Inverse telecine symptom?

[Shark]

First of all, thanks to all contributor. I finally understood the concept of differents term like telecine and similar. I still unfortunatly get some trouble to assimilate symptom for artifact i see on my video.

So, let's resume it.

i rip NTSC movie. i read on the article Force Film, IVTC and Deinterlacing that if dvd2avi give a NTSC > 95% or FILM < 5%, then i can use Forced film to IVTC. That's what i do and always did since i learn to rip. Not because i understood it, just cause it was mentionned...

But i got those horizontal line that boring me. I understand from this that it could be the result of a partial, or not perfect to be exact, inverse telecine. That some 3:2 pulldown frame added has not been well removed or analysed, right? it's the way i interpret this at least. But may be i'm wrong and that's why i post this thread.

I told myself that i should give a try to decomb to do it manually. So i used dvd2avi without forced film check. Then i used this script:

mpeg2source("E:\title.d2v")
Telecide
Decimate(cycle=5)

In fact, it gives me the same result, still theses horizontal lines.

http://membres.lycos.fr/aguirre2001fr/image/lignes.jpg

I read somewhere that uncomb could solve this. i tryied it on d2v with forced film activated, but also combinated with non checked + decomb as indicated below. Same thing.

So i ask myself: does thoses lines resulting from a "bad" IVTC or something else? Is there a way to get ride of them? how? Is my setting for decomb bad? Does it worth to use decomb as dvd2avi give me NTSC around 99% or Film 1% ? It seems that it's only NTSC DVD material that give this kind of symptom....

I would be very thankfull if someone could take the time to explain me thios or to point to some valuable articles like the one about those concept!

thanks a lot to all contributors of this forum. It's more than a gold mine, a little hard to follow but instructive ;)

Shark

[Hiro2k]

What is your film? Also you can try to use a De-interlace and see if that fixes your problem.

[Awatef]

First, I don't see anything wrong in the picture!!

Second, you got it ALL WRONG!!
When FILM > 95% you can apply Forced Film, and you'll get always or most of the time (depending on how near it is to 100%) no combing, only a few frames MAY come interlaced. The best is, you put a FieldDeinterlace(full=false) command in your AVS script to blend the few frames that may come interlaced.
If FILM < 96% (or you see only NTSC), you should apply IVTC. That means, you DON'T check Forced Film, then you apply Decomb in your AviSynth script, before any kind of resizing/filtering like this:

Telecide()
Decimate()

[Shark]

Arf, something i didn't catch so...

If i well understand your comments, if NTSC = 99% it means i don't have combing effect so i don't need to IVTC, is that what you meant?

Now, you mean also that residual come from some frame that was not progressive (ie interlaced) and that's why i need to use FieldDeinterlace(full=false), right? So my potential pb wouldn't come from a bad Telecine but from some interlaced frame left? So i suppose that when DVD2Avi detect the movie as progressive, it's a kind of average or at least he didn't detect some interlaced frame.

What i don't catch is that i thought i had to inverse telecine in any case: either forced film would do it or have to use another method if FILM/NTSC values are out of correct ranges. So why is it wrong to use decomb even if the value are correct? I didn't thought i had to de-interlace since progressive was displayed in statistic window. I based my judgment according to the doom9 article + the decomb http://shelob.mordor.net/dgraft/decomb/decomblegacy.html (review)

I will give it a try. As for the famous horizontal lines, my pointer is may be not placed at the right place. Check on the Post in the bottom left, up to the leave. I know there are not big lines, just bothering me to have this smal effect.

Shark

[neuron2]

@Shark

I see no residual interlacing at all in that linked image. I see normal digital sampling artifacts. If those are bothering you, you are going to have a very frustrating life in digital video.

>If I well understand your comments, if NTSC = 99% it means
>I don't have combing effect so i don't need to IVTC, is that
>what you meant?

DVD2AVI can't always be trusted. It could say 100% NTSC and the clip may still be IVTC'able (hard telecining). You apply IVTC when a clip has 3:2 pulldown telecining. The guide tells you how to determine if that is the case.

DVD2AVI does not detect interlaced frames. If you apply Force Film, it simply instructs DVD2AVI to ignore the MPEG RFF flags and to leave the frame rate at the film rate.

Relying on a tool's analysis is dangerous. There is no substitute for separating fields and actually inspecting the field structure of the video. That is the only reliable way to determine the nature of the source clip and hence its proper treatment. It is actually not difficult at all. One day I will write a guide about it. :)

@Atawef

>If FILM < 96% (or you see only NTSC), you should apply IVTC.

For film source, yes. If the source is pure video, then no.

[Awatef]

@ neuron2
I know, I know (just didn't want to rewrite the whole guide here ;))

@ Shark (I'll write french here because he doesn't seem to understand well)
Si FILM est *superieur* a 95% (c.a.d. NTSC inferieur a 5%), tu actives "Forced Film".
Dans tous les autres cas, tu n'actives pas Forced Film, mais tu ajoutes les lignes d'IVTC (Telecide & Decimate) dans ton script AVS.

Comme neuron2 a dit, le telecinema doit etre numerique, du 3:2 pulldown, c.a.d. un cycle de 3 images propres suivies de 2 images entrelacées (verifiable dans les scenes panoramiques), pour qu'on puisse appliquer IVTC.
Si c'est du telecinema analogique (ce qu'il a appelé hard telecining) ou si tu as une source NTSC pure (tres rare), tu n'appliques PAS IVTC, mais juste du simple deinterlacing.

[neuron2]

@Awatef

Please observe forum rule 13 in the future. You could have done that in a PM. Thank you.

[Awatef]

Actually that's why I've written the remark that I'm going to write it in french, it was kind of demanding permission to write french at that place ;)
It was a brief translation of what was written before, don't worry neuron2, I didn't insulted you there :D
But OK OK, next time PM :D

[Shark]

Thanks a lot the two of you (and for the french traduction: i understood well what you said in english but it doesn't hurt to tell it in french in case :D )

If i well understood your comment neuron2 (by the way your filters are awsome!) it seems it's a kind of artifact i cannot avoid. I thought thoses lines was resulting of something wrong in my procedure. I first thought it was related to "combing", but when i saw a sample of a video "called" combing, then i knew it was not the same problem. So i tryied to find something else as for reason.

I read more about the way half frames are put together:Here (http://www.avdeals.com/classroom/Proscanexplained.htm). When you see the samll animation, my result is like the "Field 2" in the interlace scan demo. Off course, it's not so explicit in my video, it just that it was the closest pattern i found...

Here is my video hoping you'll see what i meant wiht my "pb" (it's more visible in the bright area)

http://us.f1f.yahoofs.com/users/fafec29b/bc/Sample/sample.avi?bfScZC_ASclGFO1B

I know this is not a big defaut and when you watch your monitor more than 1 meter, it's quite invisible. It's just a matter to get the best result as possible. I also ask myself if this can be due to the fact i choose TV Scale in dvd2avi?

Moreover when you talk about digital sampling artifcat, could you be more precise? Are you reffering to blocking? In positive case, i know it's resulting from compression and i can't avoid them.

Thanks a lot for your replies

Shark