TDeint and TIVTC

[Mug Funky]

@ chipzoller:

if something was shot on film for TV in PAL land, the cameras will have been run at 25fps as there is no reason to run them at 24. so i'd assumefps, and be assured the audio pitch will be correct as it was shot at that speed to begin with (and nobody would bother with a timestretch for a 1/1000 difference in speed).

it's possible this episode was either never aired in NTSC land, or a tape could not be found and thus a conversion was done off a PAL master.

[laserfan]

Originally Posted by laserfan

tfm().tdecimate(hybrid=1)
...is [there] a new-or-better script I can or should use for these captures.

Hmmm, I wonder why no one has an opinion about this? I didn't think my question was too off-the-wall?

[ChiDragon]

laserfan,

EDIT: Nevermind, I tried my suggestion and it doesn't work... your script is the best I can think of, of the solutions that don't require manual intervention.

[Atak_Snajpera]

TDeint() generates strange artefacts.
Source is 1440x1080i (Canon HV20)
http://www.archive.org/download/jvc_hd7/hv20_05.m2t.mpg

After TDeint() look what happens



I've also uploaded sample after conversion to 1280x720 (x264)

http://www.megaupload.com/?d=71Z12QI0


Another example:
Source: http://www.archive.org/download/jvc_hd7/hv20_07.m2t.mpg

Once again strange spots show up


I used very simple script

LoadPlugin("C:\Users\Dawidos\Documents\Delphi_Projects\RipBot264\Tools\dgindex\DGDecode.dll")
MPEG2Source("c:\TEMP\job1\job1.d2v")

tdeint()
LanczosResize(1280,720)

[laserfan]

Originally Posted by ChiDragon

...your script is the best I can think of, of the solutions that don't require manual intervention.

Thanks for that--I just had to ask, given the many, many pages of enhancements discussed here, most of which I just don't understand.

But thanks for responding!

[tritical]

Almost ready to start working on tivtc/tdeint again. I read back through the last few pages and this is the list of current issues:

tdecimate mode 2 problem non-standard framerates - Xesdeeni/ChiDragon
tdecimate problem with mode 0 decimation - ChiDragon
add tfm hints to tdeint - ChiDragon
tdecimate '+' override - ChiDragon
tdecimate problem with video section decimation in mode 5 (starting in version 0.9.12.7) - Asrial

If there are any other items I've forgotten please let me know .

[JuanC]

Well, now that you mention other forgotten items, there's this old request, for an exclusion band for Tdecimate. Not really a "current issue" but it will certainly help people like me dealing with 30fps interlaced subtitles blended on top of 24fps hard-telecined material. Thanks

[tritical]

An exclusion band in tdecimate could be accomplished by using the clip2 parameter:

mpeg2source()
tfm()
saved = last
stackvertical(crop(),crop()) # set crops to remove lines containing subtitles
tdecimate(clip2 = saved)

In this case, tdecimate will calculate everything based on the input clip, but will output frames from clip2. The clip2 clip only has to have the same number of frames as the input. All other properties: height, width, and colorspace can be different. You could replace cropping/stackvertical with another method of removing the subtitles (set the subtitle area to black using overlay, etc...).

It would be possible to add y0/y1 parameters to tdecimate and have it do the above internally, but I don't think it's worth it. Also, actually having y0/y1 parameters that are taken into account during metric calculation would be too much of a pain to add atm. However, it is on the list of things to add in the tdeint/tivtc rewrite.

[DarkNite]

However, it is on the list of things to add in the tdeint/tivtc rewrite.

Did I miss something, or have you been working on a whole new code base in your secret other life while cackling like a mad scientist who just built two tesla towers with just enough space for an operating table between them? ^_^

tdecimate problem with video section decimation in mode 5 (starting in version 0.9.12.7) - Asrial

Thank you for adding that to your current to do list.

[max24]

How is one this SOURCE Deinterlacen?
I despair of it

Sampler:
http://rapidshare.com/files/31967235/Sampler.m2v.html
http://simpleupload.net/download/88293/Sampler.m2v.html

Thanks for your Help

[JarrettH]

http://forum.doom9.org/showthread.php?t=125990

Maybe I should ask this here instead.....

I'm trying to encode Pan's Labyrinth and MeGUI detects it as Partially Film (correct) with Top Field First. What's the difference between choosing TIVTC alone vs TIVTC+TDEINT?

[ChiDragon]

TFM by default will use PP=6 which is motion-adaptive cubic interpolation on frames detected as combed. TFM+TDeint by default will use TDeint's different motion masking and kernel interpolation instead of cubic.

Maybe you should post of a sample of the unprocessed VOB in a section where you say TIVTC creates jerkiness.

[JarrettH]

I think I figured it out. So apparently clicking analyze will make a different avs script than if I were to choose Partially Film + Top Field + TIVTC manually.

If I select myself:

tfm(order=1).tdecimate(hybrid=3)

If I click analyze

tfm(order=1).tdecimate(hybrid=1)

[ChiDragon]

hybrid=1 is for mostly film (blends 30p->24p), hybrid=3 is for mostly video (blends 24p->30p).

[JarrettH]

It definitely did fix it though. The stream is 23.97 fps and looks proper. So TIVTC+TDEINT should produce better quality?

Thanks Chi

[ChiDragon]

@max24, except for 2 shots of the plane in the opening (before and after the kangaroo at the end), that is just plain old easy field shifting. Use "TFM()" or even "TFM(mode=0,PP=0)". The two plane shots mentioned have some duplicated fields which will cause the video to be a bit jerky there, but they're short sequences anyway.

[ChiDragon]

JarrettH, I believe tritical would agree that TDeint's deinterlacing is superior to what's built into TFM yes.

[JarrettH]

It probably is, but I was bored and made some captures

I think this movie only begins to have interlacing problems near the end. Three shots, one TIVTC alone the other TIVTC+TDEINT

If anyone wants to compare:

http://rapidshare.com/files/32249715...DEINT.rar.html

Thanks!

[JarrettH]

Top one in each set is TIVTC alone, bottom is TIVTC+TDEINT

[foxyshadis]

I don't think the TFM clip2 parameter is working as it should. I've been using:

clipt=TDeint(field=1,emask=TMM(field=1),edeint=separatefields.selecteven.eedi2(field=1))
clipb=TDeint(field=0,emask=TMM(field=0),edeint=separatefields.selectodd.eedi2(field=0))
TFM(clip2=clipt,pp=4)

But the frames I get back are exactly the same as with TFM(), and very different from clipt. I haven't had any luck figuring out why.