/hatt/Yatta) , AnimeIVTC, pvsfunc (https://github.com/rlaphoenix/pvsfunc) and there are also some scripts and code snippets some users use.

All in all there is no magic tool which automatically produces the best results.
Test way to IVTC content is to read up on what it is one wants to do, properly analyse the source (looking at field and bobbed frame to detect patterns and problems with the source), then decide whether one of the methods mentioned can help, whether one needs to combine some of them or whether some custom written methods are needed.

Cu Selur

[Miranda]

Hybrid says it does not detect telecine. My guess is it was originally 23.976 but it has been edited on a 23.97 timeline.

[Selur]

Hybrid says it does not detect telecine.

That does not mean it's correct.

Hybrid has two methods to automatically decide what content the source is:
a. read the headers using MediaInfo (which can be wrong) (+ some additional logic I just rewrote a few weeks ago)
b. run some heuristical tests on the source (which can be wrong; and I long gave up on tweaking them, they are based on https://github.com/Selur/MPlayerInterlaceDetection)
-> Only way to know for sure is to understand how to analyse and decide for your own that your source content is and overwrite the scan order appropriately + adjust the settings Hybrid uses. Since you open a vob and not a DVD source the flags of the content might be totally wrong and the heuristics might simply fail.

You also might want to read:
https://en.wikipedia.org/wiki/Telecine
https://forum.selur.net/thread-1961.html
https://forum.selur.net/thread-1927.html
https://forum.doom9.org/showthread.php?t=176104
and also do some additional reading to understand what is happening, what you can do with Hybrid and what other methods might be possible.

Cu Selur

[Miranda]

I'm seeing:

1-2-3- frames unique / 1 dupe / 1-2-3- frames unique / 1 dupe

Sometimes 1-2 frames then a duplicate

With random interlaced fields.

The original source would have been film transferred to 25fps, then converted to NTSC 23.97.

[Selur]

Sounds like there is no fixed pattern so, QTGMC in bob-mode + sRestore is probably a good automated approach if you don't like the result you have to try tweaking the settings, use one of the tools I mentioned or come up with an alternative way.

The original source would have been film transferred to 25fps, then converted to NTSC 23.97.

There are tons of ways progressice/interlaced 25fps to 29.97 telecined could have be done, the pdf linked over at https://forum.doom9.org/showthread.php?t=176104 mentioned that and different ways to deal with it. (not with Hybrid, but in general and you might need to understand quite a bit of Avisynth and what is happening if you want to use the methods mentioned there)

Cu Selur