Help with weird deinterlace

[eXtremeDevil]

I have a bluray source (indexed in a .dgi file), PAL, and it has a weird deinterlace: it has 5 interlaced frames and 1 progressive.

Which deinterlacing filter should I use?

And other thing, how to solve the "no function named McBob" message?

Thanks.

[lansing]

upload a sample clip and post the script that you use

[Didée]

5i + 1p is a funny pattern. That's 1 additional field per 11 source fields. Ratio 12/11 is 1.09, which is (roughly) *double* the ratio of 24<>25 fps.

To avoid the message about MCBob, you simply don't use MCBob.

[eXtremeDevil]

upload a sample clip and post the script that you use

OK, this is what I do:

I have the BD fully ripped to the PC.

This is the m2ts info:

Video
Count : 201
Count of stream of this kind : 1
Kind of stream : Video
Kind of stream : Video
Stream identifier : 0
ID : 4113
ID : 4113 (0x1011)
Menu ID : 1
Menu ID : 1 (0x1)
Format : VC-1
Commercial name : VC-1
Format profile : AP@L3
Internet media type : video/vc1
Codec ID : 234
Codec : VC-1
Codec : VC-1
Codec/Family : VC-1
Codec/Url : http://www.microsoft.com/windows/windowsmedia/format/codecdownload.aspx
Codec profile : AP@L3
Duration : 3070000
Duration : 51min.
Duration : 51min. 10seg. 0ms
Duration : 51min.
Duration : 00:51:10.000
Bit rate : 20722726
Bit rate : 20,7Mbps
Width : 1920
Width : 1 920pixeles
Height : 1080
Height : 1 080pixeles
Pixel aspect ratio : 1.000
Display aspect ratio : 1.778
Display aspect ratio : 16:9
Frame rate : 29.970
Frame rate : 29,970fps
Frame count : 92008
Resolution : 8
Resolution : 8bits
Colorimetry : 4:2:0
Chroma subsampling : 4:2:0
Bit depth : 8
Bit depth : 8bits
Scan type : Interlaced
Scan type : Entrelazado
Scan order : TFF
Scan order : Campo superior primero
Interlacement : Interlaced
Interlacement : Interlaced
Compression mode : Lossy
Bits/(Pixel*Frame) : 0.333
Delay : 11650.667
Delay : 11seg.
Delay : 11seg. 651ms
Delay : 11seg.
Delay : 00:00:11.651
Delay, origin : Container
Stream size : 7952346051
Stream size : 7,41GIB (86%)
Stream size : 7GIB
Stream size : 7,4GIB
Stream size : 7,41GIB
Stream size : 7,406GIB
Stream size : 7,41GIB (86%)
Proportion of this stream : 0.85663
Buffer size : 1843200

I put the m2ts in DGIndexNV and get a DGI file.

I don't have anything in the script but the source (DGSource).

Here is the m2ts sample, trimmed with DGIndexNV:

http://www.mediafire.com/?6m46ll9ytb4tr68

About the McBob thing, I'm actually trying to solve it to use it...

[Didée]

The source is crap. It's a 25-->30fps conversion with sliding blending (similar to ConvertFPS, but actually it seems somehow different), and afterwards it has been phase-shifted.

Contact the dealer, and sing the "gimme my money back" song.

[eXtremeDevil]

OK, and in a different, constructing way, how can I deinterlace the video? Any suggestions?

[Didée]

Match the fields with just TFM(). After that, you can enjoy the blending crappiness in all its glory.

There is nothing more that could be done. The blending is done in a funky unusual way, so that the blendreversal tools (like RestoreFPS) will fail big time.

[eXtremeDevil]

OK, thanks, I'll leave it just with TFM(), if there's nothing more to be done...

What is the difference with TFM().TDecimate() and TFM(pp=0).TDecimate()?

And what about the McBob? Why my Avisynth doesn't load it?

Thanks.

[wonkey_monkey]

I had a feeling it might be Doctor Who or Torchwood!

bob.srestore (http://avisynth.org/mediawiki/Srestore) # or a better bobber (qtgmc for instance)

does a pretty good job, but the chroma still shows a little residual blending. As I recall, some episodes of Torchwood (Countrycide springs to mind) also had blending problems even in PAL format - they used it as a bit of a guinea pig for doing effects in HD, and some new Sony HD cameras - so that may not help on other episodes.

Planet of the Dead (http://en.wikipedia.org/wiki/Planet_of_the_dead) was released on Blu-ray in the same way (I did contact 2entertain to complain, but they had no plans to re-release), and I wrote an article on how I fixed it (which requires the PAL DVD):

Restoring the Doctor Who Specials (http://horman.net/doctorwho/specials.php) (scroll right to the bottom).

