Well, VHS capturing Mostly cartoons, having some questions:)

[zerowalker]

yeah noticed that, but you always learn something new!

[zerowalker]

When is the source interlaced that needs deinterlacing then?
i tried a VHS movie with a part that was interlaced, but it worked both with QTGMC() and TFM(), which makes me confused.

[manono]

If it's a movie then it's telecined 23.976fps. That is, you'd need a full IVTC of TFM().TDecimate() to return it from the interlaced 29.97fps. Are you in NTSC land? If not, it's not telecined. If for no other reason than the much faster encoding speed, field-matching is preferable to deinterlacing with QTGMC. However, with VHS tapes, I can see how QTGMC might give a more pleasing result as it has some noise removal built in, which VHS tapes usually need.

To answer the question, it's interlaced if shot using video cameras. You separate the fields or you bob the video, scroll to a place with movement, begin advancing a frame at a time in VDub(Mod), and every field (or bobbed frame) will be different. Each field is from a different point in time. With progressive sources each field will be pretty much identical to its partner.

[zerowalker]

PAL country here

Okay, but so, it´s only video cameras in the long run which should "need" QTGMC?
otherwise it´s give and take for the noise removal?

But won´t QTGMC() fail on vhs and stuff if it´s like games, where it´s progressive, it will fail noticing color changes on occasions?

[manono]

There are lots of conditions under which you could use QTGMC, particularly if the time it takes to encode is no consideration. You can, for example, run it over a progressive source to smooth out aliasing. I use it for that quite a lot when end credits are flickering like crazy. It has a special setting for that. You can use it as a bobber for blend removal filters, like SRestore. And you can use it as a deinterlacer for video sources, with either single or double framerate results.

So, no, I wouldn't necessarily say only video sources might benefit from its use.

I haven't used it for gaming footage before (I don't cap games), but I do use it on VHS captured sources. My interests are in old and black and white movies from VHS, and I have no knowledge of it messing with the colors.

[zerowalker]

Hmm, true, QTGMC got some really nice feautures like noise reduction and other things that really does a good jobb.
So it´s up to the results in the end i guess


A question, is there a way to TFM() this, without it failing thanks to (uneven field matching??).

EDIT: Is there a way to capture in virtualdub, and make it work like fraps or, with it making a new file every 4gb, but maybe a new file every 2 min or something?

EDIT 2: seems like Spill Drive is the solution i was looking for

[jagabo]

A question, is there a way to TFM() this, without it failing thanks to (uneven field matching??).

I don't now what you're asking here. But your sample video has screwed up chroma channels.

[zerowalker]

If you use TFM() one frame will be messed up, with lines, looks corrupted, i think it fails so Deinterlace or what to call it.

It has?
How can you see that?

Oh maybe you mean the dots all over colored areas?
well it´s S-video that fails there.

EDIT: ehm, i asked before when to use QTGMC, and looked at an n64 capture, and there QTGMC looked way better, and with Converttoyv12(interlaced=true)
AssumeTFF()

it worked like a charm.

TFM() worked, but was edgy, (the capture was very noisy and stuff like that)

How come Chroma isn´t failing there, is it true interlacing?

[zerowalker]

Along side the question above, i was wondering, how much noise does Component have if you compare it with composite and S-video (though i only get bad s-video though)?

[jagabo]

If you use TFM() one frame will be messed up, with lines, looks corrupted, i think it fails so Deinterlace or what to call it.

I think you're talking about residual come artifacts on the grey wall. 2x point resize:



If you turn off post processing, TFM(pp=0), you'll see that it happens because an interlaced frame has gotten through the field matcher. But the post processor didn't recognize some of the combing in the grey wall and let it through. You can try playing with the different pp modes. TFM() also has a decombing threshold setting you can try playing around with, cthresh.

It has?
How can you see that?

