What's the best deinterlacing method for VHS video for burning to DVD?

[Sharc]

Here your interlaced DVD authored with DVDStyler, with the denoised video (without audio for this demo):

What denoising filter did you use?
It looks better than I expected for interlaced video, and even Yadif de-interlaces the denoised version pretty well.

Here the script I have been using:

AviSource("Original Sample.avi")  #or use your preferred source filter
AssumeTFF()
QTGMC(preset="fast")  #double-rate deinterlacing for subsequent non-interlaced-aware denoising filter
MCDegrainSharp()   #denoising
SeparateFields().SelectEvery(4,0,3).Weave()   #re-interlacing for DVD PAL compliance
ConvertToYV12(interlaced=true)   #prepare for the final mpeg2 encoder of DVDStyler

The script can be directly loaded and executed in DVDStyler.

[hello_hello]

Bwaak,
When I referred to 25fps, I was referring to 25fps being the only progressive-scan format natively supported for a PAL DVD. Likewise it'd be 30fps progressive for NTSC, at least as far as I know. https://en.wikipedia.org/wiki/DVD-Video#Video_data

Other frame rates can be wrapped into the PAL or NTSC field rates using repeat flags, as you said. I'm not sure why DVD Architect wouldn't accept a 24fps video, given 2:3 pulldown is such a common thing.

24fps can be wrapped into PAL's 50 fields per second with a 2:2:2:2:2:2:2:2:2:2:2:3 telecine pattern instead of speeding it up, but I think the only time I've seen it used has been for DVDs containing concert footage etc, where it'd be undesirable to increase the frame rate due to having to increase the audio speed and/or pitch to match.

I'm not sure about the distinction between 24F and 24P, although the former sounds like it might be a progressive segmented frame.

Here's a DVD review from 2002 when progressive scan DVD players were all new and shiny. It shows an example of the DVD player switching from ignoring repeat flags to de-interlacing hard telecine instead of applying pulldown removal. At least I think that's what the problem was.
https://hometheaterhifi.com/volume_9_2/mrg-87-june-2002.html#A%20Beautiful%20Mind

[Bwaak]

I'm not sure why DVD Architect wouldn't accept a 24fps video, given 2:3 pulldown is such a common thing.

It is the other way around: it accepted it, and did not re-encode it for DVD, used as is. On the other hand, it did re-encode my native 25p MPEG-2 file.

I'm not sure about the distinction between 24F and 24P, although the former sounds like it might be a progressive segmented frame.

No, one is native (or interlaced with flags, so treated as native), another is 24-in-60 without flags, so treated as interlaced. 30p and 25p were PsF. I don't think there have ever been a consumer camcorder that could do 24 PsF.

[Alwyn]

Can you lot stay on topic. Ferrari is obviously a newby and needs basics only. In my assessment, that won't be AVISynth.

There was no malice, and it was clearly marked as blunt.

You need to reset your levels. That was out and out abuse. If you've nothing nice to say, don't say anything. Ironically, your ranting probably achieves the opposite; you turn people away and are ignored. I certainly wind back your comments by 90%.

[Alwyn]

DVDStyler: Wow!

Here's my attempt. I deinterlaced, cropped and resized to 768x576 in VDub2 (as per my guide here), saved as a Lagarith AVI (for something different), then imported it into DVDStyler. No denoising or other enhancements.

After download, change the extension from ZIP to ISO, then right-click>Open With VLC Player. I'm not at my big TV so can't see the result on it but it looks ok on my 'puter.

The forum should enable ISO uploads.

[Skiller]

Looks alright, but it's deinterlaced to 25p.

These kinds of shaky home videos need 25i or 50p to look like they do on tape. Personally I wouldn't want to watch more than a minute of it on a TV juddery like that...

[Alwyn]

These kinds of shaky home videos need 25i or 50p

I gave DVDStyler a 50fps deinterlaced AVI but it converted it back to 25fps.

Giving it an Interlaced AVI or MPEG 2 (from handbrake) didn't work well; the interlacing looks like it got burned-in on the ISO.

Personally I wouldn't want to watch more than a minute of it on a TV juddery like that...

Agree, although the camera is pretty close to the action.

[jagabo]

DVD doesn't support 50p or 60p. Only 25i, 25p, 30i and 30p.

[Bwaak]

DVD doesn't support 50p or 60p. Only 25i, 25p, 30i and 30p.

One would expect the authoring software to convert 50p into 25i or to at least flash a warning.

[Sharc]

DVD doesn't support 50p or 60p. Only 25i, 25p, 30i and 30p.

Yes, according to DVD specs (Although I doubt that hobbyist do have these for USD 5'000 and signing an NDA).
I have seen commercial "NTSC" DVD's 30p flagged as progressive, but I have never come across commercial 25p DVDs flagged as progressive. Progressive footage has always been in the 25 PsF format and flagged as interlaced. 25p progressive flagged "PAL" DVD's were all home-brewed. This only reflects my limited experience of course, and might be wrong. I didn't analyze all Bollywood productions ...
I think the label "NTSC" or "PAL" on the discs and packages should actually exclude progressive encoding+flagging because "NTSC" and "PAL" are TV broadcast standards which are interlaced only.

