tivtc does not work well with fading scenes?

[kcb]

I'm using the following script:


LoadPlugin("G:\Video_Tools\DGIndex\DGDecode.dll")
LoadPlugin("G:\Video_Tools\AviSynth_Scripts\TIVTC.dll")
LoadPlugin("G:\Video_Tools\AviSynth_Scripts\Decomb.dll")
mpeg2source("G:\Video_Tools\AviSynth_Scripts\test.d2v")
TFM(order=1)
TDecimate()


so far on all DVD's i've tried ivtcing (not anime) it does not handle fading scenes properly (i.e. when the video ends).

Any ideas?

[kcb]

actually it does but the problem is with fading scenes at the end and start.

[Mug Funky]

what sort of stuff are you IVTC'ing? movies? TV episodes?

because if it's just the beginning and end, it's possible the fades were done in 60i after telecine took place. in this case, field-matching will be ineffective, and TIVTC's postprocessing might not catch it (for some reason i've yet to determine, even some of the best comb masks bugger up on fades).

[kcb]

yep i'm ivtc'ing movies, yes, its just the start and end where the interlaced articfacts remain otherwise it's fine even fades during the movie. Is there any filter i can apply after the ivtc is done to blend the start and end? without affecting the rest?

[AS]

Apply a deinterlacer only at those sections, via trim().

[Chainmax]

Why loading decomb if you're not using it? In anycase, you might want to try the following IVTC methods, which were recommended to me by Revgen:

For TFF clips:

AssumeTFF()
Interp = SeparateFields().SelectEven().EEDI2(field=1)
Deinted=TDeint(order=1,field=1,edeint=Interp)
TFM(mode=6,order=1,PP=7,slow=2,mChroma=true,Clip2=Deinted)
TDecimate(mode=1)


For BFF clips:

Interp = SeparateFields().SelectEven().EEDI2(field=0)
Deinted = TDeint(order=0,field=0,edeint=Interp)
TFM(mode=6,order=0,PP=7,slow=2,mChroma=true,Clip2=Deinted)
TDecimate(mode=1)

You need EEDI2, TDeint and TIVTC.

[Daodan]

Hmm, interesting. I'll try myself to see if the results are better than with tfm+tdecimate only. But I'll bet it will be slow as hell because of that tdeint.

[Chainmax]

Well, it does combine the slowest options of TIVTC with a slow deinterlacer (TDeint) that uses an interpolator (EEDI2) as a backup. If you have the space, you're probably benefit from encoding the filterchain to an intermediary lossless format before the final encode.

[kcb]

hmm stupid question what is the difference between BFF and TFF?

[Chainmax]

If you want to learn about interlacing, telecining and field orders, I strongly recommend you to read the IVTC tutorial (http://www.doom9.org/ivtc-tut.htm) and the Decomb package's readmes, tutorial and reference guides (http://www.neuron2.net/decomb/decombnew.html). They have a lot of extremely useful info on those subjects.

[kcb]

Ok i've quiet familiar with the methods used in the tuturial and the field order. I just wasn't sure of what BFF stood for, i'm guessing Bottom Field Frame?

[foxyshadis]

Top Field First and Bottom Field First, which is the temporal order of the fields.

[Daodan]

Tried Chainmax's method and I'm pretty happy with the results. The strange thing is I now have the same problem kcb mentioned (the first few frames are blurry, not the last though). With this script I don't get something like that:
tfm(order=1,mode=5,pp=7,slow=2,micmatching=3)
tdecimate(mode=7)
Maybe you could try this method kcb, and share the results.

[Chainmax]

Just a clarification, those high quality IVTC/Deinterlace methods were first thought of by tritical (the creator of TIVTC, EEDI2 and TDeint) and suggested to me by Revgen. I'm merely passing on useful information.