Avisynth: Motion Adaptive Filtering Now Possible? - Page 5

kwag · #81 05-28-2003, 03:21 PM

Quote:

Originally Posted by bman

I'm trying to encode AVS2.51 but can't find unfilter() function

Hi bman,

Here: http://www.avisynth.org/~warpenterprises/
The filter's MECA

-kwag

bman · #82 05-28-2003, 03:38 PM

@ KWAG !
Thenx

What about vobsub or subtitles with avs2.51

bman

kwag · #83 05-28-2003, 03:42 PM

Quote:

Originally Posted by bman

@ KWAG !
Thenx

What about vobsub or subtitles with avs2.51

bman

Good question

I haven't tried that yet

-kwag

audioslave · #84 05-28-2003, 05:24 PM

Thanks a bunch Kwag! I'll get right to business

Anerboda · #85 05-28-2003, 05:40 PM

Quote:

Originally Posted by Boulder

Sampler does work in AVS2.5.1 and should work since v0.2. I use it all the time, just remember to use sampler-2.5.dll instead of sampler.dll . They're both in the same package.

True.. I just did a test with ToK and it is working... in settings just point ToK to the Sampler-2.5.dll from the package.

But I got a CQ of 62,83 with AviSynth 2.5 and with AviSynth 2,08 I got a CQ of 70,12 with the same settings in ToK for file prediction (optimal scripts for both versions of AviSynth) with the same movie.

Anerboda

audioslave · #86 05-28-2003, 06:17 PM

@Kwag

Could you PLEASE show me what lines to remove to get rid of the text on the top of the movie? I just can't seem to get rid of the text...

Please show me on the script, will you? BTW What settings do you recommend for a "normal" DVD encode? To get a "normal" amount of noise? If you know I mean...
I got the aspect ratio thing working though. I added the "AddBorders()" line I got in "MovieStacker" and simlpy removed the side borders

. Like this:

BicubicResize(448, 414, 1/3, 1/3, 9, 0, 702, 576)
AddBorders(0, 81, 0, 81)

Then, in TMPGEnc, I set the "Video Arrange Method" to "Full Screen".

kwag · #87 05-28-2003, 06:55 PM

Hi audioslave,

Get your resize values from moviestacker, and paste them on the corresponding lines. Use the "Latest Script" posted as reference.
Don't mess with blockbuster "noise" yet! I posted that just as an example

If you use the "Latest script", it will work just fine, and the subtitle (numbers on top) won't appear in your encode. Get the latest script here: http://www.kvcd.net/forum/viewtopic.php?t=3483

Which is this:

Code:

MaxThreshold=1.58

UnFilter(50, 50)
BicubicResize( Your_Resize_Values_Here )
STMedianFilter(8, 32, 0, 0 )
TemporalSoften(2,7,7,3,2) # Experimental!
mergechroma(blur(1.50))
## Dynamic linear adaptive filtering ##
ScriptClip("val=YDifferenceFromPrevious()/14.55" + "val > MaxThreshold ? MergeLuma(blur(MaxThreshold)) : MergeLuma(blur( val ))")
##
#LetterBox( Your_Values_Here ) # Depends on situation. Use MovieStacker!
#AddBorders( Your_Values_Here ) # Depends on situation. Use MovieStacker!
Limiter()

-kwag

audioslave · #88 05-28-2003, 07:19 PM

Thank you kwag! Sorry for being a pain in the ***

.
I'll stick with the Optimal Script for now...
Off to bed now. It's 02:18 p.m. here in Sweden. Yawn! Good night to you all...

ozjeff99 · #89 05-28-2003, 11:16 PM

Recognition of all contributors and puting personal attention aside for the benefit of the whole team is a mark of the finest leadership. We are here, and have come this far so quickly because of you Kwag. You inspire us all.

Thanks mate...ozjeff99

kwag · #90 05-28-2003, 11:37 PM

Thanks ozjeff99

You all make this site a great place to be, learn, study and play

-kwag

bman · #91 05-29-2003, 12:33 AM

@ ozjeff99
I'm with u on every word u sead !
KWAG - U are THE ONE !!! and I mean it !

@ KWAG
====
ScriptClip("val < 0.3 ? BlockBuster(method=N, detail_min=1, detail_max=10, variance=((1/val)*10), seed=0).Subtitle(String(Subnoise),1,30) : val >1 ? blur( 1.58 ).Subtitle(String(SubBlur),1,30):blur(val)")
=====
This line doesn't want to work for me ? Ofcourse I removed Subtitle function and instead of 1 I delete 0.1 on val :

ScriptClip("val < 0.3 ? BlockBuster(method=N, detail_min=1, detail_max=10, variance=((0.1/val)), seed=0) : val >1 ? blur( 1.58 ) : blur(val)

