Feeling Lost with Interlaced Video

[manono]

And QTGMC together with TDecimate for the wrong framerate could also make something play jerky. When making statements such as yours, you might want to make both the script you used and a short sample available so people could have a look.

Also, unless you have a reason for using QTGMC, a real IVTC would be done with TFM/TDecimate and not QTGMC/TDecimate and especially not with QTGMC/SelectEven, as vaporeon800 already mentioned. Was Golden Girls shot on film as you claimed, and not video? I don't know, just asking.

Edit: Apparently it was shot on video so no decimation should be done, just deinterlacing if it has to be deinterlaced. And for that QTGMC/SelectEven is a good way to go. These are things you should figure out before just throwing random filters at your source.

[Downgraded286]

Well, I know enough to tell telecined video apart from interlaced video at this point. If you were to watch The Golden Girls without any filtering done you would see combing from hell in nearly every frame. QTGMC cleaned this up wonderfully.

I recently did the same for Monk (ripped the dvd to put on my media server) and most of them were progressive, except for season 1. I saw that distinct 2-3 pattern, ran tfm and tdecimate and it looked fine. However, after this I did notice some shots that looked, for lack of better words, like someone set it to render at 320x240 or something. What I mean by this is excessive jagginess in some scenes, such as one showing some wires strong across telephone poles on a street with a steep hill.


I have no idea what the actual definition of film of video means, at least in this sense. When *I* said film, I meant like... not animation, like a live action tv show shot with a camera. I could be wrong there, but that's how I currently understand it.

Anyways, from now on I won't be using TDecimate and QTGMC together. I still may continue using selecteven, so far I haven't had a problem...

[manono]

If you were to watch The Golden Girls without any filtering done you would see combing from hell in nearly every frame.

Then why did you begin that earlier post with, "I ran into some issues with something else. QTGMC and TDecimate did in fact make something look quite jerky." If, as you say, you know how to tell telecined content from interlaced video, then why would you even try to decimate an interlaced television show?

I have no idea what the actual definition of film of video means...

Film or video? Film is film - shot using 24fps progressive cameras. At least that's what it meant during the era of The Golden Girls. Video means shot using 30fps interlaced video cameras. Something shot using film cameras will be either hard telecined (and encoded as interlaced) or soft telecined (and encoded as progressive) for NTSC DVD. Video is pure interlace (none of that 3/2 stuff) and is encoded as interlaced.

What I mean by this is excessive jagginess in some scenes, such as one showing some wires strong across telephone poles on a street with a steep hill.

That's often an indication of the TFM deinterlacer kicking in for one reason or another. The post-processor is TDeint and it creates that kind of aliasing when used. When using TIVTC, you might use QTGMC as the deinterlacer to prevent that kind of thing from happening:

tdeintted = QTGMC().SelectEven()
tfm(clip2=tdeintted).tdecimate()

There might be other reasons for that jagginess also.

[Downgraded286]

I am trying to use Ripbot264 now. I've gotten as far as loading my AviSynth script into it, and it seems to be doing exactly what VirtualDub already does, except with the x264 encoder. What I mean by this is when I load my AviSynth script, Ripbot264 provides me no options for processing audio, subtitles, or chapter markers. How can I preserve this with Ripbot?

I wanted to try this route because I've come to the realization that not only is there the obvious bottleneck of processing QTGMC, and then encoding the result with x264 at a later time, but also, especially now with the external HDD I am using as a "work" drive, there is the bottleneck of having to process these relatively massive intermediate files on this disk that only reads and writes so fast. I thought I could skip all this by loading my AviSynth script into Ripbot instead, which I thought would be able to gather the audio, subtitles and chapter markers as well as my AviSynth script, and process it all at once from the comparatively small source files I've made with MakeMKV into my final result already in one step. These huge intermediate files, as well as processing the video twice, is really slowing things down.

Now, I realize I could simply use MakeMKV again, only this time merge my already finalized x264 video stream with the original audio and whatnot, but I actually process the audio too, to compress it further, in my final step. That's why I want to load everything into Ripbot.

[manono]

... Ripbot264 provides me no options for processing audio, subtitles, or chapter markers. How can I preserve this with Ripbot?

It has a number of AAC bitrates from which to choose for the audio. And a subtitle section as well. Don't know about chapters as I have no interest in them.

If making MKVs you can also have soft (selectable) subs and a greater choice of audio output. I didn't see any options for creating chapters.

