Quantcast Avisynth: Some No-Good Results - digitalFAQ.com Forums [Archives]
  #1  
10-27-2003, 11:28 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
i encode harry potter - philosophica stone 4 times (152 minutes,21 seconds)


one with MA script:
blur in moviments scenes(frame 114829)
and less red in overall images(frame 120500).
without MergeChroma(blur(MaxTreshold)) the red is more natural

with the new Kwag (posted in CQMatic forum):
SwitchThreshold = (Width<=352) ? 12 : (Width<=480) ? 6 : 3
the same result(blur and less red(same frames))
without MergeChroma(blur(MaxTreshold)) the red is more natural

with the script from Phil(posted in CQMatic thread too):
BlindPP(cpu=4)
Convolution3D(1, 6, 12, 6, 8, 2.8, 0)
LanczosResize(...)
Undot()
TemporalSoften(2,7,7,3,2)
DCTFilter(1,1,1,1,1,1,0.5,0)
AddBorders(...)
searching words to explain:
convolution give some "incandescent" colors in dark scenes!

with this script was better:
Undot()
Asharp(1,4)
BicubicResize(336,224,0,0.6,0,0,720,480) # better image than lanczos
#LanczosResize(336,224,0,0,720,480) # faster than bicubic & asharp
Deen("a3d",3,4,1,4) # light clean
#Deen("a3d",3,6,1,6) # middle clean
DCTFilter(1,1,1,1,1,1,0.5,0)
AddBorders(8,8,8,
give better color and better contrast!
more sharpness in moviments scenes( frame 114189)
and better red(frame 120500)

obs:
this frames was used as examples loading the scripts(pictures)
and the resulting mpegs(pictures) in vdub to compare.
watching the movie you see this differences easily!

more:
using BlindPP(cpu=4) in the last script is slow but encrease the CQ!
without grip(script) don't loose image the top/botton borders
and don't "buble" the center of the image(linearity)!
without MergeChroma the red is better.
without lanczos have less "mosquitos".
without convolution don't "buble" the dark colors in dark scenes.

sorry for bore you my friends but...
any coments please?

the best of it all:
CQMatic was fast and perfect in all predictions with CQ_VBR (352x240)
find the CQ in only 2 cycles everytime !
seamless CQs (round 18,5)and final sizes (801 to 803mb) with all scripts!
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
  #2  
11-21-2003, 12:15 PM
Jellygoose Jellygoose is offline
Free Member
 
Join Date: Jun 2002
Location: Germany
Posts: 1,288
Thanks: 0
Thanked 0 Times in 0 Posts
Hi jorel my friend...

Very interesting testings, I would have replied earlier, but missed your post...
I'm currently trying to cram The Lord Of The Rings 2 Towers SEE on one 99min CD-R also at resolution 352x288...

Some questions regarding your tests. How does Deen compress compared to C3D (your settings)?
How is the overall compression of your script compared to the MA script?

One More Thing... Do you see advantages over CQ using CQ_VBR at that resolution?
__________________
j3llyG0053
Reply With Quote
  #3  
11-21-2003, 12:54 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
hy Jell


in poor english(trying):


1-
Deen("a3d",3,4,1,4)
is very "gentle",don't blur and don't "spot" the image.
i see some "solarisation" in dark scenes(faces,forehead) with C3D
and using mergechroma decrease all the red matizes!
deen like posted give compression without change matizes!

2-
the overall compression is seamless with CQ_VBR
round 1.3 less CQ_VBR (for this long movie)
then using MA but give better contrast and sharpness!
don't know how much compression it can give using CQ
with another resolutions(more than vcd resolution)!
maybe using CQ,the MA can give some more compression.

3-
CQ_VBR is really better in 320*2xx with long movies,
got less blocks comparing the same scenes with CQ!

Reply With Quote
  #4  
11-21-2003, 01:03 PM
Jellygoose Jellygoose is offline
Free Member
 
Join Date: Jun 2002
Location: Germany
Posts: 1,288
Thanks: 0
Thanked 0 Times in 0 Posts
Thanks, I'll try CQ_VBR and Deen on my encode @ 352x288 !
__________________
j3llyG0053
Reply With Quote
  #5  
11-21-2003, 03:42 PM
Jellygoose Jellygoose is offline
Free Member
 
Join Date: Jun 2002
Location: Germany
Posts: 1,288
Thanks: 0
Thanked 0 Times in 0 Posts
Hmmm Jorel, are you positive about CQ_VBR looking better at VCD res. than CQ? To me it seems High Action scenes get far more blocky with CQ_VBR...
__________________
j3llyG0053
Reply With Quote
  #6  
11-21-2003, 04:21 PM
Encoder Master Encoder Master is offline
Free Member
 
Join Date: Mar 2003
Location: (^_^)
Posts: 504
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Hmmm Jorel, are you positive about CQ_VBR looking better at VCD res. than CQ? To me it seems High Action scenes get far more blocky with CQ_VBR...
Normaly You're right. I would use 64min and MAX 1700. And let predict the CQ. I think this would to be an good result.