[Sharc]

Looks alright, but it's deinterlaced to 25p.

These kinds of shaky home videos need 25i or 50p to look like they do on tape. Personally I wouldn't want to watch more than a minute of it on a TV juddery like that...

Absolutely. Why butcher it by deinterlacing to 25p (and 50p is not DVD compliant).
The main problem of Ferrari420 was the wrong field order (initially), and that he wanted to put 2hours29minutes of noisy and shaky home-video capture footage onto a DVD-5, encoded as mpeg2 for DVD compliance. The macroblocking festival (even Canopus Procoder would have a problem with this) or bitrate peaks by unconstrained encoding (poor or missing VBV control by the encoder) causing stutter became unavoidable.

[Sharc]

DVDStyler: Wow!

Here's my attempt. I deinterlaced, cropped and resized to 768x576 in VDub2 (as per my guide here), saved as a Lagarith AVI (for something different), then imported it into DVDStyler. No denoising or other enhancements.

Why all this complication? Why resizing to square pixels? Just to let DVDStyler resize it again to non-square (DVD compliant) pixels?
You can crop directly in DVDStyler btw. You may have to pay attention to do the cropping interlace compliant though. (Maybe DVDStyler does it correct by itself, I didn't test).

[Anonymous5394]

Leave interlace as is. Deinterlacing most often results in more trouble!

[Anonymous5394]

[QUOTE]To be blunt, in order to be as clear as possible here:
If you want to make shitty quality video, you can certainly do so. I just hope I never have to watch your butchered video. In fact, I won't watch such crap, I'll move on to something else. That's how most people feel, though some will "be nice"/coddle your feelings about it, give the lip service, then turn it off when you're gone/QUOTE]

I bet the guy is glad now he never clicked on and asked for help from Lord S. That said sometimes you have to be cruel to be kind eh. Yo Ho Ho!

[Alwyn]

Why butcher it by deinterlacing to 25p (and 50p is not DVD compliant).

Because I thought a nice progressive source would be better than interlaced. You said that you put your QTGMC script straight into DVDStyler. You're giving it a double-rate progressive source, are you not? What's the difference with my progressive input then?

Why resizing to square pixels?

Because when I gave DVDStyler the original AVI, it messed up the aspect and pillarboxed the video slightly; the DVD output didn't expand from 720 to 768 (4:3 full screen). When I gave it 768x576, the DVD correctly displayed the video as 4:3 full screen.

I note you didn't resize in your script; why did DVDStyler correctly resize that and not the original AVI?

[lordsmurf]

I'm not sure what people expect to see from home videos made with an 80s camcorder filmed in the 90s that have been degrading for the last 20 - 32 years in heat, light, humidity, and electromagnetic radiation.

99%+ of the time, what you just stated is BS.

All of my videos suffer from pixelation which is probably due to my Panasonic DMR-EH55,

Correct.

that price scalpers are now selling for $300 (I got mine for $30 and was incredibly lucky).

Such a ridiculous statement. MSRP on those was $300+, and it takes effort and money to make them still work properly 15 years later. Your random thrift store find is not a barometer for costs.

If that looks like crap then there's little that I can do about it.

You can do plenty, with effort, with some budget. Thus far, you choose not to. That's fine, your choice. But don't get pissy when it gets pointed out to you.

... And no response. Just wanted to denigrate my video and call it "crap" to make them feel better about themselves because they have more money to spend on this than I do. Just what I expected. Video snobs. Nothing to offer except criticism.

It's the holidays. And you are not my priority, I don't owe you anything.

DVD supports progressive video but only at 25fps.

I am not sure about that.

No, it's 100% correct for PAL.

There was no malice, and it was clearly marked as blunt.

You need to reset your levels. That was out and out abuse. If you've nothing nice to say, don't say anything. Ironically, your ranting probably achieves the opposite; you turn people away and are ignored. I certainly wind back your comments by 90%.

Oh, FFS.

Must we all walk on eggshells, in order not to hurt the feeling of people that are making mistakes, spout false info, are not cooperative, and then cop an attitude when called out on it? Really?

No.

[Sharc]

Why butcher it by deinterlacing to 25p (and 50p is not DVD compliant).

Because I thought a nice progressive source would be better than interlaced.

Why? Usually not by reducing 25i to 25p for shaky home video. DVDStyler was just smart enough not to output 50p for DVD compliance, so you ended up with the 'stroboscopic' 25p. Technically doable, but not recommended.
(If you absolutely wanted 25p you could add motion blur using QTGMC(.....) in order to reduce the stroboscopic look.)

You said that you put your QTGMC script straight into DVDStyler. You're giving it a double-rate progressive source, are you not? What's the difference with my progressive input then?

Read and understand my script in post#31. I added comments to every line which should have explained the purpose. After the denoising filter I reinterlaced the video to preserve the 25i motion. Ask if not clear.

Why resizing to square pixels?