Regarding the messed up chroma: apply a simple bob (or any of the smart bobs) and watch the yellow color in the windows. See how it moves differently than the grey walls. A crop from the top of the frame:

EDIT: ehm, i asked before when to use QTGMC, and looked at an n64 capture, and there QTGMC looked way better, and with Converttoyv12(interlaced=true)
AssumeTFF()

it worked like a charm.

Basically, you use QTGMC() when you want to deinterlace truly interlaced material. But people sometimes use it on progressive material because it also does a little noise reduction and antialiasing of edges. But you can use it any time you want.

TFM() worked, but was edgy, (the capture was very noisy and stuff like that)

How come Chroma isn´t failing there, is it true interlacing?

The chroma problem is visible with Bob(), TFM(), QTGMC(), Yadif(), any deinterlacer -- because it's in the source. I think an interlaced YV12 video was treated as progressive YV12 at some point.

I'm having some problems with your sample though. When I use AviSource() to open it I get an error message from AviSynth -- "couldn't locate a decompressor for fourcc ffds". So I used ffMpegSource2() instead. I get interlaced YV12 out of that. So it's possible ffmpeg is causing the chroma problem. What are you using to decompress the video? Do you see the same chroma problem pictured above?

Along side the question above, i was wondering, how much noise does Component have if you compare it with composite and S-video (though i only get bad s-video though)?

Component should have better resolution than s-video and composite. It should be free of dot crawl artifacts. Though S-video should be free of dot crawl artifacts too so I'm not hopeful it will help you.

[zerowalker]

Okay wait i will clip the same and make it lagarith to make sure everything is fine with the decoding part

And yeah have tried many kinds of settings with TFM() but it´s hard to make it deinterlace without faling on progressive parts with some parts.

Okay, good to know as QTGMC really is awesome at some points

EDIT: It seems to work with using pp=3/4, but it deinterlace the whole picture, so the text is affected, but maybe it´s supposed to be so

[zerowalker]

Along with the above, i must ask once more.
About Decrawling etc. I always get those, Lines, like ZigZag, but it wasn´t herringbone, it was crosstalk as Jagabo said ( i think)
So can someone direct me to a, filter solution for that in the meanwhile, as i probably can´t get S-video vhs and stuff soon

[jagabo]

Your latest cap no longer has those chroma problems.

[zerowalker]

Yeah it was an decoding issue

[zerowalker]

I am trying to solve the dot crawl, and the Color fading along it with no success.
Here is a restored sample, but with dot crawl pretty much all over edges.

[jagabo]

It's not possible to entirely remove dot crawl artifacts without damaging other parts of the picture.

[zerowalker]

Well pretty muched guessed so, as well the dot crawl is pretty heavy.
But will try the way of using a DVD as passthrough to S-video if i can just get some answers later on in the other topic.
As that is hopefully one way to solve it i hope

[zerowalker]

Does anyone know of resoration on DVD recorded VHS tapes, which pretty much are full of interlacing and artifacts?
I myself am trying out some stuff, pretty much QTGMC() and Motion Compenstation.

But maybe there is more?

[jagabo]

But will try the way of using a DVD as passthrough to S-video if i can just get some answers later on in the other topic.
As that is hopefully one way to solve it i hope

You mean using the DVD recorder's passthrough to split composite to s-video? That will only help if the luma/chroma splitter in the DVD recorder is better than the luma/chroma splitter in your capture device.

[zerowalker]

Oh;S
Thought it helped if i could give the capture device S-video as someone stated could be better for some reasons.
If not, then well, the end is the same that i will need an vcr with S-video, TBC can come later.

But please watch the mpeg2, and tell me what you think (sadly i chose wrong, they are not the same, but it´s the same frames though difference episodes)

[zerowalker]

Do someone know if there is a way to prevent the Fuzzines or how to put it, that seems to come from Motion Compensation script?

