AviSynth and VirtualDub filters - which one first?

[xara]

If I set up a VirtualDub internal filter and load a AVS script with another filter`s chain, which one will be applied first?

And, what useful filter`s chain should I build from these both roots?

tks

[stickboy]

Opening an AviSynth script applies the filters in the AviSynth script.

VirtualDub applies its filters to what it opens. VirtualDub doesn't know what AviSynth is doing; it just gets whatever frames AviSynth outputs. That is, VirtualDub is working on the output from AviSynth. Therefore, AviSynth filters happen first.

What VirtualDub filters do you need to apply? Usually it's better to stick with all AviSynth filters so you can stay in YUY2 or YV12 instead of converting to RGB32.

[xara]

Hummm...I was not paying attention in these color planes management...
So, you are saying meabe I'm losing data in those convertions?

I`m doing just a "sharpening" action with VirtualDub filter.
You will say, why dont you use the Didee's LimitedSharpen function?

Well, I try, I try and I try, but I dont get it.
Outside the speed, I just cant get the same gorgeous effect that you guys are showing in the LimitedSharp thread. Sometimes it makes sense to me...3 pages later it dont. Anyway, its a "before" keyboard problem here. :)

[Didée]

When you're already using Avisynth anyway, then additional filtering in Vdub really should be avoided. Unless one needs some very special filter for Vdub that doesn't have a counterpart in Avisynth, that is. And the sharpening filter in Vdub surely is not of this kind ;)

Show a sample of what you're going to sharpen up, and we'll see what's in reach and what not. And if it has to be by all means, things can also be made fast. ;)

[xara]

Here we go

Original

http://www.duna-design.com.br/p2p/doom1-original.jpg
http://www.duna-design.com.br/p2p/doom2-original.jpg


Internal VirtualDub sharp filter

http://www.duna-design.com.br/p2p/doom3-sharp.jpg
http://www.duna-design.com.br/p2p/doom4-sharp.jpg


LimitedSharpen function

http://www.duna-design.com.br/p2p/doom5-didee.jpg
http://www.duna-design.com.br/p2p/doom6-didee.jpg

AVS script
Import("D:\Program Files\AviSynth 2.5\plugins\LimitedSharpen.avs")
Import("D:\Program Files\AviSynth 2.5\plugins\Ylevels.avs")
AVISource("H:\dvd\dragao1-2pass.avi")
saturation_v = 0.8
cv = - (1-saturation_v)*256
ColorYUV(off_y=2, gain_y=-24, cont_u=0, cont_v = cv,opt="coring")
LimitedSharpen(Smode=2, strength=100,wide=true, soft=true)

Source are not interlaced.
The details that I get in the trees by the VD filter I cant achieve with the LS function. In fact, I dont see a difference between the original and the last image.

[Anonymouses]

Could you post a small unprocessed sample of this? Also, is the source that blocky or is that just due to jpeg artifacts because that looks horrible.

[xara]

Those are jpg artifacts. I used the lowest compression.
And I choose this scene because its a mixed of real picture and CGI animation, there are static objects and there is movement.

I'm encoding another footage right now.
If its really necessary I'll post it tomorrow, covering about 200 frames of the same scene.

[Pookie]

What animal is that ? Seriously. Is that a dinosaur ? That source is not good. Try LTSMC or ff3dfilter on it. LimitedSharpen is the icing on a cake, you still need a cake to bake :)

[stickboy]

Those are jpg artifacts. I used the lowest compression.
And I choose this scene because its a mixed of real picture and CGI animation, there are static objects and there is movement.What's the point of posting screenshots that show filtering results if they're going to be butchered by JPEG compression artifacts?

[Didée]

Do I see this right, that the source you're starting out with is an XviD file, 640x368, with a bitrate of 870 kbps? These are adequate properties to make a final encoding. But *starting* with such a one is, erhm, ...

[Anonymouses]

Those are jpg artifacts. I used the lowest compression.
And I choose this scene because its a mixed of real picture and CGI animation, there are static objects and there is movement.

Next time either post pngs or high quality jpgs. All I can see is a mess of blocks any "details" are all smoothed out by the compression. Those jpgs really don't help us at all.

[xara]

That video its not my source. Its the final encode that I was playing with and showing to you.

The pictures seems pretty nice in my monitor and I dont see any artifacts. Not if you are looking for sharp differences.

Anyway, I dont have the source anymore.

So, this is another footage, captured from satelite TV by a DVD recorder.
This means its a mpeg2, interlaced, cropped from the main file without any post processing.

http://www.duna-design.com.br/p2p/doom-dino.jpg

Sample MPG (http://www.duna-design.com.br/p2p/doom-sample.mpg) (5,4 MB)

But, as far as I understood, even if you guys point me LS parameters, this will work only for this footage... Well, meabe it will be a starting point.

Thanks in advice