I know it's not right but where ???
I'm getting too much noise added to dark and and even bright scenes too .
After encoding at the very begining of clip I'm getting few frames looking like marble surface ??!!!!
I messed everything I think

bman

kwag · #92 05-29-2003, 12:49 AM

Quote:

Originally Posted by bman

ScriptClip("val < 0.3 ? BlockBuster(method=N, detail_min=1, detail_max=10, variance=((1/val)*10), seed=0).Subtitle(String(Subnoise),1,30) : val >1 ? blur( 1.58 ).Subtitle(String(SubBlur),1,30):blur(val)")
=====
This line doesn't want to work for me ? Ofcourse I removed Subtitle function and instead of 1 I delete 0.1 on val :

It should read like this:
N="Noise"
ScriptClip("val < 0.3 ? BlockBuster(method=N, detail_min=1, detail_max=10, variance=((1/val)), seed=0) : val > 1 ? blur( 1.58 ) : blur(val)")
The idea of 1/val is to work in reverse. The value of "val" fluctuates from close to 0 to around ~25. So what I did is divide the number by one, so that the higher the activity, the lower the result and vice versa. This is the effect we want, and that is to apply more noise on slow or static scenes and decrease the noise in a linear fashion as the activity increases. This way we apply the noise inversely proportional to the activity

You can play with the (1/val) to include a multiplier, like ((1/val) * 10 )
That will increase the noise, but probably too much! But go ahead and experiment

that's what this is all about

-kwag

kwag · #93 05-29-2003, 01:06 AM

Here you go guys

www.kvcd.net/small-adaptive-sampler.mpg (10 second clip)
That's what the complete movie looks like on one CD-R

That movie is 96 minutes long, and it was encoded at 528x480 and @CQ=71.2 with the current script ( For AviSynth 2.5x )

-kwag

jorel · #94 05-29-2003, 01:41 AM

@ Kwag and all

i download the full-adaptive-sampler.mpg (11.8mb)
and the small-adaptive-sampler.mpg (1.04mb).

after see the full sampler i can affirm...

great, fantastic,...

congrats Kwag the full sampler is brilliant.(i don't need to see the small)
very clear and with big details.

i need to read all this thread.(again)

DorvalCS · #95 05-29-2003, 02:26 AM

Quote:

Originally Posted by jorel

i download the full-adaptive-sampler.mpg (11.8mb)

Jorel my friend... Where did you get this full-adaptive-sampler???

I've looked everywhere and I can't find it

DorvalCS

jorel · #96 05-29-2003, 02:36 AM

@ DorvalCS and all

don't know if the download is working now...

http://www.kvcd.net/full-adaptive-sampler.mpg

bman · #97 05-29-2003, 03:07 AM

KWAG wrote:

Quote:

It should read like this:
N="Noise"
ScriptClip("val < 0.3 ? BlockBuster(method=N, detail_min=1, detail_max=10, variance=((1/val)), seed=0) : val > 1 ? blur( 1.58 ) : blur(val)")

I begane with this line from very begining

Quote:

The idea of 1/val is to work in reverse. The value of "val" fluctuates from close to 0 to around ~25. So what I did is divide the number by one, so that the higher the activity, the lower the result and vice versa. This is the effect we want, and that is to apply more noise on slow or static scenes and decrease the noise in a linear fashion as the activity increases. This way we apply the noise inversely proportional to the activity

Brilliant idea , Brilliant !

Quote:

You can play with the (1/val) to include a multiplier, like ((1/val) * 10 )
That will increase the noise, but probably too much!

I did played some but I don't know why I can't get smooth picture

Then I just decreased 1 to 0.1 and even 0.01 and it made significant difference

Quote:

But go ahead and experiment

that's what this is all about

Ofcoarse I'll do that as soon as I'll arrive home .
And exscuse me for my stupidity but BlockBuster("noise") I always used to get rid of blinking colored blockiness on dark scenes ???
I assumed that we'll use it in similar way in adaptive script ??!!

bman

rhino · #98 05-29-2003, 04:15 AM

It will be interesting to see what effects this has on long films and fitting them onto 1 CD. I was re-reading the thread about Pearl Harbour being 183 minutes, lots of action but better on 2 CD's. Wonder what effect dyanmic filtering will have on this.

Cheers,

Jellygoose · #99 05-29-2003, 05:54 AM

I'm joying, I'm cheering!!

Just watched the small-sampler over and over again! awesome!
can't wait to test this!! I'll be back in a couple of hours!

By the way bman... As for the noise, I wouldn't mess with the (1/val), to change the amount of noise, but with the multiplier... so if you want less noise added, try something like this ((1/val)*0.5)
I don't know if I understood you right, but this would add half as much noise as before...

Jellygoose · #**100** 05-29-2003, 07:01 AM

Hi all!!

I'm currently testing the new script, with a full prediciton of "The Boondock Saints" using ToK... and...
it seems to work... I'll check with ya'll later!