MakeMKV

[Selur]

can't download the video (no clue why, tried different browsers)
I guess the source is probably either telecine or telecine (soft), overwriting the scan type and telling Hybrid to use the original frame rate probably will help with Hybrid handling the source. (telling Hybrid to ignore the time codes might help too)

[davexnet]

This seems to work, field match and decimate instead of qtgmc:

aud=ffaudiosource("C:\Users\davex\Downloads\A1_t00.mkv",track=-1)
vid=ffvideosource("C:\Users\davex\Downloads\A1_t00.mkv", track=-1,  seekmode=1,fpsnum=30000,fpsden=1001)
audiodub(vid,aud)
tfm().tdecimate()

[Chris Hooper]

Nobody was able to figure it out de-interlacing with QTGMC.
My reply earlier wasn't to start back up with things that no one could figure out. It was only in reference to MakeMKV not giving exact copies like people have stated.
I've spent my hours and weeks researching, listening to others ideas, etc., which nothing worked.
I no longer have an interest in something that I know isn't going to work.
Unless there's a way to get MakeMKV to correct the issue so I can de-interlace using QTGMC, then there's no point to continue.
I don't wish to waste my time nor everyone else's time beating a dead horse. LOL!

[davexnet]

The file is not interlaced, it's telecined with 3:2 pulldown.
qtgmc is not appropriate for this

[Chris Hooper]

If that's the case, Dave, then why does it look interlaced when I turn off my auto de-interlacer in VLC?
Plus, why would QTGMC work perfectly on the first 4 episodes on disc 1, but then start messing up on the discs thereafter?
The MediaInfo on the 1st disc stated it was Interlaced. I'm almost positive. And I thought it said the same thing on the other discs as well.
I can't check at the moment because I'm currently de-interlacing a video, but I will when it's finished.

[Chris Hooper]

Waaait a minute. I remember now. Awhile back. I tried demuxing and remuxing, extracting the audio, converted the video to m2v, used DGPullDown, then recombined the ac3 audio and converted m2v back using MuxMan. I'd forgotten about that.
It did absolutely no good. That was so long ago, I'd forgotten about that. I think that was one of the first things I tried to do.

[netmask56]

I have transferred over 800 DVD's and 300 BD and UHF BD's using MakeMKV to my NAS. Absolutely no quality change, no problems with sync, nada! My playback device is a Zidoo UHD3000 media player. I have the paid version as I like to support the author and I don't have to bother updating the key every 3 months of so. I watch on an 85" SONY 4K TV with a separate sound system. I don't know what the OP is doing to get the results reported. In my experience over the years MakeMKV is rock solid and does what it says.

[Chris Hooper]

The file is not interlaced, it's telecined with 3:2 pulldown.
qtgmc is not appropriate for this

Here's an image I just took using DGIndex, Dave.
I thought I remembered it saying Interlaced, and it shows it here in DGIndex.


[Attachment 84540 - Click to enlarge]

[ForSerious]

It's only incorrect if you can prove that you don't get the issue when you rip with another program.
That mkv file you linked, plays back correctly. I can't find anything that reports it being interlaced, but I can see the scan lines if I transcode it at all.
When I run QTGMC through vapoursynth, it produces a file that is deinterlaced and plays back correctly.
Here's the vpy:

import vapoursynth as vs
core = vs.core
import havsfunc
clip = core.lsmas.LWLibavSource(source="E:/Dump/Online/A1_t00.mkv", format="YUV420P8", stream_index=0, cache=0, prefer_hw=0)
clip = core.std.SetFrameProps(clip, _Matrix=5)
clip = clip if not core.text.FrameProps(clip,"_Transfer") else core.std.SetFrameProps(clip, _Transfer=5)
clip = clip if not core.text.FrameProps(clip,"_Primaries") else core.std.SetFrameProps(clip, _Primaries=5)
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
clip = core.std.SetFrameProp(clip=clip, prop="_FieldBased", intval=1)
clip = havsfunc.QTGMC(Input=clip, Preset="Placebo", TFF=False)
clip = clip[::2]
clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)
clip.set_output()

I converted the result of that script to H.265 with ffmpeg and libx265, crf 12.

I don't see any evidence that this is MakeMKVs fault.
If you can DM me a place to upload the result, I can send you the file I got. Then you can judge.

[davexnet]

fpsnum=30000, fpsden=1001
This part fixes any inconsistencies in the frame rate, without it, it does go out of sync.
I also used it (post #32)

[Chris Hooper]

Thank You, Dave.
I definitely believe you. You've helped me before on a project I was working on last year, and it worked out beautifully.

[Alwyn]

I concur with Dave. It is not interlaced, it's 3:2 telecined. Open it in VDub and step through frame by frame. The 3:2 pattern (PPPII) is obvious.

Further, VDub is reporting it as 29.97.

[Chris Hooper]

Thank You, Alwyn.
So no De-Interlacing is required, like ForSerious is saying?
I'm just trying to clarify as to what I need to do.

[ForSerious]

I could be wrong.

The first thing I did was try to drop frames with -r in FFMPEG. That's when I saw the scan lines while playing back with MPC-HC.
There's something abnormal about this particular video though. The first scene, has no repeating frames, but the next scene does.

I'm only trying to suggest that if you see scan lines in your final resulting file in that seconds scene, then deinterlace it no matter what all the tools say.

