View Full Version : About interlaced...I know that this subject is beaten to death, but....
vigi_lante
4th January 2005, 21:22
For me, this subject is hard to understand, because when people talk about interlaced content, they don't do the differentiation between truly-interlaced, hybrid or telecined.
Well, my main objective right now is to capture, from Television, NTSC cartoons and anime. I don't know exactly how to classify my source, sometimes I see a repeating pattern of 4 clean frames followed by 1 interlaced frame, and sometimes I see 3 clean frames followed by 2 interlaced frames. Maybe it's hybrid.
I don't want to loss my time trying to figure out what kind of source I have and what settings I need to use, and since I will watch my videos on a Television, I want to leave the source as it is, preserving what is interlaced as interlaced, and what is progressive as progressive.
But the problem is: How to handle this kind of video using Avisynth and at the same time preserving the original source characteristics ? Because I want to use my filters to clean the picture, but leaving the source as it is to retain quality, avoiding any de-interlace artifacts.
I'm trying to do the script, and so far I reach this one...
AviSource("c:\capture.avi")
ConvertToYV12(interlaced=true)
even=last.separatefields().selecteven()
odd=last.separatefields().selectodd()
even = even.UnDot().cnr2().edeen(7,14,21,2,3,true)
\ .TemporalCleaner(6,12).aWarpSharp().mfToonLite()
\ .Limiter()
odd = odd.UnDot().cnr2().edeen(7,14,21,2,3,true)
\ .TemporalCleaner(6,12).aWarpSharp().mfToonLite()
\ .Limiter()
interleave(even,odd)
weave()
But I dont know if it's right or wrong, what I need to change etc...
I wonder if these settings are only used for truly-interlaced content (not my case), if I need to select "Interlaced Encoding" on XviD Configuration and go on...
Sorry guys, I know that this subject is beaten to death, but after a lot of weeks searching for it, I still could not find what I need to do in this case...
One more observation...thanks to scharfis_brain, I learned that interlaced content needs a 1:1 reproduction on a TV, so since I will playback using Xbox Media Center, the video size need to be at least 640x480 in order to keep everything 1:1, avoiding overy scaling, and having at the sametime a fullscreen playback. So I will just capture at 640x480 (and not 720x480 as I was doing before), because in this way I will not need additional resizing.
Thanks.
Mug Funky
5th January 2005, 04:39
the treatment you're giving it seems alright to me.
btw, the width doesn't really need to be 1:1 - just so long as your playback equipment can output it at the right ratio for your TV.
the real thing to worry about is vertical scaling, as this destroys the interlace lines (unless you smart-bob first, then re-interlace, but usually that's a waste of time).
of course with capturing there's not much reason to cap at 720 unless your destination is DVD - certainly you wont really get anything extra out of the signal, as broadcast TV is nowhere near even 640 lines.
vigi_lante
5th January 2005, 05:53
Thanks...
But I just wonder how all this settings, that are specially made to process interlaced video, would work if the source have more progressive frames than interlaced frames (those that are somehow telecined for example).
Using "ConvertToYV12(interlaced=true)" for example, I know that you need to use this, if the source is truly-interlaced, but if it's source is hybrid or telecined, I just don't know if it would be really usefull. Selecting "Interlaced Encoding" on XviD Configuration is another example. The need to use "SeparateFields" before applying spatial filter is another one. You see ? If the source have more progressive frames than interlaced, I wonder how these settings, made to use only with interlaced frames, would affect the frames that are progressive...
btw, the width doesn't really need to be 1:1 - just so long as your playback equipment can output it at the right ratio for your TV.
Yes. I know, but Xbox Media Player can only output 640x480 and 720x480...so I will stick with 640x480.
Thanks...
Mug Funky
5th January 2005, 09:18
as far as "converttoyv12(interlaced=true)" goes, if you're working with telecined stuff you'll need to do it this way regardless or the chroma wont match properly in some cases. of course, if you're capping in yuy2, then it's best to IVTC before you convert to yv12 except in strange circumstances. with interlaced DVDs you'll be decoding straight to interlaced YV12, so IVTC filters will handle them properly.
if you're keeping things interlaced for output, then you might as well filter interlaced - you wont gain much from not doing it that way. spatial filters will perform slightly differently, but that's about it - the encoder (set to interlaced of course) will handle it just fine.
these issues are pretty small, so it's best not to lose sleep over them :)
vigi_lante
5th January 2005, 16:03
Well, I think that I will just capture using ffv1 (YV12) and skip "converttoyv12(interlaced=true)".
Gee, I didn't know it would be so problematic to process interlaced video, it's better be worth... :)
Thanks again...
scharfis_brain
5th January 2005, 16:11
Do NOT capture analogue video using YV12 compression.
It will destroy chroma movement, because most YV12-Codecs will only downsample progressively.
vigi_lante
5th January 2005, 16:21
I see...well, thanks for the tip. So I will just use huffyuv and "converttoyv12(interlaced=true)".
