Script Advice for DVB-S Capture

[XadoX]

i need an Script advice for an DVB-S Capture from the german channel KIKA.
I am capturing the whole season from WICKIE, and wanted to archive them into MP4 (H.264). So i want to get the best from the picture.
After some first tries of encoding with Gordian Knot the picture seems not sharp enought.

# DEINTERLACING
TomsMoComp(1,5,1)

# CROPPING
crop(6,10,704,560)

# RESIZING
LanczosResize(704,512)

DENOISING:
Undot()
mergechroma(blur(1.3))

So i read a while and found some interesting filters. RemoveGrain/Dirt, LimitedSharpen and SeeSaw. But there are so much settings witch generate a bufferoverflow in my brain. Also i did't know witch one is better for the cartoon source. Maybe the deinterlace mode is the false one for such cartoons.

Can some of the script god's could give me some advice.
Here's a link to a example file:
EP02 10secs. (http://rapidshare.de/files/10324620/EP02_-_Die_Falle_cut.m2v.html)

[Richard Berg]

I haven't looked at your sample, but some general ideas:
- TomsMoComp softens the image due to the way it works. Try TDeint.
- MergeChroma(Blur(1.3)) is pretty crude. Something like FFT3DFilter(sigma=3.5, plane=1).FFT3DFilter(sigma=3.5, plane=2) should provide just as much denoising without nearly as much loss of detail.
- LimitedSharpen can very effective without adjusting any of the tricky parameters. Just use it instead of resizing: LimitedSharpen(dest_x = 704, dest_y = 512, soft=false). I prefer to do it after denoising, but it's not inherently better or worse (see the SeeSaw thread).

[Chainmax]

If you have to deinterlace, download TDeint, EEDI2 (both from this thread (http://forum.doom9.org/showthread.php?t=82264&page=25&pp=20)) and use them like this:

The configs for using eedi2 with tdeint's mode=1 are:

1.) tff

interp = separatefields().eedi2(field=3)
tdeint(mode=1,order=1,edeint=interp)

2.) bff

interp = separatefields().eedi2(field=2)
tdeint(mode=1,order=0,edeint=interp)

UnDot only removes a little bit of noise, you might want to use FFT3DFilter or VagueDenoiser for slightly stronger denoising. If you want to stick with Undot-like denoising, use RemoveGrain(mode=1) instead: it's exactly the same but unlike UnDot it denoises the borders of the picture.

[Richard Berg]

Noticed you're doing toons. As such, another denoiser to consider is TBilateral.

[XadoX]

oh thanks for the help. That gives me a few hints how to start with.

But there is another problem. At the beginning of an episode (intro) the source is interlaced and the main movie (like the short clip i ulpoad) is progressive could that be? And when it's so, do i have to deinterlace it different?

[Mug Funky]

it's perfectly possible to have parts interlaced and other parts not :) just makes it harder for us.

if you use a smart enough deinterlacer you can run it on progressive content without breaking anything (much). only thing to watch out for is sharp horizontal edges, but you probably wont miss those if you're resizing down too much.

of course you could always encode it interlaced. DVB is 99.99999999% of the time top-field first. once you know that, it's a snip to encode interlaced xvid or mpeg-2 (if you're doing mpeg-2 then you might as well just burn your captures to DVD without any filtering - guaranteed no loss there). just make sure you don't resize vertically. you can encode something as anamorphic easily enough, so it's perfectly acceptable to encode a 480x480 or 352x480 file and have it play back nice.

[XadoX]

So my actual script without deinterlacing is as follows:

# PLUGINS
LoadPlugin("C:\Programme\Video\GordianKnot\DGMPGDec\DGDecode.dll")

# SOURCE
mpeg2source("C:\TEMP\EP02_-_Die_Falle_cut.d2v")

# CROPPING
crop(6,10,704,560)

# DENOISE
TBilateral()

# RESIZING & SHARPING
LimitedSharpen(dest_x = 704, dest_y = 512, soft=false)

Here the Original:
http://img492.imageshack.us/img492/4056/original3qt.th.png (http://img492.imageshack.us/my.php?image=original3qt.png)

And here the Script one:
http://img492.imageshack.us/img492/5577/script0ph.th.png (http://img492.imageshack.us/my.php?image=script0ph.png)

And i must say the scriped image looks realy better than the original one.
So at last i have only to check the deinterlacer later with the whole episode.

