Telecined DVDs; why flags instead of player handling it all?

[shae]

Why do film DVDs rely on flags instead of just letting the player do the 3:2 pulldown?

[mpucoder]

I don't understand your question, or maybe you don't understand how pulldown works. The flags tell the decoder (player) how to perform the 3:2 pulldown.

[neuron2]

I think he means if it was coded as 23.976 without flags, the player could determine that since the output display device is 29.97, it has to apply 3:2 pulldown, without being told anything.

The problem is that video can be hybrid, with video and film intermixed. At a minimum you need to be able to flag the sections somehow. The mechanism of pulldown flags is a good general way of doing it, as evidenced for example by the ability to do things like use pulldown to display PAL at the NTSC rate.

Now one could get silly and say the player can use heuristics to distinguish film from video, but the idea fails because a static video scene could be mistaken as film and displayed at the wrong rate. Determinacy is really desirable to ensure that all players render the video the same way.

[mpucoder]

And the pulldown flags can be used in other ways. I have seen segments such as "This preview has been rated..." use the flags to reduce the number of encoded pictures (rff set for each picture requires only 20 pictures per second).

The flags seem to me the most flexible and economical way of handling pulldown. Without the flags in order to get the proper fields repeated, and that can be important, would be to encode a timestamp for each picture. And you still would have to indicate field dominance somehow. Instead we have a timestamp for each group of pictures, and just 2 bits per picture to tell the decoder how to obtain the target framerate from the pictures, and the proper field order.

[shae]

Yes, I guess it is useful for mixed. Too bad DVDs can do more things than I want them to (which is hold clean progressive video) and consequentially provide more pitfalls for incompetent producers. :)

What is the reason for problems with IVTC, BTW? If it's pure film, as it should be, the flags could be ignored completely, no? How can editing, regardless of the flags, do any harm to progressive frames (which I assume is what's used for every film DVD from any marginally reputable company)?

[neuron2]

Not all material is soft telecined, a lot is hard telecined and subsequently edited. That makes IVTC a heuristic process. Yes, if you have pure soft telecine you can just ignore the flags, which is what Force Film does in DGIndex (well, almost, that mode also tries to keep video portions running at the same speed by decimating them, etc.).

[shae]

You mean post IVTC frames with combing are always hard telecined? If yes, why would there be a few hard TCed frames in the middle of a soft TCed film?

And while these troublesome frames are typically on scene changes, it's is not always the case. I've also encountered it, on occasion, in the very middle of shots. This seems like a less likely place for editing, so what might that be?

[neuron2]

I'd have to see the source. Wanna post a stream?

[shae]

Can't do that right now, but maybe a specific example will help; a somewhat recent movie (maybe a common one) I remember with this problem was Silent Hill.

Half way through (I'm pretty sure it's a few minutes ± from the very middle), there's a scene with people running thru a graveyard to a church, and there, in the middle of a shot, artifacts after plain IVTC.