Pulldown of 25 fps to NTSC

[ad25]

I have an interlaced PAL video that I would like to create an NTSC DVD from. I just want to confirm my understanding from reading through the thread.

1. Use Avisynth to deinterlace and resize to 720x480
2. Encode to a 720x480 25fps progressive MPEG-2 file
3. Use DGPulldown to flag the 25fps file to 29.97fps
4. Author DVD

Is there anything I'm missing? Thanks in advance.

[manolito]

Everything about your procedure looks alright to me. (I think you know that you have to demux your video before step 1)

Just a little advertising here: If you happen to use DVD2SVCD, have a look at http://forum.doom9.org/showthread.ph...974#post971974
This is an automated procedure to convert PAL to NTSC with the additional bonus that the chapter points you might already have in your PAL video will also be converted.

Cheers
manolito

[Xesdeeni]

Originally Posted by ad25

I have an interlaced PAL video that I would like to create an NTSC DVD from. I just want to confirm my understanding from reading through the thread.

1. Use Avisynth to deinterlace and resize to 720x480
2. Encode to a 720x480 25fps progressive MPEG-2 file
3. Use DGPulldown to flag the 25fps file to 29.97fps
4. Author DVD

Is there anything I'm missing? Thanks in advance.

If it's truly interlaced (each of the 50 fields contains motion), I wouldn't recommend that process, which is best for progressive frame sources. You would be converting 50 "instants of time" to 25. Instead, I'd recommend that you:
1. Use AVISynth and a plugin (LeakKernelBob, SmoothDeinterlacer, etc.) to deinterlace the video to 50 (progressive) Fps.
2. Scale the video to 720x480.
3. Use ChangeFPS() or ConvertFPS() to convert/change the frame rate to 59.94 (progressive) Fps. Which one you use depends on your personal preference. ConvertFPS() blends frames. ChangeFPS() replicates frames. I prefer ChangeFPS().
4. Divide the frames into fields and throw out every other one (SeparateFields().SelectEvery(4,0,3) or SeparateFields().SelectEvery(4,1,2), depending on your destination field polarity).
5. Encode to a 720x480@29.97 interlaced Fps MPEG-2.
6. Author DVD.

Xesdeeni

[ad25]

Conclusion:

1. DGPulldown method for Progressive PAL to NTSC.
2. ChangeFPS/ConvertFPS method for Interlaced PAL to Interlaced NTSC.

Thanks for the replies.

[Guest]

We don't make blanket statements about "best" around here.

[ad25]

Sorry about that. I have reworded my reply.

[laserfan]

I resurrected this old thread because I think Xesdeeni and DG aka neuron2 have done a wonderful thing creating the DGPulldown tool. Hasn't gotten enough pub IMHO and it surely deserves it (never was a "DGPulldown" thread created per se, was there?). Anyway...

I bought a pricey DVD set from Sweden (PAL) which I did not want to convert to 720x480 because I thought my media players (networked ShowCenters) would play the original ripped 25fps MPEG2. Turns out at least my analog NTSC TV didn't like the original much (jerky, prolly had something to do with 25x2 vs 29.97x2 ). Anyway, a simple DGPulldown from 25-->29.97 did the trick.

Amazingly, I was able to re-mux (using IfoEdit that didn't care this was non-standard stuff) the original 720x576 and 25fps AC3 audio with a brand-new subtitle track I made for it and it all plays beautifully! Just want to say THANKS you guys especially neuron2 for exe-cuting this!

[Guest]

I've never heard of anyone doing that before. Thanks for bringing it to our attention, and thank you for your kind words.

[SeeMoreDigital]

I actually did some tests a few years ago with (PAL) 720x576 MPEG-2 sources running at (NTSC) 29.970fps and 23.976fps speeds.... And vice-versa.

From what I remember, most, if not all my DVD players managed to puke out an NTSC signal... which was a positive step.

I was however unable to find a DVD authoring tool that would allow me to create "non DVD standard" VOB sets suitable for burning onto DVD.


Cheers

EDIT: Found a thread: http://forum.doom9.org/showthread.php?t=83499

[laserfan]

Originally Posted by SeeMoreDigital

From what I remember, most, if not all my DVD players managed to puke out an NTSC signal... I was however unable to find a DVD authoring tool that would allow me to create "non DVD standard" VOB sets...

Well, as I said IfoEdit worked for me. I had intended to sacrifice a DVD+R to try my VIDEO_TS with, and based on your post I have a better-than-zero chance of actually seeing it work on any of my NTSC players! I'll try it today!

[laserfan]

Well, the result plays smoothly on the one Sony (NTSC) DVD player I've tried, but it isn't displaying at the proper AR i.e. it's taller than it should be. It looks like, since the player's been told this is a 720x480 NTSC picture, and it's not, it's 720x576, that it's "too tall" at least on my 4:3 analog TV. Haven't tried it yet on my 16:9, I guess that's next.

