3:2 Pulldown and Interpolation

[PuzZLeR]

Greetings my friends, I have another question for you (and hopefully others reading this can benefit).

Yes, I've done the research regarding interlacing, pulldown, etc, and even received help here regarding this (thank you!), but I'm still a bit miffed about a couple of details. None of the guides are really telling me what the difference between this and that is. But I do have enough background now to understand if someone knowledgable can set me straight.

My question:

Can someone explain, what really happens when you convert video with "inverse 3:2 pulldown" and "3:2 pulldown playback" (and/or the difference between these and their difference with using "progressive")? And what does "interpolation" really do?

The reason I ask is in particular with TMPGEnc Xpress when I pre-convert/deinterlace MPEG-2 content or frameserve from my editor in order to properly encode to MPEG-4 (DivX, AVC). I have gotten excellent results using Inverse 3:2 pulldown with the filter "interpolation" - no "interlace-raking" and high detail preserved. No, I'm not complaining, but it would be nice to know what I'm doing.

Thanks so much again. I have always appreciated help given to me.

[edDV]

I'm not sure what TMPGEnc Xpress menus are but inverse 3:2 pulldown (aka "Inverse Telecine" or IVTC) is a straight forward concept.
http://www.dvdfile.com/news/special_report/production_a_z/3_2_pulldown.htm

"Interpolation" can mean different things depending on the context but should be unrelated to inverse telecine. Inverse telecine just reorders fields so that 23.976 fps progressive frames are restored. You start with 23.976 frames, add fields to get to 29.97 NTSC and later remove fields to come back to where you started i.e. inverse telecine.

Deinterlace is a separate process where 29.97 interlace video is converted to progressive via various techniques, none perfect. Interpolation (guessing about missing information) is one technique.

Real world telecine video often has breaks in the sequence. The inverse telecine software (or you) must have a strategy for dealing with breaks in the sequence. A fallback is to blend or other method deinterlace until the sequence is re-detected. Video gets blurry when this happens.