IVTC help for newbie

[nfv]

Greetings

I will start out by saying I have no idea how to use avisynth :)

that being said I have it downloaded and installed don't know if I have correct version? anyways....

I would like to know how to convert a 29.970fps avi to 23.976fps avi

the interlace pattern is 3 interlaced and 2 progressive NOT to be confused with 2 interlaced 3 progressive

so...how would I achieve this using avi synth?

many thanks in advance..and please don't burn me at the stake for being a newbie

[ianken]

Use TIVTC or Decomb. Google it. Both filter packages have samples.

But before you try it you should at least get a simple script working. Visit the AVISynth WIKI. It has a getting started section.

[manono]

Welcome to the forum,

Your progressive/interlaced pattern is unusual and to be sure of what you have and how best to treat it we'd need a small untouched sample. Cut maybe 10 seconds of a section with steady movement and upload it somewhere (Sendspace, MediaFire) and provide the link. Based purely on how you described it, I'd doubt it's a candidate for an IVTC.

[nfv]

Welcome to the forum,

Your progressive/interlaced pattern is unusual and to be sure of what you have and how best to treat it we'd need a small untouched sample. Cut maybe 10 seconds of a section with steady movement and upload it somewhere (Sendspace, MediaFire) and provide the link. Based purely on how you described it, I'd doubt it's a candidate for an IVTC.

many thanks for your response and welcome :thanks:

what format would you like the file in? the original is mpeg2 but it was done on a dvd-ram dvd recorder very old recorder first generation...I tried to use tmpgempeg editor to cut a 10 second clip out of it but it won't recoginize it :devil:

I can convert it in procoder to any format you like what should I convert it to? for you to analyze

many thanks :)

[TheRyuu]

If it's mpeg2: dgindex, mark where you want to cut, save and demux.

[nfv]

ok...i googled dgindex..its part of another program? i'm not certain how this works?? is dgindex a stand alone program or part of another and do you have to use it with avisynth? :thanks:

[manono]

DGIndex comes as part of the DGMPGDec suite of related programs. If you don't have it yet, get it here:

http://www.videohelp.com/tools/DGMPGDec

Open a VOB in it and use the [ and ] buttons to isolate a small piece. Then go File->Save Project and Demux Video. If the resulting M2V meets the requirements I set, upload it.

No, you don't need AviSynth to cut it. You will need AviSynth to filter your video later on. And no one wants it in any format other than the source format.

[nfv]

ok...here it is

http://www.mediafire.com/?xpkkyeebg8c18ob

I think that will do

[dragon_warrior]

A simple qtgmc().selecteven would do it good. It's not telecined so no point in using TIVTC
*edit: 10 - 20-sec clip is enough for a sample ;-)

[manono]

Cut maybe 10 seconds of a section with steady movement...

Not too good at following instructions? Instead of 10 well chosen seconds we get over a minute of nearly useless intro. Next time, show people walking or something. Moving.

Anyway, there's enough there to figure out it's field-blended and can't be IVTC'd. It has to be unblended using something similar to this script:

Yadif(Mode=1)#or your favorite bobber.
SRestore(FRate=23.976)

http://avisynth.org/mediawiki/Srestore
http://forum.doom9.org/showthread.php?t=95924

This can be fixed only by using AviSynth. There's nothing else out there that can unblend. So, either take a crash course in AviSynth or forget about fixing this series.

That kind of thing is put out only by the worst fly-by-night DVD production companies. Too bad this show hasn't had a decent release on DVD.

[nfv]

sorry about clip being too long :(

when I ripped the dvd-ram it was not easy, I had to use ISObuster and it was not like any ordinary file structure like 1gig vob peices so...that was one vob and it was like 110,208 in file size

I will do my best at what you suggested

thanks

[johnmeyer]

Actually, you can use IVTC to get where you want to go, although perhaps other methods may be preferred by others. The following does remove the "ghosted/blended" field that occurs every fifth field:

tfm(source,mode=2,pp=1,cthresh=35,micmatching=3,scthresh=2,mmsco=false)
tdecimate(mode=7,rate=23.976)

This is the code I use to remove pulldown fields from transfers I do on my shutterless movie projector. It seemed to work just fine on your example.

[edit] I changed pp=0 to pp=1 after the post by manono (immediately below) pointed out my error.

[manono]

In many parts of it there are two blended fields in every five. Granted, one of them is faint, but it's still there and easy to see. With PP=0 as in the script provided, quite a few interlaced frames slip through. Perhaps PP=0 was set only for testing and was meant to be removed when doing the actual encoding. In addition, the script above leaves behind a ton of duplicate frames, with missing frames resulting from that. So parts of it play jerky. It's an interesting script but yes, other methods may be preferred by others.

[johnmeyer]

... With PP=0 as in the script provided, quite a few interlaced frames slip through. Perhaps PP=0 was set only for testing and was meant to be removed when doing the actual encoding. In addition, the script above leaves behind a ton of duplicate frames, with missing frames resulting from that. Thanks for catching that mistake. I have edited my post above, changing pp=0 to pp=1, and it now works correctly.

To check that my change really works, I did a separatefields on the original video and stepped through it while simultaneously, in another instance of VirtualDub, stepping through the results of the script above, but with separatefields added at the end of that script as well. By doing this, I could detect any situations where the script misses a field.

With pp=0 I found lots of places where the script was dropping frames, just as manono pointed out. I don't know how I didn't catch that. By contrast, with pp=1, the script works as it should, based on looking at over 500 fields.

I guess I need better version control in my scripts because, as Gavino has noted, I seem to always grab earlier (buggy) versions. I'll try to take more care next time before I quickly grab something.