Deinterlacing artifacts after inverse telecine?

[newpball]

I am working on an old film source which is encoded with a 2:3 pulldown and put on an NTSC DVD.

Applying an inverse telecine still gives me artifacts that look like interlacing artifacts:

What causes that? A bad scan, compression artifacts?

[jagabo]

Could be many things. Post a sample of your source. And what software? TFM().TDecimate()?

[newpball]

Here is a sample of the source:

sample.m2v

Yes TFM().TDecimate()

[Brad]

I would blame the horizontal timebase errors from the analog tape step, mainly.

Just add Vinverse to the end of your script IMO. I think the source is too low-detail for it to cause any softening.

[jagabo]

With that video it's just jitter from horizontal time base errors, and MPEG compression artifacts. You can follow up with vInverse() to blur them away.

[manono]

I don't think it's 'deinterlacing artifacts' as it IVTC's perfectly. Sometimes when I have similar problems with tapes I'll use QTGMC followed by SelectEven (or odd, whichever is better as sometimes one field is plainly better than the other) and TDecimate.

Much more problematic, in my opinion, are the blown out whites. But I'm sure you already plan on doing something about them.

[newpball]

I don't think it's 'deinterlacing artifacts' as it IVTC's perfectly. Sometimes when I have similar problems with tapes I'll use QTGMC followed by SelectEven (or odd, whichever is better as sometimes one field is plainly better than the other) and TDecimate.

I just tested QTGMC() followed by SelectEven() after IVTC and it seems to give a much cleaner result compared to IVTC() and vInverse().

Much more problematic, in my opinion, are the blown out whites. But I'm sure you already plan on doing something about them.

Well what can you do, when there is no information there is no information.

My plan actually was to render this in tri-toned blue.

But I am open to suggestions.

[manono]

I just tested QTGMC() followed by SelectEven() after IVTC and it seems to give a much cleaner result compared to IVTC() and vInverse().

The reason I mentioned QTGMC in the first place was because the combination I mentioned will give back a 'pseudo-IVTC' together with cleaning. If all you're using it for is cleaning then you may as well skip the QTGMC and go with a cleaner, ones such as the FFT3DFilter or DFTTest, both of which are used by QTGMC depending on the settings. If you used QTGMC at default then it was DFTTest that cleaned up your video.

Well what can you do, when there is no information there is no information.

The information isn't missing because the whites are blown out. They aren't so bad that they can't be fixed. There's no information there because this thing has been through the mill already and and whatever detail it once may have had has already been destroyed. I might do something like this:

Tweak(Bright=-10,Cont=0.9,Coring=False)

Adjust things to your taste. I think tinting is a good idea in which case you might not want to lower the contrast so much. I don't know, up to you.