[MOmonster]

The sharpen and the denoising should be done fieldwise if you don´t deinterlace the source, else you will get strange artefacts on the interlaced parts.

[XadoX]

The sharpen and the denoising should be done fieldwise if you don´t deinterlace the source, else you will get strange artefacts on the interlaced parts.

I know, thx. But at first i only whanted to sharpen the image.

[XadoX]

i've even tried to encode the whole thing (x264/MeGUI/AE-Standart-Profile Bitrate=1000). The result looks very good for me, but is was slow as hell i only get 5frames with my Pentium4 3.0HT, so i needed about 2mins for 10secs. of video.

Oh my god, with the deinterlace script by Chainmax i will only get about 2frames. This could be a long encoding time for 78 epsiodes :eek:

[MOmonster]

If time is impotant for you, don´t use tbilateral, it´s really slow. Deen and seesaw work also nice and are faster.

[mg262]

Hey... Vicky the Viking!! This is one of my favourite cartoons of all time -- I have some of the Wickie DVDs which I've been meaning to dub into English for a long time. I think this may be the first time on these forums I've seen someone working with the sort of content I'm interested in :).

Your original clip is noticeably shaky to me -- I think I could benefit from stabilisation. I think the only options are DePan and the VirtualDub filter DeShaker. The DVDs have a serious problem with spots and speckles, which isn't particularly apparent in your source fragment... but if it happens at other points in the video you might want to think about trying plug-ins designed to remove spots and scratches (in particular, look through plug-ins by Fizick and Kassandro).

If the intros are generally interlaced and the main episodes are genuinely progressive, then I would treat the main episodes as genuinely progressive... you can create one script for the intro (intro.avs), encode that losslessly (to e.g. intro.avi) and then process the main episode in your script before adding back the intro, like this:

# source
crop(6,10,704,560)
TBilateral()
LimitedSharpen(dest_x = 704, dest_y = 512, soft=false)
avisource("intro.avi") ++ last

Doing this you have to be careful that the break between intro and main episode happens at a point when the audio is silent. (In fact, if you were using relatively recent MKV options I believe you could get away with only having one encoded copy of the intro on your disk and having it play back before each episode.)

BUT: I think it is quite unlikely that the intro is genuinely interlaced. It's more likely to be field shifted or field blended... could you post a short section of it so we can check?

If you are worried about speed, there are substantially faster versions of LimitedSharpen floating around... foxyshadis has very helpfully documented them on AVISynth.org. Also, IMO these videos don't have a serious problem with noise, so I think you could get away with using a fairly light/fast denoiser. (OTOH, I don't know much about encoding to the low bitrates you are using... maybe even a little noise is a problem?.)

By the way, is there a particular reason why you are targeting 704 x 512?

[XadoX]

By the way, is there a particular reason why you are targeting 704 x 512?

The Value is calculated by Gordian Knot. It is the biggest Size reffered to the origianl i can choose without making a zoomed resize. Hope you unterständ what i meen :confused:


If you are worried about speed, there are substantially faster versions of LimitedSharpen floating around... foxyshadis has very helpfully documented them on AVISynth.org. Also, IMO these videos don't have a serious problem with noise, so I think you could get away with using a fairly light/fast denoiser. (OTOH, I don't know much about encoding to the low bitrates you are using... maybe even a little noise is a problem?.)
I also thougt so, because the source isnt that buggy. Maybe i use a faster one. I have to trie different solution on weekend if my wife's @work :p


BUT: I think it is quite unlikely that the intro is genuinely interlaced. It's more likely to be field shifted or field blended... could you post a short section of it so we can check?
Here one piece from the intro witch runs into the progressive part.
Intro 2 Film (http://rapidshare.de/files/10390280/Intro2Film.m2v.html)
Here is the Intro only:
Intro Only (http://rapidshare.de/files/10400717/Intro.m2v.html)


Edit:
After watching the whole episode through virtualdub, i found something interesting. there are some positions in the episode witch look like they where not correct deinterlaced, here a screenshot of it.

http://img421.imageshack.us/img421/7297/interlaced9qp.th.png (http://img421.imageshack.us/my.php?image=interlaced9qp.png)
But there are no other interlaced frames befor or after this frame.

so it gets more complicated i thought :(

[MOmonster]

Your intro isn´t interlaced or fieldblended, it´s simple fieldshifted. So just using tfm(pp=0) will help and will also not damage the progressive parts. The only problem is the chroma channel. I think a wrong encoding is the reason for these artefacts. But it wouldn´t be a big problem to write a small fast function against it. :)

[XadoX]

So just using tfm(pp=0) will help and will also not damage the progressive parts.TIVTC performs good work.