David

[Didée]

*deleted*

edit:

Hoho, I fooled myself by just looking at TFM, and worse, not disabling the pp. http://www.cheesebuerger.de/images/smilie/konfus/a080.gif (http://www.cheesebuerger.de/smilie.php)

Forget what I have or had written. It's indeed a blendbox-normconversion, with some chroma oddities. Srestore is mandated, but the chroma issues.

[eXtremeDevil]

Sorry, but I'm no expert on avisynth. That restore function also deletes the deinterlace, right?

So is that all I can do? I don't have the DVDs and I don't quite understand your guide...

[wonkey_monkey]

Sorry, but I'm no expert on avisynth. That restore function also deletes the deinterlace, right?

So is that all I can do? I don't have the DVDs and I don't quite understand your guide...

You need to "bob" the video first, which is what "deletes" the deinterlace, then srestore selects the least blended frames from the result, and aims to get back to the original 25fps.

That is pretty much all you can do, without studying the video and trying to determine the chroma blending pattern, and then writing your own filter to fix it... just using srestore will give you a much more watchable result though - no more juddering pans.

David

[eXtremeDevil]

So I guess I'll leave it like that. Will I see that black and white effect on action scenes or it will be almost invisible? 'Cos I remember watching "Children of earth" and every time someone was running I could see some black and white parts...

[wonkey_monkey]

Yeah, you probably will see something like that. They really did make a mess of it...

[eXtremeDevil]

I've just tryed bob() and repal(). It seems to "delete" the b&w effect. Is that OK?

[wonkey_monkey]

I can't see a lot of difference between srestore and repal on your sample clip, except that repal makes a better choice on one frame with particularly blended chroma and is slower. Maybe this b&w effect is something else and repal just happens to handle it better.

If it looks okay to you, it's okay.

David

[2Bdecided]

Guys, it's really easy...
1) move to the UK
2) capture the original free-to-air 1080i50 HD broadcasts
:)

(No, before anyone asks, I didn't do this)

Cheers,
David.

[eXtremeDevil]

Guys, it's really easy...
1) move to the UK
2) capture the original free-to-air 1080i50 HD broadcasts
:)

(No, before anyone asks, I didn't do this)

Cheers,
David.

Two Davids? :P

Then I will have the BBC HD logo haha

Look at these captures:

http://www.mediafire.com/?dmtgl1pgfjl35wg

In the source ones you can see the b&w effect with bob.srestore, although in the encodes the frames are equal in that scene.

But look at a different frame on the encodes, there's the effect, but depending of the filters the color is b&w in different places.

Which filter do you guys recommend, seeing the differences? I mean, for a proper visualization.

EDIT: Forgot to upload the screencaps xDD

[2Bdecided]

Then I will have the BBC HD logo hahaThat would be a lot easier to remove ;) It's not on the current BBC One HD broadcasts btw - they dropped it during the Olympics and thankfully have forgotten to put it back (for now).

Cheers,
David.

[eXtremeDevil]

And apart from that, what do you guys think of the different encodes?

[Didée]

I think that there are roughly 30'000 blended fields/frames in a 45min episode, and that you can conclude any-or-nothing from looking at two of them.

[eXtremeDevil]

But which avs will you use and one encode do you think will look better in movement to the human eye?

[eXtremeDevil]

Can any expert recommend me one of the two method or is it completely the same to the human eye at the end?

[lansing]

what more do you want, there's no point to choose between one crappy source over another

[eXtremeDevil]

But maybe to the human eye at the end, the final mkv may look a tiny bit better. Or not?

[Didée]

If a friend says that Pepsi tastes better than Coke, it is surely true for his taste, but not necessarily for others'.

You try both ways, compare, and decide.

[eXtremeDevil]

OK, thanks.

[Didée]

Both tools try to achieve the same result. And essentially, both tools ARE achieving the same result. But, due to different approaches, the one tool may fail on the occasional frame X (where the other tool happens to succeed), but will succeed on the occasional frame Y (where the other tool happens to fail).

[Jenyok]

This is the bestest deinterlace code.
It is very very slow.
.

LoadPlugin("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\EEDI3\eedi3.dll")
LoadPlugin("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\NNEDI3\nnedi3.dll")
LoadPlugin("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\TDEINT_20070122\tdeint.dll")
LoadPlugin("C:\PROGRAM FILES\AVISYNTH 2.5\PLUGINS\TMMV1\tmm.dll")

AssumeTFF() # Верхнее поле первым
#AssumeBFF() # Нижнее поле первым
edeintted = eedi3(field = -2, \
sclip = nnedi3(field = -2, \
qual = 2))
TDeint(edeint = edeintted, \
order = -1, \
mode = 2, \
sharp = true, \
mtnmode = 3, \
full = false, \
tryWeave = false, \
type = 2, \
emask = TMM(mode = 1))