Guide=1 w/ hybrid material?

[JohnMK]

I have hybrid material (half FILM, half native interlaced) and need to know if I should insert guide=1 into the Telecide() function parameter. What do you think?

NTSC, Region-1. Carl Sagan's "Cosmos" 1980.

Thanks. :)

[JohnMK]

Bumpershoot. Neuron2's here.

Well, he was anyway. :(

[neuron2]

We have to consider the effects on the film and non-film parts.

For the film parts, pattern guidance usually helps and is recommended.

Strictly speaking it is incorrect to apply field matching to real interlaced material (30fps), but in practice it doesn't seem to degrade the material very much. That is why Telecide can be applied to hybrid material with reasonable results. In my tests, pattern guidance doesn't seem to make much difference, probably because it rarely is "triggered" on 30fps material.

In summary, I think it is OK to use pattern guidance for hybrid material. But as always, in any specific case, the proof of the pudding is in the eating.

[JohnMK]

If you have badly telecined material though, such as the NTSC US release of Red Violin, I would not advise using guide=1.

Telecide(guide=1)
Decimate(mode=2) resulted in pronounced sutter in the badly telecined places (reported as NTSC)

Telecide()
Decimate() (mode=0 implied) resulted in perfect output. Either it was the different mode, or the guidance, but this was clearly better. I didn't try non-guided mode=2. Would this improve on the jerkiness, or do you think it was the guidance during these bad areas that resulted in the chopiness?

[neuron2]

@JohnMK

Yes, use guidance only when you are sure the corresponding pattern is actually being used for the telecining! For example, my hero 'manono' just sent me a clip that was telecined from 25fps to 30fps, and which produced a 32232 repeating pattern (3232 repeating is what you see for 24fps to 30fps). That looked like 4 progressive followed by 2 interlaced frames: ..PPPPIIPPPPII... Guide=1 won't work on this. So I plan to add a new guidance mode to handle it.

But in your case, John, I think it could well be the Decimate mode that is the significant factor for the stutter. When using mode=2, it is really important to get the threshold just right. One day I'll write a guide on how to do that. :p

Why not try guidance with mode=0?

[JohnMK]

That looks good.

I'm curious though, why is guide=1 necessary? Why can't decomb find whatever works, dynamically, as it's working throughout the encode? Is guide=1 still recommended for films that have badly telecined portions, such as in Red Violin? I assume it's original film material all the way through because not every frame is showing interlaced combing in motion scenes, in these what I believe to be badly telecined portions. If every unprocessed frame with motion showed interlacing, then yes I'd consider it true hybrid material.

[neuron2]

Originally posted by JohnMK
I'm curious though, why is guide=1 necessary? Why can't decomb find whatever works, dynamically, as it's working throughout the encode?I won't answer this until you clarify the question. Are you asking why there is guidance at all (versus having perfect blind matching), or are you asking why Telecide doesn't just decide to apply guidance and figure out which one automatically?
Is guide=1 still recommended for films that have badly telecined portions, such as in Red Violin?I don't know what you mean by "badly telecined".I assume it's original film material all the way through because not every frame is showing interlaced combing in motion scenes, in these what I believe to be badly telecined portions. If every unprocessed frame with motion showed interlacing, then yes I'd consider it true hybrid material. I need to see the clip to comment on this.