The only problem is the chroma channel.
Ahhh. :scared:
I would be very happy if someone could generate me the script for the wrong chroma channel.

[MOmonster]

Ahhh. :scared:
I would be very happy if someone could generate me the script for the wrong chroma channel.
What about this?
Function xadox2(clip clp)
{
global bobbed = clp.bob().duplicateframe(0)
global tel = clp.tfm(mode=0, pp=0)

diff = YV12Subtract(bobbed.duplicateframe(0), bobbed, tol=0, y=0, U=3, V=3)
qumask = YV12LUTxy(diff, diff.trim(2,0), yexpr="x", uexpr="x y < 0 x y = 128 255 ? ?", vexpr="x y < 0 x y = 128 255 ? ?", y=0, v=3, u=3)
global repaired = MaskedMerge(bobbed, bobbed.trim(1,0), qumask, y=0, U=3, V=3).blur(1)

scriptclip(tel, " LumaDifference(bobbed.selecteven(),bobbed.selectodd()) < LumaDifference(bobbed.selectodd(),bobbed.selecteven().trim(1,0)) ?
\ mergechroma(tel,repaired.selecteven()) : mergechroma(tel,repaired.selectodd())")
return(last)
}


Function xadox(clip clp)
{
a = clp.bob().duplicateframe(0)
global b = clp.tfm(mode=0, pp=0)

global fieldo0 = a.selecteven()
global fielde1 = a.selectodd()
global fieldo1 = fielde1.trim(1,0)
global fielde2 = fieldo1.trim(1,0)

global c3 = 0.0
global c2 = 0.0

d2=scriptclip(b, " l1 < l2 && c0 < c2 ? mergechroma(b,fieldo0) : (l1 < l2 || c1 < c3 ?
\ mergechroma(b,fielde1) : mergechroma(b,fieldo1))")

d1=FrameEvaluate(d2, " global l1 = Lumadifference(fieldo0,fielde1)
global l2 = Lumadifference(fielde1,fieldo1)
global c3 = ChromaUDifference(fieldo1, fielde2)
global c2 = ChromaUDifference(fielde1, fieldo1)")

d0=FrameEvaluate(d1, " global c1 = c3
global c0 = c2")

return(d0)
}
You need masktools and tivtc to run the function. The second should give better results, because of a simple masking, but the first is a little bit faster. Just try it out.

Edit:
Tfm is used inside the function, so just call the source and then the function.
#
source
#maybe deblocking
xadox2()
resizing
deen("a3d",2,3,7)
warpsharp()
#the source is allready sharp enough, just try not to smooth it too much

[Chainmax]

So it's not even interlaced? Good :).
About TBilateral, it's supposed to be very slow and kill too much detail, you might as well just use LRemoveDust_YV12(17,2) instead which is much faster. As an alternative, I'd recommend like Richard Berg to use FFT3DFilter (or FFT3DGPU) like this: FFT3DFilter(sigma=X,bw=32,bh=32,bt=3,ow=16,oh=16), with X varying between 1 and 3. Other good alternatives would be DeGrainMedian(mode=0) or VagueDenoiser (haven't used it in a while so can't recommend settings).

[XadoX]

Edit:
Tfm is used inside the function, so just call the source and then the function.
#
source
#maybe deblocking
xadox2()
resizing
deen("a3d",2,3,7)
warpsharp()
#the source is allready sharp enough, just try not to smooth it too much

Removing the Fieldshift with xadox2() works good, but there are still these broken frames with these interlaced artefacts.

My actual script looks as follows:

# SOURCE
mpeg2source("D:\Wickie\EP02 - Die Falle.d2v")

# DEINTERLACE
xadox2()

# CROPPING
crop(6,10,704,560)

# RESIZE
LanczosResize(704,512)

# DENOISE & SHARPENING
deen("a3d",2,3,7)
warpsharp()

[lamer_de]

I don't know, but the filtered screenshot you posted in your first post already look "artificial" to me. It's kinda hard to describe. The second one looks like a watercolour painting to me, even if all the "detail" (well, there are not really that much details to begin with) is still there. I think it's related to Limitedsharpen, which often gives a weird/oversharpened look when used with non-live-action material at the default parameters (at least that's my experience). In your recent script, warpsharp at the default level wll most likely create the same effect as well. Imho moderate to no sharpening would be enough (maybe something super-simle like unfilter at single-digit strenght?). The problem is that there are no details that can't be brought out through sharpening imho. But, well, those are just my perceptions. Feel free to disagree.

