Can't propertly deinterlace source

[kostek00]

Hello, I need your help once again. I have a DVD and I want to encode it to smaller size but the problem is deinterlacers can't handle interlacing. I tried all deinterlacers in megui and also QTGMC and only last one is somehow good but still not enough for me. If you have any ides I will be very grateful. I'm attaching sample and if you can, please, write how to use filters you recommend.

Sample file.

[Stereodude]

Why is it a .mkv file? Is it an original untouched mpeg-2 video stream in it?

If that's the original source it's blended, not interlaced.

[Jenyok]

LoadPlugin("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\EEDI3\eedi3.dll")
LoadPlugin("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\NNEDI3\nnedi3.dll")
LoadPlugIn("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\TDEINT_20070122\tdeint.dll")
LoadPlugIn("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\TMMV1\tmm.dll")

#
# Деинтерлейс видео
# До обрезки и изменения размеров !
AssumeTFF()
#AssumeBFF()


edeintted = eedi3(field = -2,                 \
                  sclip = nnedi3(field = -2,  \
                                 qual  =  2))
TDeint(edeint   = edeintted,     \
       order    = -1,            \         
       mode     = 2,             \
       sharp    = true,          \
       mtnmode  = 3,             \
       full     = false,         \
       tryWeave = false,         \
       type     = 2,             \
       emask    = TMM(mode = 1))

.
This is Very Good, but Very Slow deinterlace.
.

[aldix]

@Jenyok: looks very good. Is it meant exclusively for 100% video sources, or would it work for cases i.e. ~<90% film? U.S. releases stupid pull-down rules for TV content both annoy and confuse me, that's why I ask; currently, when I think deinterlace is required I use it through QTGMC via clip2 in tfm. Could, ideally, the above be used similarly to better results?

Thanks anyways.

[kostek00]

Originally Posted by Stereodude

Why is it a .mkv file? Is it an original untouched mpeg-2 video stream in it?

Yes, it's original untouched mpeg-2 video stream, I used mkvmerge to quickly split it, thats why it's mkv.

Originally Posted by Stereodude

If that's the original source it's blended, not interlaced.

So it's blended, what filter do you recomend for it?


@Jenyok
Your deinterlacer works even better than QTGMC. Thanks, I will keep it.

[Stereodude]

Originally Posted by kostek00

So it's blended, what filter do you recomend for it?

A deblender. I don't have any real experience with deblending, so I don't know what to suggest.

[feisty2]

srestore is great for deblending

[kostek00]

I checked few deblenders and only Srestore removed all blending from clip I provided but added a little blocking on blended frames. Still it was the lowest amount from all those I checked. I will encode whole episode and say tommorow how it looks.

[Stereodude]

Originally Posted by feisty2

srestore is great for deblending

FWIW, I couldn't come up with settings for srestore that would deblend his sample, but I'm certainly far from an expert when it comes to srestore.

[kostek00]

Around 90% of blending has been removed. Thanks for help.

[Jenyok]

Sorry for Russian letters in comment in script.
.

[kostek00]

Not a problem.

[aldix]

Could I also get the reply to my question, please? Or is it too much asked.

[Jenyok]

aldix
.
I did not understand your question fully.
What did you mean in your question "would it work for cases i.e. ~<90% film?" ?
What did you mean "~<90% film" ? What is it ?
.
Deinterlace script, I wrote upper, is for ALL video source.
Film, cartton, video VHS and so on...
.
My deinterlace script is special script ONLY for deinterlaced.
QTGMC, in my opinion, is great monster with no good (bad) deinterlace function.
.
You could try to use my deinterlace script and QTGMC function and then compare two results.
.

[aldix]

I meant NTSC DVD content which requires IVTC, but some interlace artefacts on frames might remain regardless. Would your script work in such cases where presently

deint=QTGMC(preset="slow")
tfm(clip2=deint).tdecimate()

is used? That is, would your option produce more accurate results?

Don't know if I can explain any better...