Avisynth-make scripts for VirtualDub

One more similar question.

Which is the best deinterlance filter, and when to put in this two scripts?

So, which deinterlance, and were to put deinterlance and resize in this scripts?

Code:

LoadPlugin("DeGrainMedian.dll")
avisource("C:\Documents and Settings\My Documents\Video 2.avi")
ConvertToYuY2(interlaced=true)
degrainmedian(interlaced=true)
degrainmedian(interlaced=true)
crop(8,4,-8,-12)
AddBorders(8,8,8,8)

Code:

LoadPlugin("Convolution3d.dll")
avisource("C:\Documents and Setting\My Documents\Video 1.avi")
ConvertToYuY2(interlaced=true)
SeparateFields()
odd=SelectOdd.Convolution3D (1, 32, 128, 8, 32, 10, 0)  
evn=SelectEven.Convolution3D (1, 32, 128, 8, 32, 10, 0) 
Interleave(evn,odd)
Weave()
crop(8,4,-8,-12)
AddBorders(8,8,8,8)

I will use this script in Virtualdub, because i want to make divx from this scripts. Source is Dv avi(interlaced).

best regard

any comment?

After the source loading.
LeakKernelDeint() ot TDeint(). You must specify the field order.
Mind that after deinterlacing your video is not interlaced anymore. So edit the scripts in this relation.

I always work with DV avi material(bottom field first)

Code:

LoadPlugin("DeGrainMedian.dll")
avisource("C:\Documents and Settings\My Documents\Video 2.avi")
#does here to put resize or not
TDeint()
ConvertToYuY2(interlaced=true)
degrainmedian(interlaced=true)
degrainmedian(interlaced=true)
crop(8,4,-8,-12)
AddBorders(8,8,8,8)

Code:

LoadPlugin("Convolution3d.dll")
avisource("C:\Documents and Setting\My Documents\Video 1.avi")
#does here to put resize or not
TDeint()
ConvertToYuY2(interlaced=true)
SeparateFields()
odd=SelectOdd.Convolution3D (1, 32, 128, 8, 32, 10, 0) 
evn=SelectEven.Convolution3D (1, 32, 128, 8, 32, 10, 0)
Interleave(evn,odd)
Weave()
crop(8,4,-8,-12)
AddBorders(8,8,8,8)

1. Where is better to put resize, before or after deinterlance filter?

2. Does i need to load TDeint() or LeakKernelDeint() filter? (i mean dll file)

3. Does is necessary to specify the field order?

1. after the deinterlace. For education purposes try to put it before and see what happen to the output.
2. If they are not in the plugin folder of avisynth, yes. If they are in the plugin folder they autoload (and also every dll in the plugin folder would be autoloaded, so be carefull what you put there).
3. Read the documentation that come with the dlls and you will see.
Several remarks about the scripts. After the deinterlacing the video is not interlaced anymore. So, all this (interlaced=true) are pointless. Also you don't have different fields anymore. So separate fields is not necessary and folowing interleave and weave also. In other words after deinterlacing you should treat the video as progressive.

Which deinterlance is best?

And how to use convolution 3d filter if i dont have separate fields and wave(because i want to deinterlance)?

thanks

Originally Posted by SerbianBoss

Which deinterlance is best?

I don't know.

Thanks for script. But where to put resize and limited sharpen in this script?

Code:

LoadPlugin("Convolution3d.dll")
avisource("C:\Documents and Setting\My Documents\Video 1.avi")
TDeint()
ConvertToYuY2()
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0)
AddBorders(8,8,8,8)

And where to put resize and limited sharpen and crop in this script:

Code:

LoadPlugin("DeGrainMedian.dll")
avisource("C:\Documents and Settings\My Documents\Video 2.avi")
TDeint()
ConvertToYuY2()
degrainmedian()
degrainmedian()
crop(8,4,-8,-12)
AddBorders(8,8,8,8)

In both cases before add borders and after denoising. In the second script you need to put crop before degrainmedian lines. If you want DVD you don't need to resize, because DV AVI is already at DVD resolution. That means you need to add borders to fill up the frame.
As you are cropping 16 pixels horizontaly and 20 pixels verticaly you need to add the same numbers:
crop(8,4,-8,-12)
...
...
AddBorders(8,10,8,10)
If you want to resize include the resizing parameters in limitedsharpen.

