Serial Experiments Lain challenge - revisited

[igro]

I got my hands on some DVDs of Lain from Japan (long story short - it pays to scare the hell out of your Japanese neighbours by greeting them in Japanese and then start talking some more ;)). I've heard that Lain was a bitch to encode (due to the crappy R1 source and what not), but at the same time it seemed like a perfect candidate for a vfr encode, so I wanted to give it a try.

The source seems pretty good to me (considering this is not a new show, etc) - no haloing/whatever, but I have some serious trouble getting to deinterlace it.

Here's a sample: http://www.mediafire.com/?mlxzmjx29qz

For most of the eps, everything seems to go ok: it detects film and video sections, everything is smooth, but look at the above sample from episode 11, especially these two parts:

http://xs135.xs.to/xs135/09045/snapshot20090123195731775.png (notice the obvious combing on the mouth)

http://xs135.xs.to/xs135/09045/snapshot20090123195853735.png (notice the not-so-obvious combing on the mouth ;))

No matter what I try I can't get these two parts right.
My script is pretty simple, based on the vfr examples from the tivtc documentation, NNEDI + TDeint.
I've tried many options of tfm, tdeint, but to no avail :( If someone could point me in the right direction, (even "RTFM for x filter" ;)) say if this can even be done, I'll be very grateful :)

[Nightshiver]

Well, I can suggest animeivtc, as it is rather simple to do vfr with. You would use mode=5,pass=1 for the first set, load it into vdub, run analysis pass, then change the script to mode=5,pass=2 and do another.

[igro]

I wanted to avoid using animeivtc, since, from my experiences in the past with it, it tends to blur things and do weird things with lines (but it might have been my fault - wrong parameters, etc). Since I haven't got any other idea atm, I'll give it a try :P

[kemuri-_9]

i ran the sample through YATTA briefly and the first scene had no possible matches that didn't end up requiring PP blending.

however the mouthflap scene did have matches that didn't require PP blending.

[igro]

So in order to blend those problematic frames I would have to use yatta? (or write the ranges myself ;)) Or is there a magical way to tell the universe - blend those damn frames (and still get a vfr encode)?

BTW, I tried animeivtc, not better than TIVTC, or rather worse due to the grain/noise/details-killing (pick what you see :P).

[kemuri-_9]

YATTA's PP (blending) are standard deinterlacers...
the ones i frequently choose between is TDeint and FieldDeinterlace (from Decomb)
both of them have override (ovr) options to only deinterlace frames you specify from a file.

[Nightshiver]

What settings are you using that are so killing the detail? A simple vfr with no aa shouldn't degrade quality in the slightest.

[igro]

@ kemuri-_9: thanks for the tip, I read about the ovr paramter in the tdeint documentation but didn't bother to understand it >_<;

@ Nightshiver: heh, you were right, I checked frame to frame, and animeivtc(mode=5,pass=1,precision=2,aa=0) (the way I've encoded it before) doesn't hurt the details at all >_< Probably, the last time I used it, I had one of the aa modes on. But the fact that animeivtc didn't help still holds :D

[Nightshiver]

How did it not help?

[igro]

It produced a 23.976 clip with the same (more or less, I wasn't counting the actual number ;)) artifacts as tdeint + nnedi + tfm @ defaults.