(cool avatar you have, how you did this interlaced effect ?)
scharfis_brain
5th January 2005, 16:26
why do you want to go to YV12?
stay as long as possible in YUY2 chroma.
let the encoder do the chroma-subsampling from yuy2 to yv12.
I made that interlacing effect of my avatar using AVISYnth.
vigi_lante
5th January 2005, 16:30
Well, I want to go to YV12 because 2 or 3 filters that I'm using only support YV12.
acrespo
6th January 2005, 12:51
Originally posted by scharfis_brain
Do NOT capture analogue video using YV12 compression.
It will destroy chroma movement, because most YV12-Codecs will only downsample progressively.
It's true for captures directly in MPEG2?
scharfis_brain
6th January 2005, 13:50
acrespo
No, there - if you hopefully set to interlaced encoding - you won't get chroma stutter.
but with normal lossless YV12 codec, you'll get chroma stutter on interlaced video, because they are subsampling progressively!
vigi_lante
6th January 2005, 15:49
Damn, the quality leaving the video interlaced is much worse than when I de-interlace it. So, I guess that I will just try to performace a IVTC or something like that.
Hey, scharfis_brain! Let me please ask you a question: I'm trying to understand what kind of source I have. It's NTSC Cartoon captured from TV, so it's telecined. When the camera is fixed and only the characters move, I always see a repeating pattern of 4 clean frames followed by 1 interlaced frame. But, when the camera is scrolling vertically or horizontally, I see 3 clean frames followed by 2 interlaced frames (3:2 pulldown I think). So, I wonder what I need to consider to treat the video. If it really is 3:2 pulldown, if just a IVTC will do the job, etc
Thank you...
scharfis_brain
6th January 2005, 16:15
jep, then IVTC will do the job!
you have the PPPPI - Pattern on animated characters, cause they are animated at 12fps or even lower fps (eg: 8 or 6 fps)
so your first step has to be IVTC.
do your filtering afterwards.
acrespo
6th January 2005, 17:15
I always had a doubt but not ask before.
When I do a IVTC the result video is totaly deinterlace?
I am asking because when I do IVTC I always have combs and need to apply a deinterlace method as post processing.
TIVTC + PP=7 for example.
scharfis_brain
6th January 2005, 17:18
I am asking because when I do IVTC I always have combs
then, you video is not telecined Film.
vigi_lante
6th January 2005, 17:30
Hum, thanks. But then, I would like to know, when configuring Decomb for example, if I need to assume that the video is 3:2 pulldown or not. Because 90% of the video is PPPPI, and 10% is 3:2 (PPPII).
Do you know any special settings that would be usefull dealing with only pure PPPPI ? Because I could specify manually an determined range of frames that there is only and pure PPPPI (with a special configuration that would handle PPPPI 4:1) and an determined range of frames that there is only and pure PPPII (with a special configuration that would handle PPPII 3:2).
Sorry if I did not make myself clear, english is my second language...
scharfis_brain
6th January 2005, 17:43
the 4:1 pattern arrives if two consecutive frames of the film (24fps) itself are identical, and then being telecined to 30fps.
so just set up a normal IVTC to 23.976fps and verything will be fine.
vigi_lante
6th January 2005, 17:55
So, this for example...
Telecide(order=0, post=0)
FDecimate()
...would be enough ?
I also capture some animations that have a pattern of 3:1 followed by 10:1 (that repeats itself several times, like 3:1->10:1->3:1->10:1)
A normal IVTC would also deal with it ?
scharfis_brain
6th January 2005, 18:17
probably. but this will need some source examination.
(counting how many unique frames are telecined into how many video-fields)
acrespo
6th January 2005, 18:30
then, you video is not telecined Film.
But my source is anime! As I know all anime is telecined.
I am capturing from VHS in composite input of my TV Card. The format is PAL-M.
Why Decomb and TIVTC include PP parameter if all videos telecined don't have residual combs?
scharfis_brain
6th January 2005, 18:40
have a look on the searches for
hybrid anime blending
vigi_lante
6th January 2005, 21:30
My main concern is that I want to avoid, at all costs, de-interlace artifacts on the video (like aliasing, shimmering, loss of linear details etc)...
I don't want that de-interlacing affect the video quality. I would like to keep all progressive frames with the same quality as before.
So, do you think that a IVTC, like this one...
Telecide(order=1,post=0)
Decimate(cycle=5,mode=2)
...with my kind of video, will preserve the quality ? Because if after IVTC the result video is totaly deinterlaced and progressive, I wonder if there would be any side effect.
Guest
8th January 2005, 03:12
The postprocessor in Decomb is designed to affect only those frames that come through IVTC still combed. The Decomb document describes many of the reasons why some frames can come through still combed. Thanks to this fortuitous behavior of Decomb, which is still unique as far as I know, your good progressive frames will not be degraded. Also, you can use Decomb's show or debug output to see which frames get deinterlaced. That might reassure you about your settings.
To receive reliable advice on this clip, you should make a short section of it (the unprocessed source) available for us to inspect. Post a link to it and we will be happy to do that.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.