Deinterlacing Trouble (FieldBlended ??)

[chandragupta]

Hi,
i have a clip and i'm unsure of what kind of interlace is on it.
from what i have been reading here, i feel it is fieldblended...
I have uploaded the clip here : Clip (http://rapidshare.com/files/190288166/clip.m2v.html).

i have good result with the following script:
MPEG2Source("C:\DVD\bbh.d2v", cpu=0, info = 3)
colormatrix(hints=true)

d = last.bob(-0.2,0.6).ReduceFlicker()
Interp = nnedi(field=2)
TDeint(mode = 1, order=0,field=0,type=1,edeint=Interp,emask=TMM(mode = 1, order=0,field=0))
srestore(dclip = d. crop(6,44,-6,-50))

crop(6,44,-6,-50)
spline36resize(704, 384)

please tell what kind of clip it is as i'm unsure if this the correct way of deinterlacing this....
is the above script fine??
also are there any other better ways. (excluding mcbob)
Thanks!

[manono]

It's the usual Indian (and Moserbaer) PAL2NTSC field-blended trash. Your script does a good (if slow) job on it.
also are there any other better ways.
Yadif(Order=0,Mode=1)#the sample is BFF
RePAL()

Don't know if it's any better, but it's simpler and faster.

Depending on your output format, you probably don't need the ColorMatrix filter. I don't know anything about the ReduceFlicker filter and have never used it, but the docs say to use it on the interlaced source (before bobbing).

[thetoof]

This usage of reduceflicker comes from srestore's docs for a good & fast dclip.

[chandragupta]

hi manono, thanks for the reply!
It's the usual Indian (and Moserbaer) PAL2NTSC field-blended trash.
yes, almost every DVD has field-blending here..dunno why they produce this crap, and that too everyone.

Yadif(Order=0,Mode=1)#the sample is BFF
RePAL()
i read some posts on RePal and others and tried a few things :

1.)
Yadif(Order=0,Mode=1)
RePAL()
Result (http://www.imagebam.com/image/0a673424917002)

2.)
d = last.bob(-0.2,0.6).ReduceFlicker()
Interp = nnedi(field=2)
TDeint(mode = 1, order=0,field=0,type=1,edeint=Interp,emask=TMM(mode = 1, order=0,field=0))
RePAL()
Result (http://www.imagebam.com/image/b9f26e24916983)

3.)
KernelBob(4)
RePAL()
Result (http://www.imagebam.com/image/26ab3b24916941)

4.)
nnedi(field=2)
RePAL()
Result (http://www.imagebam.com/image/42e28624916963)

5.)
d = last.bob(-0.2,0.6).ReduceFlicker()
Interp = nnedi(field=2)
TDeint(mode = 1, order=0,field=0,type=1,edeint=Interp,emask=TMM(mode = 1, order=0,field=0))
srestore(dclip = d. crop(6,44,-6,-50))
Result (http://www.imagebam.com/image/3df9ad24916992)

6.)
assumebff()
kernelbob(4)
unblendpattern(1,25,false,false)
Result (http://www.imagebam.com/image/df20c524916938)

repal gave almost similar results to srestore.
but this unblendpattern() is real good. It gave better results than 2.) and 5.) and is faster, but tedious.

will it be possible to get film (23.976) from this without having to touch the audio, or is it made at film and then sped up to 25fps -> field-blended 29.976fps??

thanks for ur time !

[manono]

Hi-

Your sample was BFF, but maybe the main movie itself is TFF. It's important to know the field order when unblending. Also, while I usually use RePAL, there's nothing wrong with using SRestore for the same thing:

Yadif(Order=0,Mode=1)
SRestore()

which gives 25fps at default settings. Yes, I used to use scharfi's UnblendPattern quite a bit, before the unblenders were available. No more, though. As you said, it's tedious. It's too much like work.
will it be possible to get film (23.976) from this without having to touch the audio, or is it made at film and then sped up to 25fps -> field-blended 29.976fps??
I do it quite a bit, slowing the audio and video to 23.976fps. No, since it's on the DVD at 24.975 or 25fps, unless you resort to blending with ConvertFPS (not such a good idea), you're stuck with the PAL framerate unless you're willing to reencode the audio. They used a 25fps PAL master for the NTSC DVD. It's saves them from having to spend money to make a proper NTSC master. It's a simple matter, of course, for us on our computers to go from 25->29.97fps without field-blending (and even to slow it to the original 24fps), but I guess the hardware standards conversion boxes they use can't do it.

[chandragupta]

Thanks for all the info you provided manono,
really helpful.
:thanks: