More Newbie questions

[o2xygen]

I have recently bought a new Panasonic GS75 PAL camcorder... Its 3ccd however it records interlaced...
I usually split my Video and Audio and I work individually and then use DVDLab to Author.

But I want to achieve better picture quality. I have read many things here about some avisynth scripts that deinterlace your footage etc (like reinterpolate) with amazing results. However I am not very familiar with VDUB and avisynth. I mostly use AEffects for my editing.

Now, If I use reinterpolate to deinterlace my footage, will i have any image degradation?
Also If i deinterlace it, then CCE will have better results (as many said that CCE is not good for interlaced footage)

Also, is AE recommended for Denoising, Deinterlacing etc (can be achived with several plugins like Magic Bullet).? Or should I use any scripts?

thanks


Also, is if you could choose, what will this be:
Converting DV to MPEG-2 with Premiere Mainconcept encoder?
Converting DV to MPEG-2 with "bb" method with CCE (the sticky thread in DV forum under converting DV-->DVD)?
Converting DV to MPEG-2 with DVD2DVD using D2SRoBa method and CCE?

[o2xygen]

also why is my exported video always darker than the original.? In Virtual DUB or After Effects is always brighter. With Mainconcept Dv codec

[theReal]

If your source is interlaced and you are making DVDs for the tv screen, leave it interlaced! Deinterlacing always means 100% recompression of the video, it reduces overall quality (not to speak of the difficulties you'll run into when you deinterlace truly interlaced footage)

I am not a fan of progressive 25fps video anyways, even if shot in true 25p mode. A standard tv screen is not made for progressive video and IMO progressive footage on the tv screen always looks like 1970's tv news (at the time when they were still shooting the news on 16mm film).
I have tried it with a Panasonic DVX 100AE camcorder (probably the best SD Mini-DV camcorder available) and the results don't look like Hollywood, they look, as I said before, like old 70's tv news... (to make it look like Hollywood you'd have to do the hell of a lot more lighting and postproduction)

[o2xygen]

So there is quality impact... :thanks:
But If your footage has a lot of horizontal lines in it then deinterlacing is the only way to go, right? Its comes handy at these occasions

I will make a DVD test on TV with the normal interlaced video and the same video progressive to see how much is the loss of quality..

[theReal]

But If your footage has a lot of horizontal lines in it then deinterlacing is the only way to go, right?
Are you referring to "combing artefacts"?
Your footage always has a lot of horizontal lines (576 horizontal lines for PAL or 480 lines for NTSC to be exact), that's normal - but if you start to see the lines in "combing artefacts" (looks like a comb at the edges) then your tv is showing progressive images. Then, of course, you have to deinterlace! Is your tv a HD tv set that shows progressive video? Or are you watching the video on a computer screen?

[mic]

If you want to use AviSynth &/or V/Dub, there are filters available for de-noise etc. My personal experience is that you'll get faster, better results then using really anything else I've come across, but I haven't used Magic Bullet so I can't offer a direct comparison.

Agree with theReal on interlacing -- a case can be made for NTSC ivt where the reduced frame count allows less compression, but at 1 fps difference, don't know it's worth the bother.

RE: darker... Are you doing anything with color spaces, maybe in default settings?

[o2xygen]

Are you referring to "combing artefacts"?
Your footage always has a lot of horizontal lines (576 horizontal lines for PAL or 480 lines for NTSC to be exact), that's normal - but if you start to see the lines in "combing artefacts" (looks like a comb at the edges) then your tv is showing progressive images. Then, of course, you have to deinterlace! Is your tv a HD tv set that shows progressive video? Or are you watching the video on a computer screen?

I don't really know what is named.Something like this here
http://www.100fps.com/crusty2.jpg
It happens every time I pass an animation through Virtual Dub (encoded in DV) without adding the deinterlace filter. After that, it is removed at some degree
I am watching on Computer Screen. My TV is relatively old

[theReal]

that are "combing artefacts" caused by two interlaced fields merged into one frame.

a case can be made for NTSC ivt where the reduced frame count allows less compression That only applies to real FILM footage (originally 24fps), where the intermediate, interlaced frames can be discarded. With video footage, you'll get 29.97fps without any intermediate frames, you can't reduce that footage to 23.96 fps!

If you're watching on a computer monitor, then deinterlacing is recommended - but you can also deinterlace on the fly with software DVD players or with ffdshow and keep all picture information in case you want to watch it on a tv screen eventually.

100fps.com also shows how to make a 50/60fps video (for PAL/NTSC). It looks better on a computer monitor than deinterlacing to 25/30fps progressive.

[o2xygen]

RE: darker... Are you doing anything with color spaces, maybe in default settings?
I have found what the problem was. I output the movie to my TV and the whites were bright.
I messed up with the outputs of MPC (from Overlay mixer to VMR9) and the white is bright as the original... I was starting to become dissapointed because I was doing something wrong... :D :D :D :D :D :D :D :D

[mic]

I have found what the problem was.
GREAT! :D

That only applies to real FILM footage (originally 24fps), where the intermediate, interlaced frames can be discarded. With video footage, you'll get 29.97fps without any intermediate frames, you can't reduce that footage to 23.96 fps!
hmmm... politely as possible, think you're right and wrong. :D From a more purist standpoint you can't take away what was never there, so you're right.:) But then with ivt it isn't feasable to go back to 24 p really accurately and painlessly, unless you're just removing pulldown flags from mpg2.

You can use ivt on 29.976 i.
Many V/dub & AviSynth filters work on what they perceive as the results of interlacing, pulldown etc... Why blend two fields completely losing speed and maybe quality, when you can work on only the difference? Running ivt & maybe deinterlace on DV or any pure intelaced video, you get a smaller file with some frames removed. Encoding to mpg2 as 23.976 or 24 p adding pulldown, on playback can approximate film. There will be some quality loss, but there will also be a difference in the amount of compression, which is quality not lost. :) Some folks prefer the look.

If you're watching on a computer monitor, then deinterlacing is recommended
Why? If you're not recording to CD or doing web streaming where size is critical...

- but you can also deinterlace on the fly with software DVD players or with ffdshow and keep all picture information in case you want to watch it on a tv screen eventually.
Exactly :) You also don't have to worry about results of deinterlace showing as blurred frames, and get smoother playback as you've noted. Also don't have to rely on DVD player or ffdshow -- wmplayer, real player etc., most anything now days will handle it,

100fps.com also shows how to make a 50/60fps video (for PAL/NTSC). It looks better on a computer monitor than deinterlacing to 25/30fps progressive.
FWIW Microsoft has docs on 90. Ask any gamer and fps rules -- not always just because they can. :D

[o2xygen]

is it the same:
converting interlaced material to 25progressive material (magic bullet)
and
deinterlacing interlaced footage to progressive scanl using for example the deinterlace-smooth filter for VirtualDub

[theReal]

Running ivt & maybe deinterlace on DV or any pure intelaced video, you get a smaller file with some frames removed.I don't think you'll ever end up with smooth video like this. Any frame that is removed will leave an empty space, the video jumps to the next frame. On FILM sources, only those frames are removed which were artificially inserted in the telecine process, the time between the remaining frames is always the same (--> no jumps)

is it the same:
converting interlaced material to 25progressive material (magic bullet)
and
deinterlacing interlaced footage to progressive scanl using for example the deinterlace-smooth filter for VirtualDub Magic Bullet also deinterlaces the video, but I don't know what method it uses. There are several methods of deinterlacing. The simplest being discard one field. interpolate fields and blend fields. Then there are more sophisticated algorithms that examine the video and decide where and when to blend/interpolate. There are quite a few of the more sophisticated filters available for avisynth (and these free filters are often better than expensive software from big companies!)

[mic]

"deinterlacing interlaced footage to progressive scanl using for example the deinterlace-smooth filter for VirtualDub"

As theReal mentioned, there are several different V/Dub filters that work on noise or artifacts of one sort or another. If you tweak these, which can be a bit of trial and error, you can get results extremely close to video that's never been interlaced. Some feel that the filters available for AviSynth exceed even this. Either way, stacking and ordering filters is possible, and in V/Dub at least, feedback near immediate.

I don't like simple de-interlace in prog. like Sony or Adobe, because 1, the extreeeeee...mly slooooooo....w speed of rendering, and 2, the increased amount of blurring comparred to filters that just fix the problem, don't try and merge/blur parts of the frame where fields are identical. If you look at individual frames, you can see the interlacing artifacts, and see how it seldom effects the entire frame.

RE: ivt: "I don't think you'll ever end up with smooth video like this. Any frame that is removed will leave an empty space, the video jumps to the next frame."

Film takes snapshots, and telecine adds some in-between frames. 29.976 i video takes the same sort of snapshots electronically -- just more often, so frames aren't manufactured. Ignoring the appearance difference between video and teleconed film, the total motion recorded, and the fps is the same. Evenly take out frames to reach 23.976, the same holds true -- same total motion and same fps.

I think running video through ivt is actually easier & can work better: video records a set fps, so you just have to remove the standard pulldown frames, or just decimate frames evenly. With telecined film OTOH, the added tweeners are not an actual snapshot of anything, so those are the frames you want to throw away, but 1st you've got to find them. ;)

Either way you've got fewer frames, so you are losing something. And interlaced video is interlaced video, so if you're after a pure film look, further deinterlace is not going to get you all the way there. Where you gain is in a smaller total frame count and file size, allowing less final compression. This was/is a very big deal for me with SVCD.

Some like the slower fps for appearances, but another place I've personally found it useful is when enlarging interlaced video. If/when deinterlace leaves some blurred motion that shows up on playback, adaptive ivt followed by encoding with pulldown can make an improvement. It may not be as smooth as the original, but it can be a decent tradeoff.

[theReal]

Film takes snapshots, and telecine adds some in-between frames. 29.976 i video takes the same sort of snapshots electronically -- just more often, so frames aren't manufactured. Ignoring the appearance difference between video and teleconed film, the total motion recorded, and the fps is the same. Evenly take out frames to reach 23.976, the same holds true -- same total motion and same fps.I can't believe that IVTC works on truely interlaced video. I mean, the time distance between two video frames is always the same, now if you take out some of the frames, then the real-time interval between the remaining frames is larger (2 times 1/29.97 second) than between the other frames, but it is shown in an interval of only 1/23.96 seconds for every frame - the video should be stuttering every second or so.

When you delete telecined frames from film, the real-time interval between the remaining frames is 1/24 second, the same as between all the other frames, so it will run smoothly.

Is there some trick you do that compensates for this?

[mic]

:) Don't think film, think rulers :)

Film a scene with PAL DV, & film the same scene at the same time in NTSC. Both show the same thing, but PAL has 1 more fps then film.

Another way to look at is with a 10 minute clip in an NLE. Change the ruler from 24 to 25 to 29.976. The video itself doesn't change, just the ruler, & the number of divisions on the ruler between start and end. :) (In this illustration, don't think of rendering changes -- just what you see)

:( Why it shouldn't work is because you're not going to 15 fps :(

That would be the only way to remove frames and preserve inter frame spacing evenly. But with 60 fields per second going on at the TV, the amount of time based distortion is no more then that added to 24 with telecined film. Adding to 24, subtracting from 29, neither is right but either works. :D

I have used is very successfully in SVCD where the difference from less compression is really helpful. On a DVD I don't use ivt or deinterlace because you can notice either when it creates a bad couple of frames (if your eyes work that way -- most don't), and space isn't as critical. Enlarging interlaced on the other hand (particularly to go w/s) has always been a pain for me, because you emphasise anything a deinterlace pass misses. Adaptive ivt in V/Dub takes out the worse offenders, and overall helps.