SetMemoryMax(700)
SetMTMode(5, 4)
MPEG2Source("F:\Desktop\Zelda test.d2v", cpu=4)
QTGMC()
SelectEven()
#DeHalo_alpha(rx=2, ry=1, darkstr=1, brightstr=1.5, lowsens=150, highsens=150, ss=1.5)
 # Motion compensation on frame -2...+2
super = MSuper ()
vf2 = super.MAnalyse (isb=false, delta=2, overlap=4)
vf1 = super.MAnalyse (isb=false, delta=1, overlap=4)
vb1 = super.MAnalyse (isb=true,  delta=1, overlap=4)
vb2 = super.MAnalyse (isb=true,  delta=2, overlap=4)
cf2 = MCompensate (super, vf2, thSAD=400)
cf1 = MCompensate (super, vf1, thSAD=400)
cb1 = MCompensate (super, vb1, thSAD=400)
cb2 = MCompensate (super, vb2, thSAD=400)

# dfttest on motion-compensated clip
Interleave (cf2, cf1, last, cb1, cb2)
dfttest (sigma=32, sigma2=32, tbsize=5, lsb=true)

SelectEvery (5, 2)

# Gradient smoothing
presmooth = last
SmoothGrad (radius=20, thr=0.15)
SmoothGrad (radius=12, thr=0.25, ref=presmooth)

# Back to 8 bits, ordered dithering

LSFMod(strength=80)
fslg_dither_to_n_bits (10, outn=true)
Dither_convey_yuv4xxp16_on_yvxx()
Distributor()

# The last part is for exporting to 10bit, other than than it´s the same as 8 bit:)

It´s the black that is like, i don´t know;S
it´s not Flat or something, don´t know the word, i think it´s from dithering, but not sure.

Hope you understand, if not will will try to explain better!

[zerowalker]

I have problem Deinterlacing a PAL DVD, i tried QTGMC with Srestore, but no matter how i use QTGMC i get ghosting.
Does anyone know why?

[jagabo]

Not all videos can be de-ghosted.

[zerowalker]

Yeah but i don´t get it with TFM(), but TFM fails on some frames, it doesn´t deinterlace etc.

And yeah, it´s cartoon, so i know that it may have Ghosting to make it "smooth" or how to put it. but it doesn´t seem to have it.

And QTGMC creates ghosting even on progressive frames on the source;S

[manono]

If you want answers about a source, don't ask questions; provide untouched samples from the source. But as mentioned, some sources are so crappy they can't be fixed.

QTGMC doesn't 'create' ghosting.

[zerowalker]

Here is a sample

And well QTGMC has created ghosting for me, when it fails to deinterlace, but TFM works then so it´s probably not interlaced i guess.

[sanlyn]

Not interlaced ? ?

[jagabo]

Here is a sample

And well QTGMC has created ghosting for me, when it fails to deinterlace, but TFM works then so it´s probably not interlaced i guess.

That video just has a weird cadence. Some of the original film frames last only one field, some two fields, some three fields, etc. I suggest you QTGMC() it an encode at 50 fps if you need it progressive. Or maybe, after QTGMC(), try TDecimate() in mode 2 or 7 to get some lower frame rate.

[zerowalker]

So it´s messed up interlace i guess, but do i really have to go with 50fps?
Is it really that many real frames?

Sanlyn, with that i mean it could be some meesed up interlace, not ordinary, which seems to be the case.

[jagabo]

So it´s messed up interlace i guess, but do i really have to go with 50fps?
Is it really that many real frames?

Sanlyn, with that i mean it could be some meesed up interlace, not ordinary, which seems to be the case.

The interlacing isn't messed up. It looks like the original was about 15 fps progressive. It has gone through some weird telecine pattern to get 50 fields per second. It may be hard to get it back to the original ~15 fps progressive frames without some duplicates and some dropped frames. Or it may be a mix of frame rates. So it's easiest to just encode it as 50 fps progressive. That way the cadence won't be any worse than the original DVD or broadcast.