Because when I gave DVDStyler the original AVI, it messed up the aspect and pillarboxed the video slightly; the DVD output didn't expand from 720 to 768 (4:3 full screen). When I gave it 768x576, the DVD correctly displayed the video as 4:3 full screen.

You may not have used DVDStyler's cropping/padding feature correctly? Right click the .avi in the assets to open the corresponing menus. DVDStyler will encode with the correct PAR's which should warrant a geometrically undistorted output. You wouldn't need the VirtualDub2 intermediate steps, unless you want to add special filters.

[lollo]

AssumeTFF()
QTGMC(preset="fast")  #double-rate deinterlacing for subsequent non-interlaced-aware denoising filter
MCDegrainSharp()   #denoising
SeparateFields().SelectEvery(4,0,3).Weave()

Yes, that's the correct procedure to denoise an interlaced video and interlace back using a spatial-temporal denoiser.
Because one the frame generated by the deinterlacer is discharged, and to be more accurate on the source processing, you can use QTGMC(lossless=1) instead of QTGMC(preset="fast"), which preserves intact the original field.

[victoriabears]

Funny, I have aload of DVD's made from vhs and they are all progressive, when i read they would typically be interlaced??

[Bwaak]

Funny, I have aload of DVD's made from vhs and they are all progressive, when i read they would typically be interlaced??

Garbage A/D converter generating 25p from 25i? Are you using the Clearclick thingy? OTOH, if you are digitizing feature movies, then 25p is what they usually are, although you still may be missing half the resolution.

[victoriabears]

Thanks they are provided by on eof those rare TV show providers, the quality is good to very good, am just usin ghandbrake to get them to mkv, again pleased overall. And you could be right that they were created using simple straightforward a/d converters of some sort.

[Ferrari420]

On the denoising strength section in the configuration of MSU Denoiser what does hardness and softness change? With full softness and full hardness the result I see is the same. There is only a difference if the MSU Denoiser filter is turned off and on.

[DB83]

Well I have been away for some days.

At least we now have a sample. So thank you for that. Whilst it may not make any difference, you were also asked for the detail of your entire capture workflow. Who knows, the actual capture device may be the 'villain' here.


But I would agree that 2hours 30 is too much for a single-layer dvd however you swing it. But you can still get reasonable results at 2 hours.

[Sharc]

...But you can still get reasonable results at 2 hours.

Reasonable results? Hmmm.... With re-encoding already compressed DVD footage yes, but more difficult and demanding for shaky and noisy VHS home video captures. May require too much denoising producing wax pictures and/or annoying encoding artifacts. It's still mpeg2 only for DVD. Better chances would be with AVC/HEVC/mpeg4.

[Skiller]

>120 minutes of shaky VHS home videos on a DVD5 needs half D1 resolution (which is really not that big of a hit with VHS).
With 80-120 minutes at full resolution, ProCoder would give good results.
Below 80 minutes almost anything looks good.

[lordsmurf]

...But you can still get reasonable results at 2 hours.

Reasonable results? Hmmm.... With re-encoding already compressed DVD footage yes, but more difficult and demanding for shaky and noisy VHS home video captures. May require too much denoising producing wax pictures and/or annoying encoding artifacts. It's still mpeg2 only for DVD. Better chances would be with AVC/HEVC/mpeg4.

Just for the sake of adding to this side discussion, masking overscan always helps, NR helps, and halving to 352 Half D1 helps. Bitrate is not a pure measure, but one of allocation. You can easily fit semi-noisy Half D1 masked with some NR to achieve a watchable DVD.

That's not at all what we have in this thread ... but it's completely possible.

(You, selur, surely know this. I'm educating anybody else that does not yet know how MPEG encoding works, or the DVD-Video specs.)

[Alwyn]

On the denoising strength section in the configuration of MSU Denoiser what does hardness and softness change?

I got a reasonable result with Preset set to "High Noise MSU" and both sliders set to 100. note that you need to reset the Preset each time you open the filter.

There is only a difference if the MSU Denoiser filter is turned off and on.

That's expected and means the filter is working.

Another filter which may help is Camcorder Color Denoiser. Down the page a bit.

[Ferrari420]

Well I have been away for some days.

At least we now have a sample. So thank you for that. Whilst it may not make any difference, you were also asked for the detail of your entire capture workflow. Who knows, the actual capture device may be the 'villain' here.


But I would agree that 2hours 30 is too much for a single-layer dvd however you swing it. But you can still get reasonable results at 2 hours.

Samsung DVD-V70 6 Head DVD/VHS Player - via RCA -> Panasonic DMR-EH55 DVD Recorder - via SCART to S-Video adapter -> Diamond VC500 -> PC

[Sharc]

Samsung DVD-V70 6 Head DVD/VHS Player - via RCA -> Panasonic DMR-EH55 DVD Recorder - via SCART to S-Video adapter -> Diamond VC500 -> PC

As you have to use the 'Composite OUT' (RCA) of the V70 VHS player make sure to enable the Comb Filter in the EH55.

[Ferrari420]

The comb filter is on on my EH55