I want to make scripts for xvid Virtualdub and this is the reason of deinterlacing and resizing.

So does this script looks better or not(make attention on deinterlance,resize,limitedsharpen and crop):

Code:

LoadPlugin("DeGrainMedian.dll")
avisource("C:\Documents and Settings\My Documents\Video 2.avi")
TDeint()
LanczosResize(512,384)
ConvertToYUY2(interlaced=false)
crop(8,4,-8,-12)
degrainmedian()
limitedsharpen ()
AddBorders(8,8,8,8)

Code:

LoadPlugin("Convolution3d.dll")
avisource("C:\Documents and Setting\My Documents\Video 1.avi")
TDeint()
LanczosResize(512,384)
ConvertToYuY2(interlaced=false)
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0)  
limitedsharpen()
AddBorders(8,8,8,8)

My finall scripts for xvid is this: I load this scripts in virtual dub and encode using xvid.

Code:

LoadPlugin("DeGrainMedian.dll")
LoadPlugin("tdeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYUY2(interlaced=false)
TDeint()
LanczosResize(512,384)
crop(8,4,-8,-12)
degrainmedian()
AddBorders(8,8,8,8)

Code:

LoadPlugin("Convolution3d.dll")
LoadPlugin("tdeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYuY2(interlaced=false)
TDeint()
LanczosResize(512,384)
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0) 
AddBorders(8,8,8,8)

And picture quality is Ok. I tested on very short material(1 minute) and i think that is OK.But it encode very very slow 8-10 FPS on my dual amd.

Maybe is better to use internal VD filters for deinterlance,resize...maybe faster

I don't believe you need ConvertToYUY2(interlaced=false) in the first script. You don't need it in the 2nd script unless you're using that old version of Convolution3D. I could be wrong, though, if your source is RGB.

If you want your script to encode a bit faster, substitute LeakKernelDeint(Order=1) for TDeint(). Almost as good and quite a bit faster.

Maybe is better to use internal VD filters for deinterlance,resize...maybe faster

Not better. Not faster.

My source is Dv avi(bottom field first),but when i put in virtualdub it become RGB write or no?
So does is better like this ConvertToYuY2()?

Wow, with LeakKernelDeint i have 35 fps instead 8 fps with tdeint

But should i set LeakKernelDeint(Order=0) instead order 1. In manual sais if have BTT to set order 0

So scripts is:

Code:

LoadPlugin("DeGrainMedian.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYuY2()
LeakKernelDeint(Order=0)
LanczosResize(512,384)
crop(8,4,-8,-12)
degrainmedian()
AddBorders(8,8,8,8)

Code:

LoadPlugin("Convolution3d.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYuY2()
LeakKernelDeint(Order=0)
LanczosResize(512,384)
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0) 
AddBorders(8,8,8,8)

Is that Ok.

Hi-

But should i set LeakKernelDeint(Order=0) instead order 1.

Yes, if your video is BFF, then Order=0. And if the video is in RGB, then yes, you need a convert call, although I think I'd ConvertToYV12() unless you're using a YUY2 only Convolution3D You can find out the colorspace by adding Info() to the end of the script and opening it in VDub(Mod). Of course, if you've already converted the colorspace, then it's going to give that colorspace. To find out the original colorspace, temporarily comment out (put a "#" in front of it) the Convert line (and maybe the filters that require it).

Also, for faster encodes, make sure you're using FastRecompress in VDub(Mod). Video->Fast Recompress, and not Full Processing Mode.

When this script i load in virtual dub mod i see:

Code:

LoadPlugin("Convolution3d.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYuY2()
LeakKernelDeint(Order=0)
LanczosResize(512,384)
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0) 
AddBorders(8,8,8,8)

I am using YUY2 version of convolution 3d.

So how does script looks like?

And how to put some sharpening for deinterlance(i try Order=0,sharp=true and Order=0, sharp=false) but i dont have any sharpening on movie.

best regard

I don't see Info() anywhere in that script. If you added it at the bottom, then you had already converted the colorspace, and you didn't comment out the lines as I suggested before. You can add Info() right after AVISource and it'll tell you what the colorspace of the AVI is.

