VOB sequence is 2frames 3frames ...

[pooksahib]

When preparing a DVD for edit and it has combing I first use Bob() and if there is motion with every field I know it's interlaced and if there's motion every two fields I know it's progressive. The DVD in this case has combing and after applying Bob() it gives me two identical frames then three identical frames (repeated). What do I have here and what's the way to handle it? Thanks.

[Sharc]

It is 2-3 telecined. You should IVTC (Inverse Telecine) rather than deinterlace the video to get the original progressive film frames back.

In Avisynth:

TFM().TDecimate()

or use a GUI which has IVTC as an option.

If it is soft telecined your player has just to ignore the pulldown flags.

https://en.wikipedia.org/wiki/Telecine

[pooksahib]

Great stuff, thanks, Sharc. Am I right in saying that if the DVD was PAL you'd omit TDecimate() from the code?

[Sharc]

Great stuff, thanks, Sharc. Am I right in saying that if the DVD was PAL you'd omit TDecimate() from the code?

3 cases:

(a) Speed up (runs slightly faster, 25/24, 4.2%)
Films on PAL DVDs are often just speed up from 24fps to 25fps. Encoded and flagged as interlaced for PAL DVD spec compliance, but the video is actually progressive and can be processed as progressive because both fields are from the same film frame (same moment in time)

(b) Telecine (keeps original run time)
If telecine is used for the speed conversion 24fps -> 25fps it will be using 2:2:2:2:2:2:2:2:2:2:2:3 pulldown (aka "Euro pulldown"). It can be IVTCed by

TFM().TDecimate(cycle=25)

to return the original 24fps progressive film frames.

(c) Phase shifted
25fps interlaced with the 2 fields taken from subsequent film frames.
For decoding just use

TFM()

which reassembles the fields belonging to the same film frame


http://avisynth.nl/index.php/TIVTC

[lollo]

(c) Phase shifted
25fps interlaced with the 2 fields taken from subsequent film frames.
For decoding just use

TFM()

which reassembles the fields belonging to the same film frame

Just a small addition to the excellent Sharc's reply.

TFM sometimes fails for phase-shifted / not phase shifted material where there is a very very small amount of movement (for instance static end credits), choosing the wrong field. In this case is better to use:

DoubleWeave().SelectOdd()

but the phase shift must be constant.

If the material has not constant / not always present phase shift, the reccomended TFM(order=1, PP=0) is more appropriate as Sharc suggested.

[pooksahib]

Thanks guys.