Weird combing artifacts

[dansrfe]

I have tried various options in tfm along with tdecimate to properly ivtc this but I still can't get chroma combing properly deinterlaced and aligned.

vob cut: here (http://www.sendspace.com/file/nromrw)

[manono]

TFM(Chroma=True,CThresh=8)
TDecimate()

Seems to take care of most of it. Lower the CThresh a bit more to make it even more sensitive. Have you already tried turning on Chroma and tweaking the CThresh? Just be thankful it's hard telecined and not field-blended like so much of the Shemaroo garbage. But it's still utter crap.

[jmac698]

but there are field blends there - I was just checking to see if it was DOCI.
It looks like the combed frames in a pulldown sequence were blended together.
There's a simple subtraction to fix that.

[manono]

but there are field blends there
No there aren't. Just the chroma lagging a field behind in a lot of the fields. Not the same thing as field-blending.

[dansrfe]

I tried cthresh and chroma but I guess I misinterpreted the manual's explanation of the chroma option. I put it on false thinking that it would mess up field matching. Also I do believe there are blends as well but yadif and srestore doesn't really fix them appropriately.

EDIT: I also see some ghosting.

[jmac698]

It's blended an you need exblend http://forum.doom9.org/showthread.php?t=157337

DirectShowSource("C:\Documents and Settings\me\My Documents\Downloads\cut.vob")
exblend

gives you progressive fixed frames at 29.97fps

[manono]

If it were field-blended it would show an entirely different progressive/interlaced pattern. This shows the standard evidence of hard telecine, the 3 progressive followed by 2 interlaced pattern. The chroma problems are in the 2 interlaced fields. There's no field blending. Exblend (and FixBlendIVTC and SRestore(omode="pp3")) is for those instances where you have hard telecine that's been blend deinterlaced. That's not the case here.
DirectShowSource("C:\Documents and Settings\me\My Documents\Downloads\cut.vob")
exblend

gives you progressive fixed frames at 29.97fps

All it gives me is a complete mess.

[dansrfe]

Yes, tfm(chroma=true, cthresh=8).tdecimate gives me a perfect output. It was just a screwed up hard telecine I guess. Standard Shemaroo garbage.

[StainlessS]

DirectShowSource("C:\Documents and Settings\me\My Documents\Downloads\cut.vob")
exblend


On this m2v clip (not a vob) Using DirectShowSource() on my
machine does some kind of blend deinterlacing, whereas
Mpeg2Source() displays combing. (Exactly what it does
depends upon which MPEG2 decoder is used by directShow).


gives you progressive fixed frames at 29.97fps

Dont think so, ExBlend(), recovers from source that looks like
completely progressive but with blended frames ie,
Film -> 3:2 pulldown, downsized vertical to 1/2 height (eg
VCD). The filter should not work correctly, on clips exhibiting
combing of any kind, as it happens, it does seem to be detecting
the combed frame pairs as blends but 'unblending' these pairs will
produce rubbish. I did (some time ago) implement vertical average
of pairs of samples in exblend to avoid noise, it seems however to
be ignoring the combing because of this. Maybe I have to add some
kind of detector to avoid detection of these combed pairs, it does
however give me reason to look further into this to see if it could
become useful for un-combing and perhaps even blended fields
recovery. I have to gather some 3:2 pulldown source on which to test,
(I tend to bin 3:2 source if I have the misfortune to come across it).