Hey... Vicky the Viking!! This is one of my favourite cartoons of all time -- I have some of the Wickie DVDs which I've been meaning to dub into English for a long time. I think this may be the first time on these forums I've seen someone working with the sort of content I'm interested in Vicky is anime (http://anidb.info/a1825) from the 70's, and the anime audience on this forum is rather huge ;)

CU,
lamer_de

[MOmonster]

Could you upload a small sample with these artefacts. Maybe a simple postprocessing or the tweaking of tfm can help. But without a sample I can´t say so much about.

[XadoX]

Here is one piece with such an error in it.

Look at Frame 14:
http://rapidshare.de/files/10442066/hmm.m2v.html

I think the movie was bad encoded, so that it cant be repaired :(

PS.: Somehow the forum edit function gets different now :confused:
Ahh now i must manuelly switch to advance mode.

[mg262]

lamer_de,

Kind of. It was a Japanese-German coproduction (see http://www.thechestnut.com/vicky/vicky.htm, at the bottom of the page) adapted from a book by Runer Johnson. I'm very far from being an anime expert, but it doesn't have any of the animation characteristics I have noticed in the anime I have seen -- i.e. big eyes, stylised/exaggerated expressions, reuse of animation, minimal foreground movement. A comparable example is the Mysterious Cities of Gold, a French/Japanese/US coproduction which again has none of these features -- and which incidentally is my favourite cartoon of all time!

XadoX,

I would also say avoid warpsharp... it makes things look very artificial. It may also be worth playing with the parameters of LimitedSharpen, particularly trying different values of LMode and also trying SMode = 4. I`ve never actually known what Gordian Knot and Auto Gordian Knot do... but choosing a different output resolution is as simple as changing the two numbers in the resize line of your script. (It doesn`t make much difference in this case, since you are using LimitedSharpen, but in general I don`t think there is anything to be gained by resizing on the last line of your script, unless you need to do so for compatibility with some standalone hardware box.)

Edit: your frame 14 looks like the field shift again. When I play it using DGIndex, it all lists as being progressive except for *one very short section, possibly one frame* which is interlaced. The earlier clip also flipped back and forth between being encoded as interlaced and being encoded as progressive. I don't know what this might mean...

MoMonster,

My eyes aren't good enough for this. If it doesn't take too long to write, could I ask what is actually wrong with the chroma?

[MOmonster]

@xadox
Yes it seems to be a encoding problem, but tfm(mode=2) catch also this problem without deinterlacing. The function on the last page is updated.
Another vote against warpsharp (I meant awarpsharp). But if speed is important and ee is allready there it is an alternative. better ways you can find here: http://forum.gleitz.info/showthread.php?t=22490

@mg262
I think it´s because of a wrong encoding. You can find it only in the fieldshifted parts. There are artefacts from the both frames around. Sometimes a wrong temporal filtering can result similar artefacts, but these are different, similar to chromablending. But luma is completly ok.

[XadoX]

@xadox
Yes it seems to be a encoding problem, but tfm(mode=2) catch also this problem without deinterlacing. The function on the last page is updated.
http://forum.gleitz.info/showthread.php?t=22490

With your updated function i get a strange effekt of gosting (egal ob xadox2 oder xadox), see here:
http://img206.imageshack.us/img206/7074/effect9ei.th.png (http://img206.imageshack.us/my.php?image=effect9ei.png)

[MOmonster]

Yes, that is why I didn´t use mode=2 before. Then such things could happen. I can change the function, that it works for you, but if you could use the old xadox2 for your intros and tfm(mode=2, pp=0, Mi=64) for the rest it would be easier for me. The intros should have similar lengths, so the scripting would be easier and faster this way.

[XadoX]

@MOmonster:
in my final encodings i also wanted to strip the intro of. Because i makes no sense for me to keep it the whole 78 episodes. So lets concentrate only on the main movie.

So i will use the NEW xadox2() on the main movie.
Hmm but the most "encoded Errors" are still there.

[MOmonster]

Then just use tfm(mode=2, pp=0, Mi=64) to avoid the artefacts. you should also use a deblocker, deblock() or cpu=4 with mpeg2source. (both content from dgdecode).

[XadoX]

there will be no end :confused:
the artifacts won't disappear.