But it looks like the DVD has to be either a PAL at 720x576 or an NTSC at 720x480, and since it's now a 29.97 program IfoEdit (authoring proggie) decided "NTSC". I suppose if I change it to PAL then my player will balk.

Will play with this some more...glad at least my media players don't care that it's 720x576 (and display at proper AR).

[laserfan]

Got 'round to trying my DGPulldown DVD (a 720x576 PAL disc that I didn't re-encode) on my 16x9 setup last night and it worked! The disc plays as if it were NTSC, but it's 576, and the AR is A-OK.

One oddity is that the DVD player's display has an odd staccato cadence about it i.e. is not at all smooth, but the picture (25fps to 29.97fps) is smooth and good.

I have another PAL disc on-the-way and will use this same method to play from my upconverting HD-XA2.

[DoctorM]

I've been using the 25 to 29.97 pulldown method for awhile and have been very pleased with the results (better than most commercial conversions if you ask me).

I'm working on a full DVD conversion now and saw a couple of strange things this time.
First, the DVD is 1:54:57 in length. While the untouched audio in the NTSC version syncs throughout, the runtime (checked in a couple different pieces of software) shows 1:54:50.

Which brings me to the second problem. I would disregard the runtime discrepancy since there is no sync issue, but the converted chapters are off. The theoretical calculation for conversion would be: 30/25 * original cell.

And this is true. Examining the time of each chapter on the NTSC conversion done this way matches the PAL original. The times also show correctly when using any software that shows the time points of the chapters (ChapterXtractor, DVDRemake, etc.)

BUT, while the times are right, the actual frames drift progressively until they are 7 seconds off at the end.
I can account for this by changing the cell calculation to: 29.97/25 * original cell. Calculated this way the chapter breaks appear in the correct location, but the actual timestamp is wrong.

Has anyone ever noticed this before? I can't make heads or tails of it.

[Guest]

Let's address the running time first. How did you create the M2V stream that you applied pulldown to from the DVD? What are the exact steps that produced it? What is the running time of that M2V before and after applying pulldown?

[DoctorM]

Originally Posted by neuron2

Let's address the running time first. How did you create the M2V stream that you applied pulldown to from the DVD? What are the exact steps that produced it? What is the running time of that M2V before and after applying pulldown?

The source was 25i. Using AVISynth, Yadif(), and some other clean up filters, I converted to 25p.
The video was encoded with CCE 2.70 with the Frame Rate set to 25.

Before pulldown the runtime shows 1:54:57.320 by VDubMod (1:54:56 according to DVD Lab Pro).
After applying 25 to 29,97, the runtime is still 1:54:56 (also by DVD Lab Pro) (VDubMod shows around 1:36:ish.).

Test builds of the DVD with MuxMan and DVD-lab Pro made no difference.
The runtime is 1:54:50 in Media Player Classic and VLC.
IfoEdit shows 1:54:50.13.
PowerDVD does show the correct 1:54:57 time but it could be ignoring the pulldown flags.

I tested on 3 stand alone DVD players.
A very old high end Sony, a few year old mid range Panasonic and a mid range 1 year old Philips.
All 3 agree that the runtime is 1:54:50.

29.97/25 * original cell still gives the correct chapters by frame but at a different time point than the original.

Which all seems to make sense since 1:54:57 / 30 * 29.97 = 1:54:50.

[Guest]

Looks like pulldown is irrelevant and you should be posting this elsewhere.

[DoctorM]

Originally Posted by neuron2

Looks like pulldown is irrelevant and you should be posting this elsewhere.

Perhaps that is true, but the issue comes from the 25 to 29.97 pulldown method being used to make an NTSC DVD. I'm sure I can't be the first person to notice this (although I may be the first one to care).

Does DGIndex default to non-drop frame in these cases when it should be drop frame?

[Guest]

Originally Posted by DoctorM

Does DGIndex default to non-drop frame in these cases when it should be drop frame?

Why are you talking about DGIndex here?

Try setting the drop frames explicitly by clicking until it is checked but not gray.

I never do any chapter stuff so don't really know what you are talking about. If you can identify a problem with DGPulldown and tell me what needs to be done I will do it, but I have no time to get all wound up in chapters and authoring right now.

[DoctorM]

Originally Posted by neuron2

Why are you talking about DGIndex here?

Try setting the drop frames explicitly by clicking until it is checked but not gray.

I never do any chapter stuff so don't really know what you are talking about. If you can identify a problem with DGPulldown and tell me what needs to be done I will do it, but I have no time to get all wound up in chapters and authoring right now.

Sorry, I mis-typed. I did mean DGPulldown not DGIndex.
The chapters aside, I just worry when the runtime changes when it isn't suppose to. I'll test with drop frames and see if it corrects that.

Edit: No, it's worse with DF. The audio desyncs and the runtime becomes 1:59:44. I guess the change in runtime is just a quirk associated with doing this. Thanks for the help.

[Guest]

You said the runtime before and after pulldown was the same. So why would that be a DGPulldown issue?