Quantcast Avisynth: Motion Adaptive Filtering Now Possible? - Page 24 - digitalFAQ.com Forums [Archives]
  #461  
06-08-2003, 06:43 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by black prince
@Kwag,

I was browsing thru old posts and came across this link:

http://shelob.mordor.net/dgraft/msmooth/msmooth.html

I haven't tried MSmooth
I have tried it, and the result is
Just load the dll and add "MSmooth(strength=7,debug=true)" anywhere on the current script. You'll see what I mean
You can even drop the strength down to 1, and still, it will mess up your picture. Maybe on cartoons it works better, but it definitely doesn't on regular movies.

-kwag
Reply With Quote
Someday, 12:01 PM
admin's Avatar
Site Staff / Ad Manager
 
Join Date: Dec 2002
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
  #462  
06-08-2003, 06:46 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by jorel

the quality remains after the change?
i don't see more posts about this.
no one more try the change?
I've tried on two movies, and the result is far better with "MotionEstimate". I guess the current script is doing a good pre-filtering job, that we can now use "MotionEstimate" instead of 'High quaity".
At least, I'm getting MUCH better results this way

-kwag
Reply With Quote
  #463  
06-08-2003, 06:53 PM
jorel jorel is offline
Invalid Email / Banned / Spammer
 
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by kwag
Quote:
Originally Posted by jorel

the quality remains after the change?
i don't see more posts about this.
no one more try the change?
I've tried on two movies, and the result is far better with "MotionEstimate". I guess the current script is doing a good pre-filtering job, that we can now use "MotionEstimate " estimation instead of 'High quaity".
At least, I'm getting MUCH better results this way

-kwag
thank Kwag


i have tons to encode and the new scrip is really cool
and changing ToK to "faster" is fantastic

my 60gb hd is full of vobs.

Reply With Quote
  #464  
06-08-2003, 06:58 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by jorel

my 60gb hd is full of vobs.

Yeah mine too
Now I have to go back an re-encode a lot of my DVDs
The difference is just too great to previous encodes
I'm going to look for my "K-Pax" movie. That movie, encoded with this script, should look like the original DVD

-kwag
Reply With Quote
  #465  
06-08-2003, 07:37 PM
mrj03 mrj03 is offline
Free Member
 
Join Date: Nov 2002
Posts: 46
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by kwag
And here's the showoff for the day
www.kvcd.net/704x480-showoff.mpg ( CQ=68 MIN=300, MAX=2,500 ~6MB 25 seconds , which should be about the right CQ to fit this particular movie on one CD, looking just like this sample )

Enjoy!,
-kwag
I don't know about everyone else but I'd REALLLYYYY like to DOWNLOAD ALL of kwag's samples along with which template he used for each one. This would be GREATLY appreciated.......

THANKS....... GREAT SAMPLES......
Reply With Quote
  #466  
06-08-2003, 07:46 PM
black prince black prince is offline
Free Member
 
Join Date: Jul 2002
Posts: 1,224
Thanks: 0
Thanked 0 Times in 0 Posts
@jorel,

jorel wrote:
Code:
hey bp, 
read my last post about the change of "video.en1" file. 
works ok and faster.
Thanks jorel Works great now

@Kwag,

Your right. MSmooth() was pretty bad at any setting, but UnDot()
doesn't show much improvement either. Encoding speeds up without
it too.

-bp
Reply With Quote
  #467  
06-08-2003, 08:28 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by mrj03

I don't know about everyone else but I'd REALLLYYYY like to DOWNLOAD ALL of kwag's samples along with which template he used for each one. This would be GREATLY appreciated.......

THANKS....... GREAT SAMPLES......
Hi mrj03,

All the samples in this thread were done with ToK. In ToK, there's no notion of a tempate. You just select the resolution you want, and ToK will make your KVCD mpeg with all parameters already set
That is, the Q. matrix and GOP settings are already taken care by ToK. All samples I did in this thread were done at 704x480

-kwag
Reply With Quote
  #468  
06-08-2003, 08:30 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by black prince
UnDot()
doesn't show much improvement either. Encoding speeds up without
it too.

-bp
I'm taking a look at that right now
Maybe UnDot is good for some other sources, but not for DVDs
I'm testing now..

