PAL DVD Ghosting

[Skiller]

The frame rate would be at 23.976 after SRestore() as suggested by manono. It's just not because pub used SRestore(frate=25) which is not appropriate.

[jagabo]

The frame rate would be at 23.976 after SRestore() as suggested by manono. It's just not because pub used SRestore(frate=25) which is not appropriate.

Yes, I reviewed the thread and updated my post earlier.

[pub]

I got my hands on a R1 DVD (s2d2) and it looks and sounds the same as the output from my script and commandline. If I remove frate=25 and do not use tsmuxer it looks and sounds like my R2 DVD.

AssumeFPS didn't work, it stays at 23.976 fps

See my attachments for the 29.97 fps dvd source, 25 fps dvd source and my finelooking output

[jagabo]

25 PAL DVD SOURCE.mkv only needs TFM().AssumeFPS(24000,1001).

[pub]

25 PAL DVD SOURCE.mkv only needs TFM().AssumeFPS(24000,1001).

Instead of QTGMC, SRestore, or both?

[Skiller]

Instead of both.


25 PAL DVD SOURCE.mkv is not a fieldblended normconversion, it is indeed a speedup. It somehow got it's fields shifted but a simple field match, such as TFM, fixes this perfectly. There is nothing to deinterlace (QTGMC) nor to deblend (SRestore).

So a slowdown from 25 fps to 23.976 fps would indeed restore the original speed.

But I wonder, why were we discussing restoring a fieldblended source, and posting scripts when it's clearly not fieldblended?

[manono]

I wanted to get rid of the 4% speedup done by the NTSC->PAL conversion.

As mentioned, there is no speedup with your source. That happens when film is speeded up to 25fps. This type of conversion keeps the original film length while poorly converting it to 25fps.

SRestore works best when used for the correct source framerate, in this case its default 23.976fps. By doing what you did you actually wound up with a video at slower than film speed and out of tune audio - slightly deeper than it was supposed to be. It doesn't need to be slowed down, just as the video doesn't need to be slowed down.

But I wonder, why were we discussing restoring a fieldblended source, and posting scripts when it's clealy not fieldblended?

Because the sample provided in the very first post in this thread is field-blended.

[Skiller]

Well, the latest sample isn't. So that makes sense.

Can't use the same workflow for both.

[pub]

@manono So what you're saying is the NTSC dvd is slowed down and the PAL dvd has the correct length and audio pitch?

If I use TFM().AssumeFPS(24000,1001), I get the same audio length and pitch as the output from my SRestore frate 25+tsmuxer slowdown and the output from the NTSC DVD, but I still have some fieldblending

[Skiller]

You need to differentiate between the field blending normconversion as seen in your first sample (post #4), and the latter sample. It appears the PAL releases of the series were not treated equally across the seasons or maybe even episodes. Normconversions need deinterlacing and deblending but no speed correction, while speedup conversions need TFM and optionally a slowdown.

[manono]

You need to differentiate between the field blending normconversion as seen in your first sample (post #4), and the latter sample.

Right, I apologize for saying the samples were in the first post. I did a quick check when writing before and saw links and quickly assumed they were the samples when they weren't.

Like you, I don't know if the samples are from the same episode, different episodes within the same DVD, or from different DVDs altogether.

@manono So what you're saying is the NTSC dvd is slowed down and the PAL dvd has the correct length and audio pitch?

I'm saying both the NTSC and field-blended PAL versions should have the same lengths and audio pitch.

[Skiller]

I'm saying both the NTSC and field-blended PAL versions should have the same lengths and audio pitch.

Correct.