[Chris Hooper]

I think the scan lines that you're speaking of, Serious, can be seen even when it isn't interlaced but is telecined. That's what makes it difficult. Hence the reason why Alwyn and Dave checked for the 3:2 Pulldown pattern.

[Alwyn]

As @ForSerious says, the first scene is different; the frame pattern as I see it is roughly 1 2 3 4 5 5 6 7 8 9 9 10 11 12 13 14 14 15 16 16 17 18 18 19 20 20 21 22 23 24 24 and so on. There's a mix of repeated frames.

The rest of the movie appears to be straightforward Telecine 3:2. I tried Dave's AVISynth script in post #32 and it clears up most of the duplicates in the first scene, as well as removing the telecine combing effects for the rest of the movie, and it's in-sync at the end.

So yes, as Dave says, it doesn't need de-interlacing, it needs Inverse Telecine.

[Alwyn]

can be seen even when it isn't interlaced but is telecined

Yes, the telltale sign. An interlaced file would have jaggies on every frame, whereas a telecined file only has them only when the telecine routine needs to insert an extra field to increase the frame rate up to 29.97 for TV. Depends on the pattern used, normally it's 3:2: PPPII.

Easily spotted with VDub because it has no auto-deinterlacing features like VLC has. You can, of course, deinterlace to double frame-rate manually to observe each field.

BTW, it looks to me like that file is 4:3.

[Chris Hooper]

Speaking of VDub2, Alwyn.
Couldn't I just load the video file into that, use the IVTC filter, and save the video using that to inverse the telecine?

[Alwyn]

Couldn't I just load the video file into that, use the IVTC filter, and save the video using that to inverse the telecine?

Definitely, but TFM and TDecimate do a better job. Some VDub IVTCs I've done still have residual artifacts including blending or combing. I think it depends on the source how well VDub does.

In this case, it looks pretty good (teeth):


[Attachment 84550 - Click to enlarge]

[Alwyn]

Here's AVISynth/TFM-Tdec verses VDub TIVTC (Reduce FR, AutoDetect, DAPTIVE). VDub does very well!


[Attachment 84551 - Click to enlarge]

[Chris Hooper]

Thank You, Alwyn!
Yeah, those were the setting I used on VirtualDub.

[Chris Hooper]

Okay, so I noticed discrepancies in the video that didn't look quite right like with the other videos that I'd de-interlaced using QTGMC.
So I checked the video with MediaInfo that I had reversed or inverted the telecine on and sure enough, it was showing that it was interlaced.
So the issue was, the video was both telecined and interlaced. Once I reversed the telecine then de-interlaced it, it came out looking perfect just like the others did, and without any audio problems.
So now I know that some videos can actually be both telecined and interlaced.
Good to know for future reference. LOL
Here's the end result of the video after I reversed the telecine and then de-interlaced it.

https://1drv.ms/v/c/fd0d7e8f253b6e5b/EfUDY_XODrpFhUwuFl7VsKQBed0kwZgg0zQ-OPN0FBL1Zg?e=kQFjaR

[davexnet]

Hello I took a look. The problem is the non-standard framerate; every other frame is a duplicate of the frame before it

[Alwyn]

So I checked the video with MediaInfo that I had reversed or inverted the telecine on and sure enough, it was showing that it was interlaced

The interlacing should have been obvious in movement scenes.

[Chris Hooper]

Hello I took a look. The problem is the non-standard framerate; every other frame is a duplicate of the frame before it

How does everything look now, Dave?

[Chris Hooper]

Hello I took a look. The problem is the non-standard framerate; every other frame is a duplicate of the frame before it

If you're referring to the video I just uploaded, Dave, then I don't think you're looking closely enough at each scene. I'd opened it in VDub, made it full screen, and I can easily see that each frame is different than the other. There are no duplicates. Small minute changes perhaps, but no duplicates.
I'm guessing you're talking about something else.

[Chris Hooper]

Okay, I see the whole reversing or inverting the telecine is going to have to be scrapped.
I was watching the videos after inverting the telecine using the drop frame technique, which takes it to 23.976. Whether before or after de-interlacing I noticed the problem is, when watching it in real time, it's causing the video to jitter.
That sucks. I thought this would've worked. I was so excited about the audio being in sync, I didn't think about the smoothness of the video in real time playback.

[Chris Hooper]

I just found this in the MakeMKV forums, so I'm guessing I'm not alone.
Once again, goes back to my original reply to this post. LOL


https://forum.makemkv.com/forum/viewtopic.php?t=26109

[davexnet]

I just found this in the MakeMKV forums, so I'm guessing I'm not alone.
Once again, goes back to my original reply to this post. LOL


https://forum.makemkv.com/forum/viewtopic.php?t=26109

Yeah it seems odd. However, if you take the mkv and remux it to mpg in Avidemux,
the frame rate is good - dgindex says it's interlaced because it 29.97 fps (23.976 with hard pulldown) .

It still can be inverse telecined using tfm().tdecimate()
Here's a scene close to the end. DO you see any problems with it?

aud=ffaudiosource("c:\users\davex\desktop\A1_t00.mpg",track=-1) # << from your source in post #29
vid=d2vsource("C:\Users\davex\Desktop\avs_test\ch2.d2v")
audiodub(vid,aud)
tfm().tdecimate()
trim(62000,64000)

[Chris Hooper]

I'm currently inverse telecine-ing using your method so that I can check the video throughout.
The video you uploaded is only 90 seconds of very little movement, so it's very difficult to see if there's any jitter.