I'am at the same time to put this on an 800 MB rohling with SVCD size and max bitrate 2200kbps.

I post the results later.

Here some of 168min on 800 MB:
http://samsonite.webspace4free.biz/DerPatriot.html
Enjoy it.
Reply With Quote
  #7  
11-21-2003, 04:34 PM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
EM I hope you didn't misunderstand like me which Version of LOTR II Jellygoos intends to encode!
Its a special extended Edition shipped on 4! DVDs where the movie itself occupates 2 DVD Medias!

So keep that in mind ....


EDIT: I See you linked to samples of "Der Patriot" watch the postings of Jellygoose: Lord of the rings! Diff. Movie diff. settings, diff. results!

BTW: The Hair of Mel G. in the second sample sems to be totally "plained"! Noise-Filter-Party?
Reply With Quote
  #8  
11-21-2003, 04:42 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 Jellygoose
Hmmm Jorel, are you positive about CQ_VBR looking better at VCD res. than CQ? To me it seems High Action scenes get far more blocky with CQ_VBR...
in my tests for 320x240,CQ_VBR with min300 and max2000
was better...remember that i only did this (big) movie (low action)
with this parameters.
the last kvcd 320x240(mpeg1)that i did before this was in december 2002
using avisynth 2.0x and "old" scripts
(temporalsmoother,unfilter,lanczos),
tmpgenc in high quality(slow) and don't know "how" matrix!
was monsters inc!
all the others was with more resolutions and in mpeg2.

but Jell, you're a expert, nothing is better than tests!
post more results and details,your tests are precious!
Reply With Quote
  #9  
11-21-2003, 04:52 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
i forgot....sorry:
harry potter is 4:3 and ~152 minutes(give low CQ_VBR or CQ)
and i got better sharpness than the Encoder Master pictures....
that seems "anamorphic"( Mel G. is "egg-head" ? )

Reply With Quote
  #10  
11-21-2003, 08:40 PM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
@ Jorel

He posted the pics as still 480x576 .. therfore not resized to 4:3 aspect ratio ... thats why

.... well in case of comparing exact encoding outputs I also don't resize them to 4:3 when posting in "encoding testing threads"
Reply With Quote
  #11  
11-22-2003, 01:31 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
Quote:
Originally Posted by incredible
@ Jorel

He posted the pics as still 480x576 .. therfore not resized to 4:3 aspect ratio ... thats why

.... well in case of comparing exact encoding outputs I also don't resize them to 4:3 when posting in "encoding testing threads"
oh yes,i know incredible,they are not 4:3....!
i am trying to say that the pictures are "egg-head"...
the "appearance" of the pictures "seems" alongate,
something is wrong with the resize,the faces are "too tall", don't?!

Reply With Quote
  #12  
11-22-2003, 03:05 AM
Encoder Master Encoder Master is offline
Free Member
 
Join Date: Mar 2003
Location: (^_^)
Posts: 504
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Noise-Filter-Party?
@inc.

Can't be because I've use the Optimal Script of Kwag.
And the DVD hasn't a good Quality I think but it's an good result.
Later I post from Harry Potter.
Reply With Quote
  #13  
11-22-2003, 06:56 AM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
Quote:
Originally Posted by jorel
i am trying to say that the pictures are "egg-head"...
the "appearance" of the pictures "seems" alongate,
something is wrong with the resize,the faces are "too tall", don't?!

I copied the image into Photoshop reiszed it to full screen width and to me it seems ok?
Reply With Quote
  #14  
11-22-2003, 08:10 AM
Encoder Master Encoder Master is offline
Free Member
 
Join Date: Mar 2003
Location: (^_^)
Posts: 504
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
I copied the image into Photoshop reiszed it to full screen width and to me it seems ok?
Like ever. Inc. was right.
Reply With Quote
  #15  
11-22-2003, 08:12 AM
black prince black prince is offline
Free Member
 
Join Date: Jul 2002
Posts: 1,224
Thanks: 0
Thanked 0 Times in 0 Posts
@Hi Kwag and Everyone,

I tried a backup with the latest CQMatic X3 and Optimal script on "Alien"
a favorite sci-fi oldie of mine. The results were faster encode, more
accurate file size and great looking picture for 352x240 on 1 CD 74
minutes. There is this new problem. The picture has a thin white line on
the right side that appears every 24 or so frames throughout the whole
movie. On standard TV you can't see it due overscan, but on laptop
using wmp it's very noticable. I tried the same movie with the older
MA script and there was no pulsating line to the right of the screen.
"Alien" is 116 minutes, widescreen, and below is my avs script.
Any suggestions about this problem would be appreciated

