Using VirtualDub's Smart Deinterlacer filter with GKnot

[bananix]

i'd need to do this in Gordian Knot before resampling using VirtualDub's Smart Deinterlacer filter (http://neuron2.net/smart/smart.html) so perhaps i can do it by adding few lines to .avs fle

or if i select the TomsMoComp deinterlacing in GKnot is it quite similar as Smart Deinterlace and even better and does it before resampling?

if the video is converted fron NTSC interlaced to same resolution (720x480) xvid so w-zoom and h-zoom are 100% (no resizing) do i need to still use deinterlace before this resampling?

[manono]

Hi-

There are ways to import VDub filters into GKnot's AviSynth script. It's a little bit tricky, but there's more information here:

http://www.avisynth.org/Section+4%3A+Importing+filters+from+VirtualDub#s4

I'm not really sure why you'd want to do that, as there are AviSynth filters as good or better than most of the VDub filters. For example, the best deinterlacer available directly in GKnot is the KernelDeinterlace Filter. However, the most recent version of SmartDeinterlace does have a DCDi emulation which I'm curious about. Have you tried it yourself? Anyway, the best deinterlacer of all is tritical's newly released EEDI2, available here:

http://forum.doom9.org/showthread.php?p=744308#post744308

You'd have to edit the GKnot generated .avs to Load the Plugin, and then add the filter into the script. It's very good, the best I've seen, and very slow. I'm not kidding - very slow. Unless you have a very fast computer, you might be better off using KernelDeint. It's fast and good. Before doing the encoding you can try them out by opening the script in VDubMod and checking. Or you can encode a small part of the video using the Trim command

[bananix]

manono thans't for the info...you didn't mention about TomsMoComp so do you think KernelDeinterlace is better? and if i add any filter to AviSynth script are they all used before this resampling occurs (i'm not sure what it is)?

[manono]

Hi-

I didn't mention TomsMoComp directly because I said that I thought that KernelDeinterlace was the best of the deinterlacers included in GKnot. Tom's is also in GKnot. Therefore...

and if i add any filter to AviSynth script are they all used before this resampling occurs (i'm not sure what it is)?

I'm not real sure what you mean either. However, the deinterlacers are near the top of the script for a reason. They used to be lower down, but people kept cropping incorrectly before using them, and got frustrated when they got nasty artifacts. When cropping for the YV12 colorspace (if making an AVI from a DVD, for example), you have to crop the height by Mod4 (top and bottom cropped in multiples of 4). There's more information here:

http://www.avisynth.org/Crop

In general, you add any new filter to the GKnot generated script in the same area as the already existing similar filters. Deinterlacers near the top, and spatial and temporal denoisers near the bottom, with spatial filters before temporal filters.

[Guest34343]

However, the most recent version of SmartDeinterlace does have a DCDi emulation which I'm curious about.

It's rather primitive. I wouldn't get too excited about it.

Anyway, the best deinterlacer of all is tritical's newly released EEDI2

Used alone it's nonadaptive. You need to use it with a special mode of the newest Tdeint(). I agree that it is very fine.

[bananix]

does the EDDI2 lose 50% of image cos the example had separatefields() and selecteven() in thread of it.

and how do i use it with this special mode of the newest Tdeint()? where to get newest tdeint and what lines to add to my .avs in gknot? i think my source has bottom field first...

[bananix]

i've used TomsMoComp which is good but smoothens way too much so perhaps i need to use Kernel Deinterlace or this new one if i get it to work...and if you install mpeg-2 decoder with ffdshow which is not selected by default you can test TomsMoComp and Kernel Deinterlacer with different options when you play the dvd with for example with MPC.

[Guest34343]

does the EDDI2 lose 50% of image cos the example had separatefields() and selecteven() in thread of it.

Yes, but Tdeint() uses the EEDI2() output only in combed areas of the picture.

and how do i use it with this special mode of the newest Tdeint()? where to get newest tdeint and what lines to add to my .avs in gknot? i think my source has bottom field first...

It's a bit tricky but worth the effort. First, here are the links:

Tdeint():
http://bengal.missouri.edu/~kes25c/TDeintv1RC1.zip

EEDI2():
http://bengal.missouri.edu/~kes25c/EEDI2v09.zip

Here is tritical's explanation of usage; choose the appropriate case:

-----
1.) bff clip, keep bottom field, interpolate top field:

interp = separatefields().selecteven().EEDI2(field=0)
tdeint(order=0,field=0,edeint=interp)

2.) bff clip, keep top field, interpolate bottom field:

interp = separatefields().selectodd().EEDI2(field=1)
tdeint(order=0,field=1,edeint=interp)

3.) tff clip, keep bottom field, interpolate top field:

interp = separatefields().selectodd().EEDI2(field=0)
tdeint(order=1,field=0,edeint=interp)

4.) tff clip, keep top field, interpolate bottom field:

interp = separatefields().selecteven().EEDI2(field=1)
tdeint(order=1,field=1,edeint=interp)
-----

So, for example, my BFF DV video uses this script (case 1):

avisource("wedding.avi")
interp = separatefields().selecteven().EEDI2(field=0)
tdeint(order=0,field=0,edeint=interp)

Note: This works only for single-rate output right now; do not try to use it as a bobber.

When set up correctly, the results are outstanding. Who needs Faroudja?

See the thead starting here:

http://forum.doom9.org/showthread.php?p=744308#post744308

[bananix]

EEDI2 with Tdint gives fine results, but now i have different dvd it's tff and the others were bff i belive this is also purely interlaced but it has lots of interlacing left after using this method. also the vob was joined from two cut vobs because the scene was not in one vob and had error probably because of it in one part with dgindex "Too many pictures per gop (>=500)" so i had to skip those few proplem frames when making the d2v.

can anyone tell what might be the difference with this?

[bananix]

well seems like LeakKernelDeint worked for it with this line

LeakKernelDeint(order=1,sharp=true,threshold=4)

so perhaps that EEDI2+TDeint is not good for deinterlacing all types of interlaced videos...