I always capture with canopus 110 in DV avi, so when that import in VD mod i have this:

Code:

avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Info()

From that script i make these scripts(for compressing in virtualdub)

Code:

LoadPlugin("Convolution3d.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYuY2()
LeakKernelDeint(Order=0,sharp=false)
LanczosResize(640,480)
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0) 
AddBorders(8,8,8,8)

Code:

LoadPlugin("DeGrainMedian.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYuY2()
LeakKernelDeint(Order=0)
LanczosResize(640,480)
crop(8,4,-8,-12)
degrainmedian()
AddBorders(8,8,8,8)

I changed resolution to 640x480. How now looks scripts, and how to add little sharpening?

Not correct.
You are converting to YUY2 interlaced material in these scripts. Because your converting line is before the deinterlacing line.
Now if I go to make them they should look as follows:

Code:

LoadPlugin("Convolution3d.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
LeakKernelDeint(Order=0,sharp=true)
ConvertToYV12()
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0)
LanczosResize(634,464)
AddBorders(8,8,8,8)

Better get the YV12 version of Convolution3D from http://www.avisynth.org/warpenterprises
name is convolution3Dyv12.
The second script

Code:

LoadPlugin("DeGrainMedian.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
LeakKernelDeint(Order=0, sharp=true)
ConvertToYV12()
crop(8,4,-8,-12)
degrainmedian()
LanczosResize(634,464)
AddBorders(8,8,8,8)

To avoid further color space conversion in VDub use Fast Recompress. Thus all the steps would keep YV12.

When i first put LeakKernelDeint and then ConvertToYV12() i get this error:


But when i first put ConvertToYV12() and than LeakKernelDeint that is OK.

My dv avi material is RGB24, is there difference if i converttoyv12 or converttoyuy2?

And one more, with sharp=true i dont see any sharpening in my film, maybe another metod?

Hmm, damned, your source is RGR24 and LeakKernelDeint support "only" RGB32.
Then:

Code:

...
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYV12(interlaced=true)
LeakKernelDeint(Order=0,sharp=true)
 ...
...

for both scripts.
sharp=true will preserve some more detail in the areas it deinterlace. It will not sharpen.

thanks for quick answers

I am curios, why is better to use ConvertToYV12 instead ConvertToYuY2 ?

Because the output of XviD compressor is YV12. So you have convert only once - through Avisynth. Otherwise you convert twice -RGB24->YUY2->YV12. Less color space conversions - better quality. And the second - avisynth filters operate in YV12 a little bit faster.

thanks, i didnt know that xvid output YV12

So, can i put ConvertToYV12() instead ConvertToYV12(interlaced=true). I think that i neednt interlaced=true ?

And one more. I have Ok results with these scripts, but its little smoth. So where in scripts i can add little sharp filter?

Originally Posted by SerbianBoss

So, can i put ConvertToYV12() instead ConvertToYV12(interlaced=true). I think that i neednt interlaced=true ?

No. Well, I want to say it is not correct, but if you want - put it. You may have color issues.
The logic of scripts follow the logic of events. You load the source. It is interlaced. Until deinterlacing it will stay interlaced. Correct is
Deint()- you deinterlace
ConverttoXXX() you convert the color space. (the color space conversion is interlaced=false by default)
Or,
ConvertToXXX(interlaced=true) - you convert interlaced source
Deint() - you deinterlace
The script is "executed" line by line.

Originally Posted by SerbianBoss

I have Ok results with these scripts, but its little smoth. So where in scripts i can add little sharp filter?

Before addborders line seems OK for me.

yes i now understand why i must put ConvertToYV12(interlaced=true). So script looks like:

Code:

LoadPlugin("Convolution3d.dll")
LoadPlugin("LeakKernelDeint.dll")
avisource("C:\Documents and Settings\Nenad\My Documents\Video 2.avi")
Trim(0,3442)
ConvertToYV12(interlaced=true) //because source is interlaced,we must to put interlace=true
LeakKernelDeint(Order=0,sharp=true) // now deinterlance it
LanczosResize(640,480) // i found that script is faster when here is resizing(after deinterlance)
crop(8,4,-8,-12)
Convolution3D (1, 32, 128, 8, 32, 10, 0)
//here some sharpen
AddBorders(8,8,8,8)

Which sharpen is good?