LoadPlugin("G:\DVD BACKUP\1 - DVD2KVCD\3 - MOVIESTACKER\Filters25\MPEG2Dec3.dll")
LoadPlugin("G:\DVD BACKUP\1 - DVD2KVCD\3 - MOVIESTACKER\Filters25\GripFit_YV12.dll")
LoadPlugin("G:\DVD BACKUP\1 - DVD2KVCD\3 - MOVIESTACKER\Filters25\asharp.dll")
LoadPlugin("G:\DVD BACKUP\1 - DVD2KVCD\3 - MOVIESTACKER\Filters25\UnDot.dll")
LoadPlugin("G:\DVD BACKUP\1 - DVD2KVCD\3 - MOVIESTACKER\Filters25\UnFilter.dll")
LoadPlugin("G:\DVD BACKUP\1 - DVD2KVCD\3 - MOVIESTACKER\Filters25\STMedianFilter.dll")

Mpeg2Source("F:\Temp\movie.d2v")

## Main section and static filters ###
#
undot()
Limiter()
asharp(1, 4)
GripCrop(352,240,overscan=1,source_anamorphic=true ,dest_anamorphic=true)
GripSize(resizer="BicubicResize")
STMedianFilter(3, 3, 1, 1 )
MergeChroma(blur(1.5))
MergeLuma(blur(0.1))
#
#

## Linear Motion Adaptive Filtering ##
#
# ( Portions from AviSynth's manual )
# This will apply variable temporalsoften
# and variable blur.
# Both filters are active at all times, and work inversely proportional to the
# activity, measured from current frame to next frame.

ScriptClip(" nf = YDifferenceToNext()" +chr(13)+ "unfilter( -(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100)) ).TemporalSoften( fmin( round(2/nf), 6), round(1/nf) , round(3/nf) , 1, 1) ")

#
#
#

GripBorders()
#LetterBox( Your_Values_Here ) # Depends on situation. Use MovieStacker!
Limiter()
ConvertToRGB24()
#
#
## Functions ###

function fmin( int f1, int f2) {
return ( f1<f2 ) ? f1 : f2
}

#
####


-BP
Reply With Quote
  #16  
11-22-2003, 08:22 AM
Boulder Boulder is offline
Free Member
 
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
I think that's aSharp. I've seen that behaviour occasionally with the filter.
Reply With Quote
  #17  
11-22-2003, 09:55 AM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
or maybe gripFit! As I had exactly the same problem yesterday

Try to do one scripting using conventional resizing


Maybe that helps you too ... just try


@ EncoderMaster
I'm definitely not right everytime *lol, ... like seen in my tests in the "New MA script Vers." where I tried something -- but on the other hand nice try and error
Reply With Quote
  #18  
11-22-2003, 10:06 AM
Encoder Master Encoder Master is offline
Free Member
 
Join Date: Mar 2003
Location: (^_^)
Posts: 504
Thanks: 0
Thanked 0 Times in 0 Posts
@black prince

What's you TMPGEnc Settings?
Reply With Quote
  #19  
11-22-2003, 01:02 PM
Abond Abond is offline
Free Member
 
Join Date: Mar 2003
Posts: 243
Thanks: 0
Thanked 0 Times in 0 Posts
@black prince
The same here.
As a workaround I put in TMPG Clip Frame to mask left-right with 8 pixels - no more pulsating lane - it is overlayed
Reply With Quote
  #20  
11-22-2003, 03:04 PM
black prince black prince is offline
Free Member
 
Join Date: Jul 2002
Posts: 1,224
Thanks: 0
Thanked 0 Times in 0 Posts
@Thanks everyone,

I decided to try Abond's advice and use Tmpgenc's Clip Frame to mask
this area out. I hope someone will find a solution in the meantime

-BP
Reply With Quote
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
Avisynth: Interesting results with YlevelsS supermule Avisynth Scripting 2 08-06-2006 11:59 PM
Avisynth versions and results jorel Avisynth Scripting 22 07-05-2004 11:40 AM
Avisynth: Good results with Vague Denoiser? Jellygoose Avisynth Scripting 5 10-06-2003 03:55 AM
Avisynth: Good filters and plugin for Avisynth 1.05 ? GORI Avisynth Scripting 2 02-10-2003 05:27 PM
KVCD: Good results with 90-minute CD-Rs! Pnumonic Video Encoding and Conversion 1 05-14-2002 06:30 PM

Thread Tools



 
All times are GMT -5. The time now is 03:47 AM  —  vBulletin Jelsoft Enterprises Ltd