2 problems: Anime IVTC and DVD-film 97.5% progressive.

[rrats]

ProgSADMask=20 can help with jagged lines (QTGMC)

[jagabo]

Try something like:

Mpeg2Source("VTS_01_1.[0].d2v")
q = QTGMC()
s = ChangeFPS(q.framerate)
ConditionalFilter(s, q, s, "IsCombedTIVTC(cthresh=9)") # if there's combing in s use q, otherwise use s
Santiag(0,1) # or 1,1

[poisondeathray]

poisondeathray: The most obvious part where you see aliasing is when the boy jumps at the beginning. Check the line at his arm pit. It staircases through the whole shot in the vob sample. Here's a capture from my QTGMC encode: Capture

But something else is going on in terms of the aliasing.

This is what I get from the same frame
MPEG2Source("VTS_01_1.[0].d2v")
QTGMC(fpsdivisor=2)

IVTC

[Attachment 60606 - Click to enlarge]

QTGMC(fpsdivisor=2)

[Attachment 60607 - Click to enlarge]

Your screenshot

[Attachment 60608 - Click to enlarge]

And it's not as bad in terms of aliasing as your screenshot. My screenshots were 1:1 output (720x480) from the script, not AR corrected or resized by the player.

Maybe you have other filters or shaders applied in the player during playback ?

But notice the blended horizontal lines in QTGMC compared to the IVTCed version.

In general you don't want to deinterlace progressive content. Adding additional artifacts from incorrect processsing is why you should be filtering selectively

So how do I go about encoding it VFR, as in the same frame rate as the original video?
Thank you for all your suggestions.

It's too much to explain in a post. There are guides if you look around, also in anime forums, and the TIVTC readme. There are some posts on VFR on the avisynth wiki and doom9 as well

Short version:

If you just want just rates 29.97p and below, usually 2 pass TIVTC is what is used .

If you want to keep 59.94p sections, that's more complicated because TIVTC is actually broken. The way I do it is ChangeFPS (60000/1001) on a "base" TIVTC version, then just replace the actually 59.94p sections with the QTGMC version frames with RemapFrames. Then run dedup which decimates according to thresholds outputs timecodes. It requires some adjusting threshold for ideal results so you run it in debug mode to tweak the settings - but certain types of animation usually have similar settings. In general for any timestamp based VFR method, you output VFR timecodes and mux them in with mkvmerge/mkvtoolnix

A negative of VFR is it's really only intended for final delivery/playback. It's less suitable if it's going to be processed later, such as in a video editor. Most editing programs, avisynth included, use a CFR timeline, and you 'd have to re-introduce duplicates and convert back to CFR

[ericf]

I did take a look at an Avisynth VFR page and concluded that it might be too much studying for me to be able to make it work.

I use MPC HC and MPC-BE to play my videos. I use EVR (MPC-BE) or VMR9 (MPC HC) for the rendering. I don't think there are any filters applied.
I have madVR installed, too, but reverted to the current ones because I had some issues with soft subtitles.

I don't know what to do then.
Should I still do a straight TIVTC or should I use AnimeIVTC (Hard Telecined) on it (I've actually encoded two volues with AnimeIVTC already but resized to 640x480)? What would the difference in output be? There doesn't seem to be any difference in the sharpness between the two when I take a capture from my samples.
And should I just let the AR be 720x480 and let the player or the setting in an mkv resize it to 4:3? Staxrip contains info on SAR but Megui doesn't include that in encodes so they play back as 720x480 in software players. Do I need to include such information in an avs if I use VirtualDub2?

Also, how do I analyze the video to see what frame rates are contained in the file? I would prefer to encode it by how many % of one type there is. Say, 75% 23.976. Then, I'd do a straight IVTC.
I can live with some choppyness if the animation is clear and sharp.

[manono]

Also, how do I analyze the video to see what frame rates are contained in the file? I would prefer to encode it by how many % of one type there is. Say, 75% 23.976. Then, I'd do a straight IVTC.

First, all NTSC DVDs output 59.94 fields per second. There is no 23.976. You probably meant to say, "23.976fps with soft pulldown".
Second, make the D2V using 'Honor Pulldown Flags and open it using Notepad. At the bottom you'll find the Film (or Video) percentage. Unless the film percentage is 100%, don't use "Forced Film". Use 'Honor Pulldown Flags' and then the D2V option for TFM. Read the doc or the explanation on the TIVTC page.

1.) NTSC Film, normal source (not anime or cartoon). One pass.
mpeg2source("c:\oursource.d2v")
tfm(d2v="c:\oursource.d2v")
tdecimate()

It'll treat the soft pulled down sections as if they had been 'Forced Film' and IVTC the rest. Faster and more accurate than a full IVTC of the entire thing.

[poisondeathray]

Actually I see the aliasing in your "VTS_01_1.[0]QTGMCFPSD2.mp4" video - so it's not a player config issue

But I cannot reproduce it with the script - something wrong with your encoding procedure, maybe script, because you resized too, or maybe other filters are being injected that you don't know about. I'd look at log files if you're using some GUI

Again, all I did was

MPEG2Source("VTS_01_1.[0].d2v")
QTGMC(fpsdivisor=2)

and the aliasing that is in your encode and screenshot, is not there

That specific video should be kept at least 29.97 , because there are many 29.97 sections

If not going VFR, I'd just TFM(pp=0) for the main video, and if there are a few interlaced sections like that PIP overlays, I'd use RemapFrames to replace them with the QTGMC version

[jagabo]

Here's another way of bobbing to 60p that doesn't require a runtime filter:

function TFMBobQ(clip v, int "pp")
{
    pp = default(pp, 6)

    q = QTGMC(v)
    qe = q.SelectEven()
    qo = q.SelectOdd()
    v.GetParity() ? Interleave(TFM(v, field=1, clip2=qe, pp=pp), TFM(v, field=0, clip2=qo, pp=pp)) : Interleave(TFM(v, field=0, clip2=qo, pp=pp), TFM(v, field=1, clip2=qe, pp=pp))
}

Mpeg2Source("VTS_01_1.[0].d2v") 
TFMBobQ()
Santiag(0,1)