digitalFAQ.com Forums [Archives] - Avisynth: No need for Blockbuster ?

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)

- Avisynth Scripting (http://www.digitalfaq.com/archives/avisynth/)

- - Avisynth: No need for Blockbuster ? (http://www.digitalfaq.com/archives/avisynth/3406-avisynth-blockbuster.html)

Hi girv, i'm gonna try out ur script, coz i'm doing pure divx/xvid conversions myself, and will give u some feedback!

btw...does the script have to be in same order as u wrote it, or doesnt it matter?

Cheerz
TheDJ® 8)

girv, believe me.

first we see some strange in temporalsmoother, don't?

incredible...
i'm testing blockbuster in the new script from 3 days ago at now 8O
tons of samples with and without this filter.
see my first post in this page:
http://www.kvcd.net/forum/viewtopic.php?t=4043&start=48

i'm making hard tests here friend.
wait for my results to compare with yours :wink:

are we "paranormals"? ... 8O
:lol: :lol: :lol:

Quote:

Originally Posted by TheDJ®

Thanks for testing DJ® :) Yes, the order of the filters is important.

Here is the latest version. It seems to give better compression (500Kb less on a 19Mb encode) and better smooths out blocky colour gradients in flat backgrounds.

It does take a long time to run compared to Kwag's DVD optimal script, but IMHO the quality improvement for xvid sources is worth it. Roughly it takes 5.5 hours to encode 90 minutes on an XP2000+ / DDR226.

Code:

AddBorders(32,32,32,32) 

Blockbuster(method="dither",detail_min=1,detail_max=10,variance=0.3,seed=5832) 

MSoften(3,3,4,3,pre=1,post=1,speed=4,showmask=false) 

Crop(32,32,Width()-64,Height()-64) 

Blockbuster(method="noise",detail_min=1,detail_max=8,variance=0.5,seed=5823) 

FluxSmooth() 

Sharpen(0.3)

BicubicResize(432,336)

Let me know your results!

Hi girv,

You might as well drop "Blockbuster(method="dither..", because the result is "un-natural" :!:
The "noise" creates a more natural effect, and the "dither" creates a picture like if you are looking through a "window screen".
This was talked a lot some time ago, and "dither" was an experimental option provided by SansGrip.

-kwag

Quote:

Originally Posted by kwag

You might as well drop "Blockbuster(method="dither..", because the result is "un-natural" :!:

...but with very low variance the added noise is invisible to the eye yet still helps MSoften reduce flickering DCT blocks on low detail areas. The idea is that added noise is averaged out over the low detail area by MSoften to give a smoother surface.

I know what you're saying re: method="dither", and in fact I started with method="noise" for just that reason, but this gave me slowly "wandering" patterns of pixels (if you've ever messed with "life" or other cellular automata you'll know what I mean) so I changed to method="dither" which gave me a steadier picture.

Quote:

Originally Posted by girv

so I changed to method="dither" which gave me a steadier picture.

Did you try the "noise" method with a seed of 0 :?:
I believe that will create a static noise pattern instead of random pattern on every frame, and it will achieve the same as dither, but without the dithering pattern :!:
I recall SansGrip mentioning something about that a while ago :roll:

-kwag

@kwag: from the Blockbuster docs:

Quote:

The seed can be any number between 0 and 2,147,483,647. When zero, the system's current time is used.

Quote:

method="dither"

This method is very similar to the "noise" method, with the only difference being this method will add the same noise to each frame of the clip, whereas the "noise" method will add different noise to each frame.