Hybrid DVD

[x265]

How do i deal with a hybrid source?

[x265]

http://imageshack.us/f/59/capturepuuk.png/

Code: Select all

# timecode format v1
Assume 23.976
0,744,29.9702
745,32380,23.9760
32381,34601,29.9702

<< timecodes taken from an encode of this show by a group called Figmentos.

[x265]

<< ED

<< OP

[Mister Hatt]

I wouldn't hold much value in the timecodes from a random fansub group. Let's start with a dump or screenshot of the DGIndex info box output. Run it in both honour pulldown and ignore pulldown modes.

[mirkosp]

Actually Figmentos' release is rather reliable. There are still some mistakes, since it's a long series and so some scenes went accidentally overlooked, but the encoder did properly go over the whole series in yatta, so at the very least the vfr timecodes are reliable.
Either way, screenshots like those don't help, but I've had the first DVD in my hands, and as far as I recall the only things with fuck with the metrics are the scrolling text bits. Rest is a simple 3:2 pulldown. This seems to be confirmed by the timecodes of figmentos, which just vfr the ending (the animation on the side actually is 23.976 tho, so the best solution would have been to ivtc_txt60mc, assuming the text was 60i and not 30p, otherwise try to interpolate the text on its own ─ simple enough being white text on black ─ and IVTC the image and put them together through a simple stackhorizontal) and a bit at the beginning, so just the fading text with the animation behind (they both are 29.97).

[x265]

HONOR PULLDOWN FLAGS


IGNORE PULLDOWN FLAGS

[x265]

Actually Figmentos' release is rather reliable. There are still some mistakes, since it's a long series and so some scenes went accidentally overlooked, but the encoder did properly go over the whole series in yatta, so at the very least the vfr timecodes are reliable.
Either way, screenshots like those don't help, but I've had the first DVD in my hands, and as far as I recall the only things with fuck with the metrics are the scrolling text bits. Rest is a simple 3:2 pulldown. This seems to be confirmed by the timecodes of figmentos, which just vfr the ending (the animation on the side actually is 23.976 tho, so the best solution would have been to ivtc_txt60mc, assuming the text was 60i and not 30p, otherwise try to interpolate the text on its own ─ simple enough being white text on black ─ and IVTC the image and put them together through a simple stackhorizontal) and a bit at the beginning, so just the fading text with the animation behind (they both are 29.97).

I'm confused.

[Mister Hatt]

What Mirko meant was that it isn't hybrid, just the ending credits use content with different framerates. There are two ways to handle it (or actually three but the last one is a terrible yet hilarious idea I'll leave as an exercise to the reader, think oldschool). The first is to use ivtc_txt60mc to get the framerate correct, and the second is to split it into two clips, IVTC the first as normal (the animation), and then interpolate the text to 23.976fps and stick them together again. It's really up to you but I would recommend ivtc_txt60mc because it's good, and also helps you out with the blip at the beginning thanks to the fade.

[x265]

Could you post the script for doing this?

[Mister Hatt]

It's not that difficult. You put ivtc_txt60mc() with whatever arguments you want. If you need to split, you use trim and crop and regular variables, then stackhorizontal(). I don't see the difficulty.