PAL to NTSC question

[rayden54]

I've been using the Pulldown method illustrated here to do PAL to NTSC conversions, but I've got a few questions.

First, what would happen if I demuxed the DVD, and used DGPulldown on the original m2v file without re-encoding? I'm guessing it won't work because the video's the wrong size (720x576 @ 29.97 FPS). Muxman doesn't appear to accept it anyway.

Second, the source appears to be interlaced. Does it have to be deinterlaced or can I just leave it alone?

Please don't suggest getting a PAL capable DVD player. The whole point of me doing this is to try to make discs that will play on other people's players.

[jman98]

1) Don't know what would happen.
2) You don't HAVE to deinterlace. To deinterlace requires you to re-encode and at that point you might as well just do a proper conversion instead of trying the shortcut method you are using to avoid re-encoding.
3) There are no guarantees that this method will produce a disc that all NTSC DVD players can play. Reports are that it works most of the time, but I'm sure players exist that won't touch such discs.

[jagabo]

If your source is really interlaced (as opposed to progressive frames encoded in interlaced mode) you have to deinterlace before using the DgPulldown method.

[rayden54]

If your source is really interlaced (as opposed to progressive frames encoded in interlaced mode) you have to deinterlace before using the DgPulldown method.

How can you tell the difference?

[jagabo]

If your source is really interlaced (as opposed to progressive frames encoded in interlaced mode) you have to deinterlace before using the DgPulldown method.

How can you tell the difference?

Do you see comb artifacts when there is motion? If not, the frames are progressive. If you do see comb artifacts apply a bob filter. Is each field different? If so you have interlaced video. If each pair of files is the same you have out of phase PAL and can restore the progressive frames.

[rayden54]

I know there are comb artifacts. Then I'm a bit lost.

I've loaded the demuxed m2v file into VirtualDubMod, and tried applying a Bob filter. I'm not really sure what settings to use or what to look for next.

Sorry, I'm a bit of a n00b. I've tried this stuff before, but I was dealing with Progressive video before...and then they finally released U.S. versions. This isn't a very popular show. It's Mortal Kombat Conquest.

[jagabo]

Try VirtualDub and its Bob Doubler filter. Or, if you know how to use AviSynth, you can apply a simple bob:

Mpeg2Source("filename.d2v")
AssumeTFF() # or AssumeBFF()
Bob()

Of just upload a small sample extracted with DgIndex.

[rayden54]

Sorry I got busy doing other things for a bit.

I tried messing with it for a bit, but I'm still not sure what I'm supposed to be looking for. I believe there's a sample attached to this post (assuming I even managed to upload the right thing).

[jagabo]

That video is out of phase PAL. It was originally progressive, broadcast as interlaced, then recorded with the opposite field order. So each frame contains fields from two different frames:

progressive frames: 1 2 3 4...
broadcast interlaced: 1b 1t 2b 2t 3b 3t 4b 4t... (bottom field first)
captured out of phase: 1t+2b 2t+3b 3t+4b 4t+5b... (top field first)

You can easily fix that in AviSynth:

Mpeg2Source()
AssumeTFF()
SeparateFields()
Trim(1,0) #discard the first field
Weave()

If the video switches between out of phase and in phase it's safer to use:

Mpeg2Source()
AssumeTFF()
TFM()

The output of those scripts is 25p. You can then resize to an NTSC frame size, encode at 25p, and use DgPulldown to add pulldown flags that tell the player how to output 59.94 fields per second.

[rayden54]

If the video switches between out of phase and in phase

Which, of course, it appears to do. I think. Using your first filter, there are still places where it looks like two interlaced images were superimposed on each other.

I did manage to find the TIVTC/TFM plugin. Should TFM be followed by TDecimate?

[jagabo]

Should TFM be followed by TDecimate?

No. TFM() will leave you with a 25 fps video. Remember you need to resize to an NTSC frame size -- 720x480 or 704x480.

Mpeg2Source()
AssumeTFF()
TFM()
LanczosResize(720,480)

[rayden54]

Awesome! Thanks very much. I was way out of my depth.

I think I'm finally starting to understand some of it now though.