[Downgraded286]

It's looking like I have to demux the source from MakeMKV and plug them in, as the avs script loads video only...

[manono]

What's your script that doesn't handle the audio? I do my audio separately and when done load it separately.

However, I have had instances where it just wouldn't do the audio, even when included in the script with an Audiodub line. You might ask at Doom9 where most questions about the program are handled. And since your source is DVD (right?) then I don't know why you make an MKV rather than doing it the 'normal' way with a D2V file made using DGIndex and an MPEG2Source line to open the video.

[Downgraded286]

Yes, the source is DVD. I've been using MakeMKV to rip them because it seemed to be a simple, quick and efficient program.

Back on Ripbot, it's doing all sorts of odd things. I've seen it stretch the video to be 16:9 when it's supposed to be 3:2, I've seen it set incorrect play time lengths (seems to be roughly double the length it should be), it won't accept the .sub/.idx files I demuxed (unless I burn them into the video as per your example, but I don't want that, I want to add a subtitle track I can switch on/off), and if I convert the sub files I ripped with MKVCleaver to .srt with Subtitle Edit, they'll be there in the resulting video file, but they... don't work. I can select them, but I never see any subtitles.

This is the script (draft preset just for quick temporary results as I try to figure this thing out)

setmtmode(5,4)
ffmpegsource2("A:\AviSynth Worker\The Golden Girls s06e01.mkv")
setmtmode(2)
qtgmc(preset="draft", edithreads=2)
selecteven()

That said, let's try another route. The main problem is the huge intermediate files. My disk is slow enough to make MKVToolNix take hours to process a TV series season, combining the AviSynth output with the original audio, subtitles and chapter markers to load into a file which then gets loaded into Handbrake for final compression. The amount of data to be moved during this intermediate stage is in the hundreds of gigabytes, which is quite slow on a slow external disc (which isn't even a regular external drive, it's an original 2.5" PS4 HDD plugged into a Thermaltake BLACX). It doesn't help that this disc is feeding data to another computer which is running Handbrake at the same time I'm doing this (this is my intermediate work drive, and now one computer handles AviSynth processing, while the other handles Handbrake).

If I could get the file size down, there wouldn't have to be so much data moved, so this process would be faster. I could also queue up more work to be done at once without worrying about running out of space. I've tried both Lagarith and UT Video Codec, and they both produce pretty big files. Is there another lossless codec that might work better? Is there a lossless x264 or possibly even x265 codec I could run in virtualdub to cut the size down?

[manono]

I didn't understand half of that. Like, why is Handbrake even in the equation, if you're using AviSynth.

Back on Ripbot, it's doing all sorts of odd things. I've seen it stretch the video to be 16:9 when it's supposed to be 3:2,

Then check your SAR settings and change to 1:1 if need be. If you can't get the subs into RipBot264, mux them in afterwards using MKVMerge GUI, during whch time you should be able to create the chapters as well.

I use Lagarith for lossless AVI. The sizes produced are among the smallest among the lossless AVI codecs. If you complain about the file sizes, maybe it's time for a new hard drive. Otherwise, just take the AviSynth script and encode drectly from it. Always test your scripts in VDub before sending them to the encoder. And I haven't much interest in helping anyone encoding from an MKV whose source is DVD when it's much more reliable to use MPEG2Source. Good luck.

[Downgraded286]

What's wrong with ripping a DVD to MKV? I've been using MakeMKV to rip DVDs since I've started doing this, it seems to be a simple program that produces good results. What makes mpeg2source better, and what other tools would I need in order to be able to use this function (what would I replace MakeMKV with)?

I'm not too worried about the filesize of the intermediate files because I'm low on HDD space (although smaller files would allow me to do more at once). I'm worried about it because VirtualDub/AviSynth only produces video, so I have to remux them with MkvToolNix to put the audio, subtitles etc back, which copies the very large video... and dealing with that much data is slow. I plan on changing things up a bit soon and getting a faster disc, but still, even then it would be slow.

[manono]

Better never than late. If working with DVDs repackaged as MKVs works for you, then keep at it.

[jagabo]

This is the script...

setmtmode(5,4)
ffmpegsource2("A:\AviSynth Worker\The Golden Girls s06e01.mkv")
setmtmode(2)
qtgmc(preset="draft", edithreads=2)
selecteven()

Add atrack=1 (or whatever the audio track number is) if you want audio too:

ffmpegsource2("A:\AviSynth Worker\The Golden Girls s06e01.mkv", atrack=1)