New KVCD Script Candidate for Optimal Script!

Dialhot · #1 10-07-2003, 05:03 PM

Okay, I think is time to tell you about the last things I did on script for Divx.

The following script give purely fantastic results. The only counterpart is that it is quite slow (but not so much : I encode a 0.25x on my 1.3Ghz P4).

You can remove the second blockbuster if your original source isn't too much blocky to gain a little encoding time.

Code:

AviSource("PATH\NAME.avi",false)
BlindPP(cpu=4)
Blockbuster(method="noise",detail_min=1,detail_max=5,variance=0.1,seed=5823)
VagueDenoiser(threshold=1.5,method=1,nsteps=6,chroma=true)
GripCrop(WIDTH, HEIGHT, overscan=1, source_anamorphic=false)
GripSize(resizer="LanczosResize")
Undot()
TemporalSoften(2,7,7,3,2)
DCTFilter(1,1,1,1,1,1,0.5,0)
Blockbuster(method="noise",detail_min=1,detail_max=10,variance=0.3,seed=5623)
GripBorders()

Note1 : be carreful to have the last version of vaguedenoiser ( minimum 0.28 ). See http://kurosu.inforezo.org/avs/VagueDenoiser/index.html

Note2 : if vaguedenoiser complains taht your source dimensions are too small to make a step 6 denoising, just reduce the value of "nsteps" parameter).

Lefungus · #2 10-09-2003, 11:56 AM

Hi
For the vaguedenoiser part, Have you also tried method=2 ? It may keep more details, and you'll be able to use a little higher threshold

Dialhot · #3 10-09-2003, 03:51 PM

Lol.

I used 0, 1 and 3 but didn't try with method=2

.

I will. But doesn't it affect speed ? For the threshold, I found in debug mode that 1.5 is "far enougth but not too much".

Zolie · #4 10-09-2003, 04:16 PM

First off, you said not to get anything below 0.28 for VagueDenoiser, but the site you gave only gives links for 0.27.1 as the highest. and second, what are ther average times for encoding as opposed to your other optimal scripts?

Dialhot · #5 10-09-2003, 04:29 PM

Quote:

Originally Posted by Zolie

First off, you said not to get anything below 0.28 for VagueDenoiser, but the site you gave only gives links for 0.27.1 as the highest. and second, what are ther average times for encoding as opposed to your other optimal scripts?

There is an error in the text of the page : you can see that there are two 0.27.1 links, but the second one actually DL 0.28

The time is 25% more that a script using deen (that is not very fast itself).

A movie encoded in 7 hours on my PC was done in 9 hours with vaguedenoiser.

Zolie · #6 10-09-2003, 08:29 PM

is the difference really worth the 2 extra hours, or is it just a bit better. Oh and how much larger are the files, or smaller?

Edlund · #7 10-09-2003, 11:58 PM

This script is really slow for me, I will see later what quality does it produce.

incredible · #8 10-10-2003, 05:13 AM

Hi Dialhot,

Well Im trying a lot with Vague Denoiser on also other projects.
It seems that it works a bit heavy by "plaining" a lot of details for example in skins or hair-structures.
But in combination with TemporalSoften(2, .......) it deletes a lot

?

zes · #9 10-10-2003, 06:41 AM

i too think that it softens the picture too much. the previous script kept detail better imho.

incredible · #10 10-10-2003, 07:42 AM

I think we have to keep in mind that every Dvix/Xvid Source got its difference. So this seems to be a "general" Script.

If I got for example a very good encoded 1400MB Dvix of a movie I don't have to clean it that much and on the other hand on already bad encoded Dvix's, more cleaning makes big sense

zes · #11 10-11-2003, 03:43 AM

so that would result in changing which values to what?

Edlund · #12 10-11-2003, 06:08 AM

You can use another script for clean sources

Dialhot · #13 10-11-2003, 02:09 PM

Hi all,

Not a lot of time these days so I can't answer to everything and everyone. It seems that the main idea of all your test on this new script is that is to heavy on details removing.

Well. In fact I focused myself on block removals (that is the goal of a script for avi after all) and perhaps did not do enought comparison with previous script on this point (detail level).

I'm quite sure the problem isn't in vaguedenoiser but in temporalsoften. I add this line a long time after all my tests just to have a encoded size equivalent to the previous script. Perhaps the valus are to strong. Try to check this id you have time. I plan to make test using ATC insteed of temporalsoften. Sure it will be lighter (and faster).