PAL truly interlaced looking DVD that is in fact telecined???

[niamh]

I realize this is limit off topic in the avisynth forum, but i cant figure where else to put it, please move me if needed :):)

hhhm , here goes :)
I'm backing up season 1 of the Water Margin (asian<<==:rolleyes: again ;)TV series of the 70's, broadcasted in England)

Anyway, episodes 1 to 8 were your average interlaced PAL video (or so i thought ;)), so i dutifully used tomsmocomp and a couple of slowass filters , since the source is ..... disgusting :D

then i realized that episode 9 was totally progressive(!).Anyway episode 10 is back to being interlaced, so here i got thinking that it might not be interlaced after all but only field shifted, since it makes no sense to have some episodes that would be interlaced, and others progressive.(then when does it ever make sense?)

Back to the drawing board, i learnt all i could about IVTC, telecide and such(PALland idiot here)I used:
telecide(guide=2)and miracle, the output is perfect (compared to the source)even without any noise filter!
Having had a look frame by frame through vdub, i can assure you though that there is NO progressive looking frame in there whatsoever, and this is what the telecide tutorial says :
Of course, if your clip does not have progressive content, it will not help you, and you shouldn't be using Telecide() at all!
which makes absolute sense due to the nature of interlacing ; but THIS clip looks totally interlaced and it turns out it isnt, as such.

so here are my questions:

Did i find THE exception or has anybody else encountered this?
how many PAL DVD's are actually field shifted instead of interlaced?
Am i going to be stuck forever double-checking my ''interlaced'' PALs before deciding what course of action to take, or is there a simpler method someone more intelligent devised?
Did i stumble onto something interesting, or am i just rambling? :D (dont answer that one, my feelings get hurt easily ;))


Of course, deinterlacing did a very decent job of it, and there won't be any bitrate gain since there is no decimating involved, but it can't be argued that reconstructing progressive frames is more precise than deinterlacing, as well as being faster and much lighter on the subsequent noise filters that i usually use to cover up the damage ;)

I'd appreciate any feedback from PALers who have encountered this and do use telecide, even though i've read over and over it shouldn't really be used on PAL movies.
I've found a certain number of threads through search too, but most are concerned with PAL conversion or the answers are scattered throughout various threads and i couldn't find anything that was really relevant.Sorry if I missed something :)

[Guest]

Originally posted by niamh
Back to the drawing board, i learnt all i could about IVTC, telecide and such (PAL land idiot here). I used: telecide(guide=2) and miracle, the output is perfect (compared to the source) even without any noise filter! You get post-processing turned on by default, so you haven't given us enough information to know if it was getting good frames through field matching or through post-processing.

Having had a look frame by frame through vdub, i can assure you though that there is NO progressive looking frame in there whatsoever Having a look at what, the raw source or the filtered output? Anyway just looking at frames is not reliable. You need to look at fields (see below).

which makes absolute sense due to the nature of interlacing ; but THIS clip looks totally interlaced and it turns out it isnt, as such. So maybe you had field-shifted PAL. What's the big deal?

Did i find THE exception or has anybody else encountered this? Who knows, you haven't told us enough. Even so, it's highly unlikely you've found any exception. You're asking if anyone else has encountered field-shifted PAL?

Am i going to be stuck forever double-checking my ''interlaced'' PALs before deciding what course of action to take, or is there a simpler method someone more intelligent devised? Of course you have to check your source before deciding how to render it. Yes, there is an intelligent way. Do separatefields() on the raw source and step through the fields in a motion scene. If you see a new picture every field then it is true interlaced. If you see 2 fields for each picture then it is progressive.

it can't be argued that reconstructing progressive frames is more precise than deinterlacing, as well as being faster and much lighter on the subsequent noise filters that i usually use to cover up the damage Correct. If it's progressive, then use field matching; if it isn't, don't. Are you looking for reassurance here, or what?

I'd appreciate any feedback from PALers who have encountered this and do use telecide, even though i've read over and over it shouldn't really be used on PAL movies. Use your own reasoning and rely less on "authority".

[niamh]

thanks Donald for answering me :) I did spend 3 days before posting this, and Murphy's law made me find the answers just after I posted, as usual :) . Well, maybe my silly posting will shed some light for other PAL users .



You get post-processing turned on by default, so you haven't given us enough information to know if it was getting good frames through field matching or through post-processing.


Yes. i'd thought of that, sorry :) the result is the same without post-processing.It comes out clean(i only checked a few hundred frames this time)through field matching only.

Having a look at what, the raw source or the filtered output? at the d2v,loaded into vdubmod through avisynth, unfiltered

So maybe you had field-shifted PAL. What's the big deal?
I'd never seen one or heard of one before :o now I have.Every single discussion i've seen on telecining and field shifting has been by NTSC people, with a few vague recommendations for PAL at the bottom, and a lot of ''i'm not too sure'' . It isnt that helpful nor documented.(yep, i saw documentation in one of your docs, after i posted :D I swear it wasn't there before !! lol)

Of course you have to check your source before deciding how to render it.
sounds obvious :) but again, for PAL, I've always read it was either progressive or interlaced.I've never seen a detailed howto in any ripping help guide on field shifted PAL's.This is why i was wondering too, was it a rare occurence?

Correct. If it's progressive, then use field matching; if it isn't, don't. Are you looking for reassurance here, or what?

hhhmmmmmmmm yeah :D

Use your own reasoning and rely less on "authority".

Hey, i just did!;)

Oh, and as for that
Of course, if your clip does not have progressive content, it will not help you, and you shouldn't be using Telecide() at all!
it's my fault for equating progressive with a "non-combed" picture in dvd2avi,for example, which i had been taught to do(wrongly) :)

[Guest]

Originally posted by niamh
it's my fault for equating progressive with a "non-combed" picture in dvd2avi,for example, which i had been taught to do(wrongly) :) You could plausibly argue that it is my fault for using imprecise language. :)

Still, I'm sure it is useful to others to know of your experiences. Thanks for letting us know about them.

[niamh]

Well, thank you again for answering, and hopefully this will ram the point home about field-shifted PALs :)
The SeparateFields method to check on the source is the easiest, shortest and clearest method i've read about too, it's very much appreciated :)

And you don't use imprecise language, it's the rest of us who do ;). Maybe it's time some nut did a dictionary for encoding terms, or who knows, maybe someone did already?