[scharfis_brain]

have fun:

mpeg2source("D:\FORUM\XadoX\hmm.d2v") #,cpu=6)
TDeint(mode=0,type=3,full=false,tryweave=true)

these comb artifacts look like dropouts from analogue Broadcast Videotapes.

[XadoX]

with Scharfis_brain Script i get this result:

http://img216.imageshack.us/img216/4404/hmm3nw.th.png (http://img216.imageshack.us/my.php?image=hmm3nw.png)

i also looks strange.

[scharfis_brain]

You can't do it better, unless iy you're editing out each crappy frame manually.

Also you won't notice, that strange look while normal speed playback.

[MOmonster]

I don´t get the problem, tfm works for me. No combing or similar artefacts left. Or are there other interlaced parts in your source:scared: .

[XadoX]

we have all the same source "hmm.m2v". Maybe other plugin versions.


A little overview for me about all mentioned functions witch where interesting:

Postprocessors:
mpeg2source CPU=4/6
deblock()

Fieldshift:
xadox2() # not perfect
tfm(mode=2, pp=0, Mi=64, display=true, sco=0) # works supirior
TDeint(mode=0,type=3,full=false,tryweave=true) # not perfect

Denoiser:
FFT3DFilter()
VagueDenoiser
RemoveGrain(mode=1) # Like Undot
TBilateral() # For Toons SLOW
deen("a3d",2,3,7) # Faster than TBilateral
LRemoveDust_YV12(17,2 # Faster than TBilateral
DeGrainMedian(mode=0)

Sharpener:
LimitedSharpen() #SLOW
SeeSaw
warpsharp() # Faster than Limited
awarpsharp() # Faster than warpsharp

[MOmonster]

I tested once more. Yes we speak about the same source. I just call tfm(mode=2, pp=0, Mi=64) after mpeg2source and the problem seems to be fixed. Tivtc should be the last version on my pc. Just update the version. It´s important that you use really similar (the same) parameters.

use awarpsharp, it´s faster and better than warpsharp, but still not most recommed.

[XadoX]

i use the 1.0 Beta7 from 20.12.2005. thats the newest i thought.
Maybe it also depends on DGindex? I use V1.4.6 Beta 2.

I will trie it now with an older one! With V1.4.5 the same effect.

[MOmonster]

This is the simple code:MPEG2Source("D:\dvd.d2v")
tfm(mode=2, pp=0, Mi=64, display=true)
This is my output:
http://img302.imageshack.us/img302/8573/104jy.th.jpg (http://img302.imageshack.us/my.php?image=104jy.jpg)
I have dgdecode 1.44 and the tfm version you can see on the picture. It looks still not perfect, but the u-match avoid the source artefacts. I can´t imagine what is the problem.

Edit: I was wrong, it´s really a older tfm version. Maybe the different sco cause this problem.

Edit2: With the newest dgindex version it is still all fine, but the newest tfm version don´t work for me. The same picture with the newest tfm version:
http://img285.imageshack.us/img285/8218/02hw.th.jpg (http://img285.imageshack.us/my.php?image=02hw.jpg)

Edit3: Indeed the sco parameter cause this behaviour. My mistake. This should work for you:MPEG2Source("D:\dvd.d2v")
tfm(mode=2, pp=0, Mi=64, display=true, sco=0)

[XadoX]

This should work for you:
tfm(mode=2, pp=0, Mi=64, display=true, sco=0)
And how it works. It works supirior. Your are THE man MOmonster. :thanks:

Now i can go over to choose final filters for denoising and sharpening. There are enought combos in this thread. For me the combo of TBilateral and LimitedSharpen produces the best result. I also have to trie out awarpsharp.

Is there still something to say about my resizing?

Maybee i will trie IIP :) How is the german saying:
"Mit Kanonen auf Spatzen schießen"
"to use a steamroller to crack a nut":D


# SOURCE
mpeg2source("D:\Wickie\EP02 - Die Falle.d2v", CPU=4)
Trim(14276,17600)

# DEINTERLACE
tfm(mode=2, pp=0, Mi=64, display=false, sco=0)

# CROPPING
crop(6,10,704,560)

# DENOISE & SHARPING & RESIZING
deen("a3d",2,3,7)
awarpsharp()
LanczosResize(704,512)


With this script i reach about 10frames in Firstpass and 5frames in the second (MeGUI/x264/AE-STANDART/1000Bitrate).
Here a short encoding about 2mins:
http://rapidshare.de/files/10495261/DEEN_AWARPSHARP.mp4.html

[MOmonster]

Then have fune with playing around. In the german forum (I linked) you´ll maybe find some other useful explainations, but it is not optimized for such comic art. Use the OLD xadox2 for the intro, because there it does, what it should do (the post is changed back).:D

Edit: RemoveGrain(mode=1) after resizing can improve the compressibility a bit more without any bad effect.
Your encoding looks good, maybe not so strong settings for awarpsharp and removedirt to stabilize the source a little bit more.

