When a "well IVTC configuration" give us a "BAD IVTC"?

[FreierGeist]

at this trhead: http://forum.doom9.org/showthread.php?s=&threadid=23776 hakko504 ended with a doubt...
and i'm not completely sure about it too...

When it's impossible to perform a good IVTC?
Why the "BAD IVTC" occours?

i think that bad ivtc happens when we try to perform a IVTC in a "original" 29.970fps source

that means: when the source wasn't Telecined

and to perceive if that's the kind of source we're dealing with is to see if DVD2AVI show us a "PURE" INTERLACED source!

resume: the non-Telecined sources (original 29.970fps) it's shown as INTERLACED @ DVD2AVI

so.. this kind of source we can't IVTC

am i right or am i right? :p

thanks!!!

[Hanty]

All sources can be telecined, with varying results.
You'll have to break it down, it is very rare that any encode is 100% free form interlaced frames/parts. It might not even bw caught by the eye unless you watch it frame-by-frame in vdub but it might still be there.

[FreierGeist]

i'm not looking for this perfection
:D
what i want to know is: when will appear "bad ivtc" when watching the movie in some player (WMP 6.4 for example)...

not a frame-by-frame inspection, but with an accurate view @ some player (sorry for my bad english...)

another question: is it possible to IVTC a "pure" NSTC with a "pure" INTERLACED frame type?

and... please... if so, just give me some clue on how to perform this IVTC

because i've already dealed with this kind of source and it was "unIVTCable"

:)

THANKS!

[]s

[FreierGeist]

@ Hanty

thanks for trying to help!
you wrote: "All sources can be telecined, with varying results."
i agree with you, but i'm looking for good results and possibilities os IVTC, not Telecine.

[]s

[Kedirekin]

I don't think there is any perfectly reliable automated way to determine if video will IVTC well or not.

For example, dvd2avi may report 100% NTSC on a title that will IVTC very well. A lot of anime is like this.

On the other hand, dvd2avi can report 96% FILM, and it's possible that neither Force Film nor IVTC will produce good results. 4% of a 2 hour movie is nearly 5 minutes - if all of the 4% occurs in one spot and it's fully interlaced, you'll certainly get unwatchable results in that 5 minutes.

I think the best way is to do a frame-by-frame spot check. It doesn't take too long. Just check 10 to 15 frames every 3-to-10 minutes throughout the video. If you see a consistent pattern of 2 combed and 3 non-combed frames (c-c-nc-nc-nc - c-c-nc-nc-nc), it'll probably IVTC well.

[manono]

Hi-

Kedirekin is right on the money. Forget DVD2AVI for determining whether or not it can be IVTC'd. Look at the frames.

[FreierGeist]

thanks!

but... since i know, there's a lot of Telecine methods
and, how do i know if it's "IVTCable" or not (perceiving the 3:2pulldown Telecine method is a piece of cake:), what about the others Telecines) ?
is there a kind of source wich IVTC is impossible?

:confused:

THANKS A LOT!!!

[]s

[FreierGeist]

sorry if i am being redundant/repeating myself...
but i didn't found the answer to my doubts...

[FreierGeist]

@ Kedirekin

thanks for your answer
but what i was thinking is if the Frame Type is the important thing to a "good or bad IVTC"...
i've already "well IVTCed" a 100% NTSC source
but i have troubles with 100% NTSC + 100% INTERLACED

so... i think that the 100% INTERLACED it's the trouble...

[manono]

Hi-

trbarry has spoken of a 2:2 type pulldown on some PAL material whose frames can be reconstructed, but having never seen any like that, I can't comment. But both GreedyHMA and Decomb can reconstruct them without changing the frame rate.

but i have troubles with 100% NTSC + 100% INTERLACED

Again, look at the frames. If, in motion scenes, all you see is interlacing (combing) in every frame, then it can't be IVTC'd, but only deinterlaced.

[FreierGeist]

@ manono

thank you very much!!!
you're (as always) one of the most kindly member trying to help!

"If, in motion scenes, all you see is interlacing (combing) in every frame, then it can't be IVTC'd, but only deinterlaced."

only @ "motion scenes" ?

[]s

[manono]

Hi FreierGeist-

I say look in the motion scenes because that's where it's easiest to tell. If there's no motion, then the 2 fields that make up a frame won't be displaced from each other (did I say that right?), and you won't be able to see the interlacing.

[Kedirekin]

Unfortunately 100% NTSC plus 100% Interlaced isn't very reliable either. The 100% interlaced is based on the MPEG encode method dvd2avi sees on the frames, and there is nothing sacred about that. It is perfectly valid to use the interlaced encode method on all frames, regardless of if they have combing or not (a bad use of bitrate, but perfectly valid), and it is the combing pattern you're interested in for IVTC, not the encoding method.

You're right, there are a lot (or at least several) telecine methods. Here's a description of a particularly perverse method from the AviSynth filter manual (see PeculiarBlend):
But some use a much nastier system in which the crossover to the next frame occurs in the middle of a field--in other words, individual fields look like one movie frame on the top, and another on the bottom.
To the best of my knowledge, the IVTC methods available to us only work well on pulldown type telecine, which is the one with the pattern of 2 combed and 3 non-combed frames described earlier.

monano's is right, it's easiest to determine the presence or absence of a pattern by looking in motion scenes. I especially like medium-speed panning scenes. They show combing artifacts well, and they last long enough to really check for patterns.

This also brings up another point - it's much easier to identify bad candidates than good candidates. If you consistently see 4 or more combed frames in a row, it almost certainly won't IVTC well. However, seeing the pattern of 3 and 2 doesn't necessarily mean that the pattern runs the entire length of the video. Often, the only way to really tell is to encode it, then watch it at normal speed all the way through (and don't get distracted - you need to pay attention for the entire length of the video - very troublesome).

Admittedly it's bothersome to not have some reliable tool to identify if the video can be IVTC well or not. Perhaps one of these days we'll have a tool to better determine this, but for now we muddle through.

[FreierGeist]

@ Kedirekin

WOW!
THANK YOU VERY MUCH!
now i can see that this "problem" is bigger than i was expecting to be...

that kind of video capturing "must" be eliminated from the planet (for the sake of the digital-video-interested-people happiness)! :p:p

and i hope that someday an "universal video system" takes place in this "chaotic world"!
:D

[]s
FreierGeist