Another audio sync question

[varelse]

I've browsed/searched the fourms and the guides and still don't understand what is wrong.

Here is what i've done.
ripped a DVD with SmartRipper
DVD2AVI to extract the AC3 track
use TMPGEnc to IVTC and manually reselect frames it missed
VFAPI converter to change .tpr to .avi
nandub to filter video and recombine to audio and saving as an .avi

It took me a few tries to get the final size exactly right but I'm still having an audio sync problem. It is perfect at the beginning but slowly become out of sync. by the end (57 mins) it is off by about 5 seconds.

I assuming the problem is the frame rate change from 29.976 down to 23.976 form the IVTC but i don't know how to correct this.

Can anyone help?

[jggimi]

Welcome to the forum.

A year ago, I was having problems with audio sync. I switched from TMPGEnc to Gordian Knot around November. Haven't had any sync problems since.

GKnot is used for producing .avi's with the DivX video codec. I recommend it.

[Yvez]

If you try to change the framerate with AviFrate or something like that, i don,t recommend you change 6fps..., try changing 0.02 fps at a time.

5 sec over a 90 minute movie isn't all that much...

anyway, with GKnot, I've never had a problem either, and the quality is more than respectable, it's great! :)

[varelse]

I originall tried Gordian Knot but the resulting file had a LOT of interlaced frames.

I had to switch to TMPGEnc to manually IVTC to get rid of them all.

Also with Gordian Knot, I couldn't set the Author/Copyright tags in the properties of the resulting .avi, which I can with Nandub.

Is it possible to fix this without using GKnot?

[DJ Bobo]

Then you missed the point in GKnot: you can tweak the AVS script as you want.

For almost perfect IVTC results, combine IVTC2.2 (download this from the download page of Doom9) & Decomb this way:

loadplugin("...\ivtc.dll")
loadplugin("...\decomb.dll") <- optional
ivtc(44,11,95)
FieldDeinterlace() <- optional

ivtc is your garantee for most accurate inverse telecine, fielddeinterlace() is optional and you can use it if you're too sensitive against interlacing lines, though ivtc does a good job on its own and there won't be much for fielddeinterlace to clean ;) (I personally rarely use fielddeinterlace after ivtc)

[varelse]

I tried it without the decomb/FieldDeinterlace and though the IVTC took care of almost all of the interlacing lines, there were still enough to notice. Next time with the decomb/FieldDeinterlace on and it turned out perfect.

Thank you very much DJ Bobo

P.S. What do the numbers in the ivtc command represent?

[DJ Bobo]

The numbers .. hmm ... what should I say ... I'll tell you my story :D
First I started with the help file recommended settings (32,8,15)
Then kept experimenting a lot with many Anime DVDs, first one to change was the second one, I changed gradually from 8 to 11 until it has detected interlaced from non interlaced frames correctly. Next was the third one: 15 gradually up to 110 until it detected pattern changes correctly. Next was the first parameter, which can detect less interlaced from more interlaced frames and that was done with 60 correctly.

Then came some Awatef guy and posted some (44,11,95) settings. After checking, I noticed that my values 60 and 110 were a little bit exagerated, I mean 44 and 95 were high enough to get proper IVTC.
Since then I wasn't able to tweak anymore, I think this is already the best what one can get with this automatic filter.

NB: some guy here in the forum said that (60,11,110) were actually better for him than (44,11,95). I can't confirm and I can't deny (wasn't able to catch any differences). I'm sticking to (44,11,95) since I feel the reduced parameters will limit detection errors.