MeGUI didn't detect correct interlacing mode?

[StephenChow]

Hello,

When I used Deinterlaced Material Detector in MeGUI Avisynth script creator, it selected TIVTC+TDeint(EDI) -- slow as default.
http://img475.imageshack.us/img475/3604/meguiasc1ep.png
It generated file .avs as

DGDecode_mpeg2source("C:\NEW\VIDEO_TS\VTS_01_1.d2v",info=3)
ColorMatrix(hints=true)
edeintted = AssumeTFF().SeparateFields().SelectEven().EEDI2(field=-1)
tdeintted = TDeint(edeint=edeintted,order=1)
tfm(clip2=tdeintted).tdecimate(hybrid=3)

and after it finished encoding, I have the .mp4. I used x264.
But when I saw it on my Media Player Classic, it looked like:
http://img478.imageshack.us/img478/6666/tivtctdeintedislow29jy.png
and
http://img487.imageshack.us/img487/8382/tivtctdeintedislow5cl.png

I noticed that every 5 frames, there are 3 progressive frames, and 2 interlaced frames (after and before interlacing - I mean DVD source and the final x264 mp4 file).
I don't know how to fix this. I read the guide on Doom9.org, they said that I need IVTC'd, but MeGui detected my source is hybrid?
MeGUI didn't detect source correctly, did it? How can I fix it?

Help me
Thank you

[Carpo]

only thing i have seen simular to this (the strange pixaling on the edges of things) was due to having colour correction ticked, dont know about the interlacing tho

[tritical]

DGDecode_mpeg2source("C:\NEW\VIDEO_TS\VTS_01_1.d2v",info=3)
ColorMatrix(hints=true)
edeintted = AssumeTFF().SeparateFields().SelectEven().EEDI2(field=-1)
tdeintted = TDeint(edeint=edeintted,order=1)
tfm(clip2=tdeintted).tdecimate(hybrid=3)

If using colormatrix on interlaced yv12 (i.e. before ivtc or deinterlace), then "interlaced=true" needs to be specified. So the script would be:

DGDecode_mpeg2source("C:\NEW\VIDEO_TS\VTS_01_1.d2v",info=3)
ColorMatrix(hints=true,interlaced=true)
edeintted = AssumeTFF().SeparateFields().SelectEven().EEDI2(field=-1)
tdeintted = TDeint(edeint=edeintted,order=1)
tfm(clip2=tdeintted).tdecimate(hybrid=3)

it would probably be better to use the following (assuming the colorimetry does not change at some point in the clip):

DGDecode_mpeg2source("C:\NEW\VIDEO_TS\VTS_01_1.d2v")
edeintted = AssumeTFF().SeparateFields().SelectEven().EEDI2(field=-1)
tdeintted = TDeint(edeint=edeintted,order=1)
tfm(clip2=tdeintted).tdecimate(hybrid=3)
ColorMatrix(d2v="C:\NEW\VIDEO_TS\VTS_01_1.d2v")

Anyways, I would try one of those and see if it fixes the problem.

[StephenChow]

Hi tritical,

Your second script works great except every 6 frames, the 5th and 6th frame is almost same (5th frame has a shadow, 6th frame is clear).
So I used Decimate do remove 5th frame and seem it's working good now.
MeGUI selected TIVTC+TDeint(EDI) as default, but I can select TVITC and Decomb IVTC, if I choose TVITC or Decomb IVTC, the result is ok, but I don't know if TIVTC+TDeint(EDI) is better than the others?

Source DVD ( Decrypted by DVD Decrypt, Trimmed by M2-Edit Pro)
http://rapidshare.de/files/23950987/video.zip.html
I posted this video so you can test with MeGUI.

[berrinam]

I think I will change the default to from TIVTC+TDeint(EDI) to TIVTC, because it's not worth the slow-down. I'm not aware that it caused any problems as you described, though.

[Revgen]

@berrinam

For the record, the video Steven Chow is trying to encode is a Hong Kong DVD. I doubt it's telecined correctly.

[StephenChow]

In this video TIVTC+TDeint(EDI) isn't enough, the final script I must add TDecimate(cycleR=1) so the full script is:
DGDecode_mpeg2source("C:\NEW\VIDEO_TS\VTS_01_1.d2v")
edeintted = AssumeTFF().SeparateFields().SelectEven().EEDI2(field=-1)
tdeintted = TDeint(edeint=edeintted,order=1)
tfm(clip2=tdeintted).tdecimate(hybrid=3)
TDecimate(cycleR=1)
ColorMatrix(d2v="C:\NEW\VIDEO_TS\VTS_01_1.d2v")

TDecimate(cycleR=1) is required to remove duped frames.

But this video works perfectly if I select TIVTC
TDecimate(cycleR=1)
OR
Decomb TIVTC
follow the doom9 guide ( Inverse Telecined )
AssumeTFF()
Telecide(post=0,guide=1)
Decimate(cycle=5)

The DVD Source is Interlaced NTSC, not FILM NTSC.

[Sharktooth]

Uhm colormatrix should be moved as the last entry in the script too.

[Butterfingers]

Sounds like its a 24p Film > 60i NTSC > 50i PAL transfer. This would explain 3 out of 5 frames having combing and the other 2 frames being perfect without any interlacing.

[Deep_Eyes]

I have noticed that when it detects progressive content it says "#Not doing anything because the source is progressive" I'm not positive on this, but wouldn't deinterlacing be need? .. I'm doing a animed source, and its a movie. I'm not totaly sure if this is a good idea to do "nothing" on progressive content.

[communist]

Deinterlacing turns interlaced material into progressive. If you start with progressive material you dont need (and shouldn) deinterlace it.

[Deep_Eyes]

wow, learn something new everyday.

@commuinst

Thanks for the info :)

[cbarroso]

Uhm colormatrix should be moved as the last entry in the script too.

It Sounds like a big bug in the AVS creator…

[Wilbert]

Uhm colormatrix should be moved as the last entry in the script too.
That's not necessary if you specify the d2v file instead of using hints (provided you don't convert to RGB before that).