[Mug Funky]

hmm. looking at the short clip with the glitches in it, it looks like it may have been restored by the same folks (or the same equipment?) that did the He-Man DVD masters. i've seen some similar artefacts in that. it's sort of like the artefacts you'd get with the original removedirt.dll, but with blending as well. i suspect it's a deblending/denoising technique to turn poor telecines into good ones. on bits like that you can kind of imagine what the original tapes looked like (most likely field-blended native PAL telecine, but it's hard to be sure without seeing a little more of it).

[XadoX]

DeSpot by Fizick seems a good Spot remover. Maybe i'll give it a trie on weekend.

Another thing where i have fear to ask is. Because the logo isn't transparent there is no good way to remove it, right?

[Didée]

Because the logo isn't transparent there is no good way to remove it, right?
No, there isn't. Results will be good when the hidden background is uniform, but horrible when something is moving in the affected area.
Eventually, a "pressdown" of the logo like I mentioned in the delogo thread on Gleitz' could be acceptable. (Shortly: perform logo removal, but keep 10%~30% of the logo to make it "fake-transparent".)

[XadoX]

No, there isn't.

Would to good to be true :D


Another question again. If i use the trim option, in combo with AudioDub and nics Audio Plugin. How can i recode only the trimmed audio, if i demux the trimmed audio stream in Virtualdub i get only crapy sound files.

Next week i will get my new system. And then i will trie to encode the whole episodes.

:thanks: Thx to ALL for whole the scripting advice. :thanks:

Now i will trie to learn a little bit more about x264 for encoding.

[sillKotscha]

good evening...

yesterday I was reading this thread. Because of your suggestions to XadoX I've adopted your advises but I'm not so sure if I can transfer your proposals [nearly] 1:1 for my DVB-T capture...

maybe you can help me out a little bit.

this picture shows all infos by DGIndex:

http://tinypic.com/k382he.png

Here (http://s32.yousendit.com/d.aspx?id=3N3MFY2HSE6RK29F9LJ5GA0ULC) I have uploaded a small part of the capture [9856 KB] for testing purposes.

my script is as follows:


import("LimitedSharpenBeta.avsi")
LoadPlugin("DGDecode.dll")
LoadPlugin("ColorMatrix.dll")
LoadPlugin("EEDI2.dll")
LoadPlugin("TDeint.dll")
LoadPlugin("RemoveGrain.dll")
LoadPlugin("MaskTools.dll")
mpeg2source("squeeze.d2v", info=3)
ColorMatrix(hints=true, interlaced=true)
interp = separatefields().eedi2(field=3)
tdeint(mode=1,order=1,edeint=interp)
crop(2,8,716,560)
LimitedSharpen(dest_x = 640, dest_y = 352, soft=-1)

what do you think? Does this script makes sense...

cheers

P.S.: my aim is an 1CD Xvid backup

[Chainmax]

Are you sure this is interlaced? You might be having progressive frames, making deinterlacing unnecessary. Read this (http://www.doom9.org/ivtc-tut.htm) at the "FOR PAL AREA PEOPLE:" section to learn how to see what kind of PAL source you have.

[sillKotscha]

I have trusted DGIndex and my eyes... for proper deinterlacing my knowlegde is too limited and I'm glad a place like this forum/ the german doom9 forum exists. I am familiar with the concept of deinterlacing but nonetheless I can't really see the difference between hybrid clips or shifted frames for example.

I try to gratify my eyes by using different deinterlacers, different scripts, just reading here and adopt things but that's it. If things are properly done by myself is another story... that's why I am asking here if the above mentioned script made any sense for the posted clip.

I did get advise by scharfis_brain and he suggested the following cl:

separatefields().selecteven().eedi2().crop().LanczosResize()

concerning Deinterlacing I just trust him as I would trust my mother :D

the second pass is at 68% - I'm looking forward to see the result...

thanks for your thoughts Chainmax

cheers Sill