TIVTC is a poor deinterlacer?

[killerteengohan]

Someone is trying to say do not use it because it does not work properly.

Is this false information?

[poisondeathray]

That person doesn't know what s/he's talking about

Removing every 5th frame is a decimation operation, not a deinterlacing operation .

If it says source type "film" in DGIndex , and it's 100% , you just use "force film" and you don't have to do anything else, and it will be perfect. This means it was "soft telecine" source

But on other DVD's or if it's not 100% film - You can delete every 5th frame straight up if you want to. The problem is with edits and cadence breaks. That means the wrong frame can be deleted if you delete every constant 5th frame

[killerteengohan]

I think they're just a Vapoursynth fan or a noob encoder with a know it all attitude, trying to tell everyone how to do things. That or they have their terms and understandings all mixed up.

[poisondeathray]

That or they have their terms and understandings all mixed up.

That's what it sounds like. You should help them out, otherwise misinformation gets spread and that helps nobody. I advise you to do politely, and nicely or it will probably blow up into some flame war. Or just ignore it

[manono]

An IVTC is NOT a deinterlacing operation. TIVTC and Decomb do have deinterlacers built into them for fields that can't be matched properly. It can be switched for about any other deinterlacer if preferred.

Besides being a semi-literate moron, he's an ignorant fool. Guys like that don't want to be taught better. They only want confirmation for their biases. I don't see why people like that are deserving of patience, politeness or respect.

Of course, this is a small world we travel in so he might show up here to spout some more nonsense.

[killerteengohan]

yeah Im not even gonna bother arguing with them. All they want to say is they know what they're doing and give examples saying they did this in Vapoursynth.

(Hotlinked image removed)

[poisondeathray]

Yes, it is possible in vapoursynth . With ESRGAN ( AI based training) . That model used is specifically trained for American Dad from PRAGMA . And that trained model should be able to be used for other types of cartoons with similar drawing styles. (You wouldn't be able to apply that model to ,say, an anime style source)

You can see other examples here
https://forum.videohelp.com/threads/393630-x265-exe-Inputting-frames-from-folder#post2554459

https://gitlab.com/imPRAGMA/VSGAN

[poisondeathray]

Actually, that probably is PRAGMA . My detective skills say the avatar is the same in that screenshot

I think the same model was used as this one
https://upscale.wiki/wiki/Model_Database#Drawings_2

You can ask him to post the script . Some pre-processing is typically done, because these GAN methods to tend amplify noise and artifacts. There are other types of models, denoising models, dehaloing models, etc...

[killerteengohan]

It's fine poison. It's not a big deal to me who, how or what they did. I was just wanting to clear up whether or not the IVTC was as crap as they were claiming. Thanks for clearing up my original question you two.

[dannyboy48888]

TIVTC suite is my goto now and days. If after stepping through the video if see a pattern I try :

Telecide(guide=0,post=0)
TDecimate(m2PA=True,maxndl=7,Mode=2,Rate=23.976)

OR

tfm(order=-1, mode=5)
TDecimate(m2PA=True,maxndl=7,Mode=2,Rate=23.976)

rate can be 23.976,24,25,etc.

On some "made for TV" stuff like legends of tomorrow, Dolly Pardons a Coat of Many colors, etc. the 5th frame is a full frame so I just use TDecimate only

After a test encode if I dont like motion or if I didn't see a pattern and it's interlaced I just deinterlace and let it be 60fps like so:

tdeint(mode=1, tryweave=true)

Star Trek the next generation DVD's, Doctor Who 2005 Region 1 DVD's, etc I do this to. I may have a blend here and there, but overall there isn't too many "jaggies", it usually stays in sync, the motion is smooth, they're fast, most progressive frames are left alone, and it works under wine in linux via AVISynth. I've used srestore, QTGMC, etc. They are fine tools but can give odd results at times for mixed material. for noise I may use hqdn3d(2,2,3,3) give or take the strength of the first 2 numbers. If down the road I hate the way it looks I can re-encode from my master, but so far I have yet to be overly distracted with the final encode. Reason Why I listed both telecide and TFM is there's times one will get it perfect while the other will mess it up. Example is A-X-L or Captain Marvel on DVD. some scenes TFM messes up the field match no matter how I tweak, while telecide does it right. I usually try telecide first now and days.