Edit:

UnDot() stays
It makes a HUGE difference
The sample I posted before "red-planet-19second.mpg", took 59 seconds to encode without undot, and 60 seconds with undot. Only one second longer.
But the file size without undot is 2,148KB and with undot, it's 2,109. And that makes a big difference on a complete movie


-kwag
Reply With Quote
  #469  
06-08-2003, 08:43 PM
ovg64 ovg64 is offline
Free Member
 
Join Date: Jan 2003
Location: Puerto Rico
Posts: 423
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via MSN to ovg64
know wonder CQ when down when i took out the filter
Reply With Quote
  #470  
06-08-2003, 09:13 PM
ovg64 ovg64 is offline
Free Member
 
Join Date: Jan 2003
Location: Puerto Rico
Posts: 423
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via MSN to ovg64
@ all,

You might also want to take a look at Cnr2 filter this is a noise reduction filter on at 15 sec clip i got,

Frames 368 Default setting Cnr2()
with Cnr2 = 2182kbs
with out Cnr2 = 2204 kbs

thats a 22 kbs differece, encode speed seams the same, quality is so good with 252 that i couldn't tell you that it looks any better.
Reply With Quote
  #471  
06-08-2003, 10:02 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by ovg64
@ all,

You might also want to take a look at Cnr2 filter
Cnr2 is a winner
The file size is now even smaller, and the time to encode was just the same as without the filter
Thanks ovg64
I just updated the current script

EDIT: CNR IS A LOOSER
Look at this without CNR:


And look at this flaw using CNR2:


I just noticed this going back and forth in Vdub, and I noticed something strange
So I went back to the scene and look at that thing. It screwes up the chroma big time
It could be, maybe because of the dynamic filtering being applied. I don't know, but it's out of the script for the time being
There is one frame difference between the screenshots, but the one that looks bad, looks correct without Cnr2.

-kwag
Reply With Quote
  #472  
06-08-2003, 11:28 PM
jorel jorel is offline
Invalid Email / Banned / Spammer
 
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by kwag
Hi jorel,

Maybe it's your 29.97fps movie you're trying to encode
Have you tried the script on a 23.976 Force FILM .d2v

-kwag
hey Kwag,
you give some more hints here.

if my movie show this values in the
statistics window from dvd2avi 176:
aspect radio - 4:3,
frame rate - 29.970 fps,
video type - ntsc,
frame type - interlaced...

can i use forced film to get more CQ

thanks
Reply With Quote
  #473  
06-08-2003, 11:48 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by jorel

can i use forced film to get more CQ

thanks
Not if the source is interlaced and was not originally FILM

-kwag
Reply With Quote
  #474  
06-09-2003, 12:04 AM
jorel jorel is offline
Invalid Email / Banned / Spammer
 
Join Date: Aug 2002
Location: Brasil - MG - third stone from the sun
Posts: 5,570
Thanks: 0
Thanked 0 Times in 0 Posts
ok,thanks
i always forget this "feature".

..cnr2 got a pleasure to stay some minutes in the new script.
the picture with cnr2 have a horrible error.

is need changes or some filter to turn the script better
Reply With Quote
  #475  
06-09-2003, 12:29 AM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by jorel
is need changes or some filter to turn the script better
The only thing I changed in the script was the scd_trigger = 25 # Scene change trigger value.
It was 15, and it was "falsing" on some scenes as if they were scene changes. So at 25, I have not seen that problem. I also slightly adjusted asharp( -(fmin((nf/30), 1)), 0 )
I had 35 before. Seems that 30 is more withing the "bluring" range we need. Over 35, we start loosing some blur on some key movements, so we loose compression.
These are minor tweaks, so I would say that the script is 100% useable as it is right now, and even the one posted before this update, excluding the one that had Cnr2

-kwag
Reply With Quote
  #476  
06-09-2003, 03:51 AM
Bchteam Bchteam is offline
Free Member
 
Join Date: Jan 2003
Posts: 275
Thanks: 0
Thanked 0 Times in 0 Posts
Hi Kwag,

Do you finally recommend to use "MotionSearchEstimate" instead of "Hihgh Quality"
Reply With Quote
  #477  
06-09-2003, 08:33 AM
Icarus3000 Icarus3000 is offline
Free Member
 
Join Date: Feb 2003
Posts: 258
Thanks: 0
Thanked 0 Times in 0 Posts
Wow! you guys are picky! ( I guess that's why we count on you to come up with the perfect script! )

For those of us with less discerning eyes, can you explain where in the sample I should be looking for this terrible flaw?

Thanks,
Icarus
Reply With Quote
  #478  
06-09-2003, 09:16 AM
ovg64 ovg64 is offline
Free Member
 
Join Date: Jan 2003
Location: Puerto Rico
Posts: 423
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via MSN to ovg64
Quote:
I just noticed this going back and forth in Vdub, and I noticed something strange
So I went back to the scene and look at that thing. It screwes up the chroma big time
It could be, maybe because of the dynamic filtering being applied. I don't know, but it's out of the script for the time being
There is one frame difference between the screenshots, but the one that looks bad, looks correct without Cnr2.

-kwag
I dont know Kwag I went frame by frame with bigger
samples and i didn't see any of that on mines, could it be
the position where you put the filter in the script or maybe just a conflict between to filter than again i dont use STMFilter in my script i use another Spatial filter. But i got to admit the image is dischomanise, is that a word
Take a look at my script:

LoadPlugin("C:\Documents and Settings\Osvaldo\My Documents\Avi Synth 2.51\MPEG2Dec3.dll")
LoadPlugin("C:\Documents and Settings\Osvaldo\My Documents\Avi Synth 2.51\asharp.dll")
LoadPlugin("C:\Documents and Settings\Osvaldo\My Documents\Avi Synth 2.51\deen.dll")
LoadPlugin("C:\Documents and Settings\Osvaldo\My Documents\Avi Synth 2.51\Dnr2_for_25.dll")
LoadPlugin("C:\Documents and Settings\Osvaldo\My Documents\Avi Synth 2.51\deen.dll")
LoadPlugin("C:\Documents and Settings\Osvaldo\My Documents\Avi Synth 2.51\undot.dll")

MaxTreshold = 1.50
scd_trigger = 15 # Scene change trigger value.
nf = 0 # Next frame.
Mpeg2Source("C:\Documents and Settings\Osvaldo\Desktop\Red Planet\RP.d2v")
Limiter()
asharp(2,0)
BicubicResize(528, 366, 0, 0.6, 5, 0, 710, 480)
Undot()
Dnr2()
deen("a2d",4,10,12)
TemporalSoften(2,7,7,3,2)
MergeChroma(blur(1.50))
ScriptClip("nf = YDifferenceToNext()"+chr(13)+ "nf > 2.5 ? asharp( -(fmin((nf/35), 1)), 0 ) : \
TemporalSoften(2,7,7,3,2) ")
ScriptClip("nf > scd_trigger ? asharp( -2,0) : asharp(0,7) ")
AddBorders(0, 57, 0, 57)
LetterBox(0, 0, 16, 16)
function fmin(float f1, float f2) {
return (f1<f2) ? f1 : f2
}

Remember the changes on the adaptive line is just 4 me.
Reply With Quote
  #479  
06-09-2003, 09:35 AM
black prince black prince is offline
Free Member
 
Join Date: Jul 2002
Posts: 1,224
Thanks: 0
Thanked 0 Times in 0 Posts
@Kwag,

Kwag wrote:
Quote:
I'm taking a look at that right now
Maybe UnDot is good for some other sources, but not for DVDs
I'm testing now..

Edit:

UnDot() stays
It makes a HUGE difference
The sample I posted before "red-planet-19second.mpg", took 59 seconds to encode without undot, and 60 seconds with undot. Only one second longer.
But the file size without undot is 2,148KB and with undot, it's 2,109. And that makes a big difference on a complete movie


-kwag
Maybe it's time to look at DCTFilter. SansGrip was experimenting with it
some time ago to correct mosquitos and found it produced a 30%
decrease in file size without any noticable effect to picture quality
This tweaking is really fun, since picture quality is already great and issues
of encoding speed and filesize remain.

-bp
Reply With Quote
  #480  
06-09-2003, 10:08 AM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by Bchteam
Hi Kwag,

Do you finally recommend to use "MotionSearchEstimate" instead of "Hihgh Quality"
With this script, Yesssss

-kwag
Reply With Quote
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
Avisynth: Dynamic Linear Adaptive Filtering and Scene Change Detection supermule Avisynth Scripting 3 09-15-2006 01:45 AM
Avisynth: Message from linear motion adaptive filtering? holgerschlegel Avisynth Scripting 7 08-27-2003 04:08 AM
Avisynth: Help with Linear Motion Adaptive Filtering pitoman Avisynth Scripting 2 08-05-2003 12:51 PM
Avisynth: Motion adaptive filtering good enough? bicho_visacoso Avisynth Scripting 6 06-15-2003 06:30 AM
Avisynth: AVS 2.5x Script, Motion Adaptive Filtering problems? Bchteam Avisynth Scripting 15 05-31-2003 12:38 PM

Thread Tools



 
All times are GMT -5. The time now is 04:09 AM  —  vBulletin © Jelsoft Enterprises Ltd