Avisynth Smart deinterlace vs VirtualDub deinterlace blend

[thefox149]

I have capped a VHS which appears to have interlacing lines (I think they are whats known as progressive not sure capping soecs are 352x576 pal)

When I use the avisynth smartdeinterlacer...it only reduces minimally (I check this when I open the AVS file in virtual dub) I have compared this to
the virtualdub filter deinterlace blend best. and it removes it entirely.

My questions

1. Am I using the right deinterlacer in avisynth..
2. Am I even making sense
3. Virtual dub is slower than avisynth,......is there a better deinterlacer for
avisynth than the standard one that come with avisynth ....if so which one

[thefox149]

Do I even need to de -interlace?
From what I have read in the forums...no .....but people still do it.
VCD can't be interlaced ...

[D_Knife]

If your capping at 352x576 and your producing VCD then sooner or later you need to reduce your vertical resolution so this effectively de-interlaces for you.
However most people don't like the look of this and de-interlace anyway - but you knew this right.

You didn't mention which VirtualDub de-interlacing filter you are comparing to avisynth de-interlace filter - there are at least 6 of them
you can also get a smart de-interlacer for VirtualDub

[DJRumpy]

Converting to VCD will do a perfect deinterlace as a by product of reducing the vertical resolution by half. It will be much cleaner than any deinterlace run beforhand. Deinterlace efforts will try to blend the to fields, giving a distored view when you reduce to VCD resolutions. Reducing the vertical simply keeps one field. Since your already going to have to reduce by 2, your going to lose half your detail no matter what you do.

Just resize to VCD resolutions, and don't worry about deinterlacing.

To answer your questions regarding the difference between the two types of deinterlacers, the Smart deinterlacer will try to identify frames that appear to be interlaced, and deinterlace them. The VDub deinterlacer simply trys to deinterlace everything. Since your source is noisy (analog VHS), the Smart deinterlacer will have more problems unless it's tweaked to better identify frames that have combing and frames that don't.

In either case, you don't need the filter, as it's done automatcially when you convert to VCD.

That said, you should at least check to see if your VHS capture was telecined. If it was, you should perform IVTC on it first, to remove extraneous duplicate fields first. This will essentially give you a true progressive source to begin with, with no interlaced frames, and a major reduction in the number of frames to be encoded, improving quality as a result. Of course your source has to be telecined. If it's not, do not perform IVTC. You can find more info here ( http://www.lukesvideo.com ). It will show you how to spot a telecined video.

[thefox149]

my end result will be CVD..Sorry I wasn't clear on that what I meant was that VCD was not interlaced no matter what.

I suppose me question should have been would it matter if I deinterlaced with and end result of CVD

[DJRumpy]

It all comes down to one simple question. Is your CVD for TV viewing? If so, then don't bother de-interlacing. Deinterlacing will only degrade the quality by doing so. The only time you should consider deinterlacing is when you plan to view your video on a PC display.

CVD supports both interlaced, and progressive sources, so your ok either way.

[thefox149]

Thanks DjRumpy

[FOO]

Hey RJRumpy, Does IVTC necessarily produce a progressive result ?
It seems to me that it wouldn't have to.

[DJRumpy]

If a video is telecined, and you perform IVTC on it, then yes, the output should be progressive. The only exception would be video that has sections that are true interlaced. These 'Hybrid' clips are the bane of video hobbiests everywhere, as they throw a wrench into our efforts every time.

Most video is pretty consistant though. If was originally 23.976 fps, and it was telecined for DVD, VHS, or broadcast, then you should perform IVTC before encoding it to MPEG.

[bpjenn]

Even if you playback CVD (MPEG2) on a computer only, you probably don't need to encode deinterlaced since most, if not all now, software players (ex: Cyberlink PowerDVD) will deinterlace the video upon playback. The playback will be much better quality than if you had deinterlaced upon encoding. I've tested this thoroughly. I used to deinterlace and then discovered that it was not needed since the player did it anyways. Of course, with other file formats (avi, mpeg1) you will need to deinterlace or capture/encode at lower resolution (320/352x240).

[FOO]

OK I guess I have to do experiments now.
My understanding is that IVTC removes fields. Two of ten.
It has to separate the frames into fields first because
a telecined frame contains fields from different film frames
I don't see why that forces the video to be frame based
Why couldn't it be either way, 8 fields or 4 frames

The assumption here is that an AVI or mpeg2 can be a stream of pairs of
xxx by 240 fields or a stream of xxx by 480 frames. I actually
don't know. Isn't mpeg1 limited to frame based ?

When an MPEG2 encoder asks if you want to "encode interlaced"
what does that mean ? It implies to me that if you say no, the video
is encoded as a stream of fields. And that is the case whether or not
the video is 24fps film or 30 fps telecine.

[DJRumpy]

It can be either. The IVTC filter simply processes the video field by field. All video is made up of fields, so it's a simple matter to read it in and process it that way.

As to the encoder setting for interlaced video, that would be encoder specific. That said, I would guess that it would optimize it's settings for encoding interlaced video. It would change the Block Scanning order to 'Alternate', to handle interlaced video. It would not flag each frame as 'progressive'. Again, these are just guesses. You would need to refer to the documentation that came with your encoder.

[FOO]

I'm afraid I have to take issue with the statment
"All video is made up of fields"

My capture software has a setting for "Progressive source"
What do you suppose thet means ?

What does a Progressive scan DVD player do ?

[DJRumpy]

Progressive video is also made up of fields. An upper, and a lower field. The only difference being that both fields are from the same point in time, and they are both displayed at the same time. How else do you think they copy individual fields to telecine a progressive video (23.976 fps), to NTSC (29.97 fps)?

[FOO]

Well, they could take every other line from a frame.

It's really a moot point anyway. I think I was just concerned about
how it is stored. I don't know why it matters actually.

If I really cared I'd go look up the data stream format . I
think I have the details of MPEG2 program streams somewhere.

Sorry to be so argumentative, I have nothing to do today
but cause trouble.