View Full Version : convolution3d vs SSHiQ in anime DVDs


jarthel
17th September 2002, 13:35
any feedback? I asked in the convolution3D thread but didn't get any reply. What's faster? Is convulotion3D designed only for noisy sources? Any other things on the comparison? Thank you.

Jayel

vlad59
17th September 2002, 13:48
Honestly, when I began to code Convolution3d, I didn't know there was a Avisynth port SSmoothHiq (sorry ShOdan ;) ). And I made no tests to compare those two filters. I just learn how to use it afterward.

In theory :
Convolution3d's main advantage is to use temporal informations.
SSHiq's main advantage is to have different radius support 3, 5, 7 (maybe more) and a smarter way of removing bad informations (diamond search).

For the rest you should make a test and post screenshot.

EDIT : Oooops I forgot to answer to one question. My first aim was to code a noise remover. Iago (for example) found that it is usefull (with low tresholds) to increase the compressibility of any movie.

jarthel
17th September 2002, 14:04
how do I get screenshots of movies?

thanks

Jayel

sh0dan
17th September 2002, 14:06
Load the avs into Vdub - use CTRL+1 to copy the frame to the clipboard, or use "save image sequence".

@vlad: Don't feel bad - your filter kicks a$$ :)

toshi
19th September 2002, 23:41
Jarthel, when you use 2DNR such as SSHiQ, you'd better use with temporal NR. (ex.TemporalSmoother)