View Full Version : The old Gknot cleaners in new avisynth 2.5?

11th February 2003, 09:34
Iwe been wanting to switch to the new avisynth 2.5 alpha for a while so i can use the new dup filter (which has been discontinued for the old avisynth). More spesifically i need the chroma test thats in the new version or else dup will make fade scenes in my anime "jump" because it dosnt detect the darkness/brightness difference (or i have to set threshold so low that it dosnt do anything good for the rest of the anime).

The main reason for now switching allready is that i cant seem to find (or find info about) the old cleaners which were used in Gknot as the part of the noise reduction options. These have worked better then most custom filters for me, so id love to be able to continue using them, or atleast have the option to use them. Im looking the the spatial and temporal smoother functions, and one of them didnt even have its own plugin, so is this somethng thats included in avisynth itself or something? ..

Thanks in advance for any assistance you can provide =)


11th February 2003, 10:06
Are there any "cleaners" that you're talking about in particular? Most of the information about avisynth 2.5 filters can be found with a simple search or checking the two sticky topics (the YV12 one has a link to a list of avisynth 2.0x and 2.5 filters). Many of the gknot cleaners are out of date and can probably be replaced by other newer filters using similar but more efficient techniques. Please browse those two stickies and use the search feature to see what else has been posted on a related topic and hopefully you'll find what you're looking for.

Mr. B

11th February 2003, 12:17
Why don't you just open the old scripts and see? :confused:

11th February 2003, 14:46
I should have posted this in the first post, but i didnt have any .avs availiable to look at at my school. These are the 2 im looking for:

TemporalSmoother() (dosnt seem to need a plugin dll)
SpatialSoftenMMX() (uses Avisynth_Spatial.dll)

Im fully aware that there are other filters for cleaning, both for the new 2.5 and the old, but as iwe said iwe tried a lot of them and i still think these do an exelent job on anime DVDsources (and normal movies too) even though they are pretty slow.

Iwe tried searching and looking through the stickies, but i wasnt able to find anything, but of course i may have missed something, so il check em again to make sure.


11th February 2003, 16:45
Temporalsoften is still the same in 2.5 - except faster and with a new mode, and scenechange detection.

Spatialsoften - you could use deen3d in 2d mode or convolution3d with temporal parameters at 0.

Experiment with temporalsoften - it can provide you with very excellent results at quite reasonable speeds.

12th February 2003, 00:59
Hmm.. i jsut cant seem to get the old temporalsmooth() command to work in 2.5 (no such command). I tried temporalsoften() just in case since i noticed you used that name.

I think it was a part of the old avisynth, how can i use it now?

Im probalby just beeing stupid and overlooking something obvious, but bear with me please =/


12th February 2003, 05:10
I posted a thread about TemporalSmoother here:

It doesn't look like it will be out for a while.

12th February 2003, 12:23
Ah! so atleast there IS such a function included in avisynth. i Just guess i have to read over al the avisynth docs until i can findthe command to use it then.


12th February 2003, 13:15
Let me clarify:

1) There is no function TemporalSmoother in AviSynth v2.5x nor in v2.0x.

2) Using mpeg2dec.dll for v2.0x there's a TemporalSmoother available: http://users.win.be/dividee/avisynth.html.

3) Using mpeg2dec.dll for v2.5x (which outputs to YUY2, see www.avisynth.org) I think the same TemporalSmoother available. However I'm not entirely sure whether Donald also included those functions. So you have to try it out whether there is a function TemporalSmoother in mpeg2dec.dll for v2.5x.

12th February 2003, 17:45
Thank you wilbert =) Thats exactly what ui was wondering. Right now im using Marc's mpeg2dec3.dll (which dosnt seem to have the function)so il see if there are any other mpeg2dec dll's availiable. thanks for all your help.


12th February 2003, 23:35
OK, just one last question ^_^

I got temporalsmoother to work by using Dividee's mpeg decoder dll instead of Marc's

So now all i need is spatialsmootherMMX. The old dll for that dosnt work in 2.5 but i heard you talk about using it in another thread Wilbert (while i was searching for answers). However you didnt mention there where to get it.

Sorry for all the dumb questions, but its just been a while since i changed my filter collection so i guess im outdated on the news front on 2.5 ^_^


13th February 2003, 10:19
So now all i need is spatialsmootherMMX. The old dll for that dosnt work in 2.5 but i heard you talk about using it in another thread Wilbert (while i was searching for answers). However you didnt mention there where to get it.
Seems unlikely because there is no spatialsmootherMMX for 2.5.

1) Your encoding will be slower if you use plugins (such as temporal smoother) which work in YUY2.

2) Why don't you use other plugins like temporalsoften (internal filter) and deen3d (external plugin, see faq Q3.4) like Sh0dan mentioned ? These work with YV12.

13th February 2003, 11:54
I was refeering to this:
Added Q35:
Q35: When i try to use SpatialSoftenMMX, it crashes in VirtualDub ?
A: You probably used this filter before resizing. Apperently SpatialSoftenMMX doesn't like resolutions that cannot be divided by 8 or 16. So if you use it after resizing it works, of course.

from the thread named avisynth Q&A. My mistake.. i just assumed it was 2.5 Q&A, but obviously it was not then.

I will take your advice and check out other filtering possibilities. Do you knew what the biggest difference betweenb spatialsmoother and spatial soften is?


13th February 2003, 12:18
Q&A is for both v2.0x and v2.5x, but not everything was updated when v2.5 came :)

I don't use spatialsmoother and spatial soften so I can't comment on that. For analog capture I use (only included the smoothing):


or with two captures from different broadcastings: