Quantcast Do Divx Settings Affect MPEG Output? - digitalFAQ.com Forums [Archives]
  #1  
09-11-2003, 09:44 AM
zes zes is offline
Free Member
 
Join Date: May 2003
Location: Antwerp, Belgium
Posts: 43
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to zes
should i leave post-processing options for divx disabled or has it no effect on the final mpeg output?
i ve read somewhere that the latest divx scripts 'spray' noise to reduce blocks. so what if i enable deblocking and deringing (divx 5.1) and then 'spray' even more noise for mpeg?
just a thought
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  
09-11-2003, 11:41 AM
Dialhot Dialhot is offline
Free Member
 
Join Date: May 2003
Posts: 10,463
Thanks: 0
Thanked 0 Times in 0 Posts
Let every postprocessing off. Scripts are designed to deal with all artifacts, and if you removed them with the codec feature, the filters used in the scripts will be too heavy. Your picture will probably become too blurry.

Deringing for instance blurs the picture a lot !
Reply With Quote
  #3  
09-11-2003, 12:29 PM
zes zes is offline
Free Member
 
Join Date: May 2003
Location: Antwerp, Belgium
Posts: 43
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to zes
so it does have an effect
explains why my mpegs were kinda blurry
thanks!
Reply With Quote
  #4  
09-12-2003, 08:50 AM
J-Wo J-Wo is offline
Free Member
 
Join Date: Nov 2002
Location: Toronto, Canada
Posts: 454
Thanks: 0
Thanked 0 Times in 0 Posts
where does one change these divx settings?
Reply With Quote
  #5  
09-12-2003, 10:34 AM
vmesquita vmesquita is offline
Invalid Email / Banned / Spammer
 
Join Date: May 2003
Posts: 3,726
Thanks: 0
Thanked 0 Times in 0 Posts
I think post-processing settings are not used unless you use DirectShowSource instead of AVISource (but DirectShowSource gives me problems). Right now I am doing some tests exchanging undot for BilndPP() on the version 2 of the script and getting nice results. But I haven't tried version 1 script yet. BlindPP is a postprocessing filter included in MPEG2DEC package.

[]'s
VMesquita
Reply With Quote
  #6  
09-12-2003, 10:36 AM
Dialhot Dialhot is offline
Free Member
 
Join Date: May 2003
Posts: 10,463
Thanks: 0
Thanked 0 Times in 0 Posts
vmesquita we are talking about the postprocessing done by the codec itself ! So they are used when you decode the video with avisource.
And the proof is that CQ found by CQMatic raise when you set the post process settings on.
Reply With Quote
  #7  
09-13-2003, 06:55 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
And encoding gets a lot slower too. Try encoding a small DivX clip with zero postprocessing and then at full PP in the codec settings. You'll see the difference
Reply With Quote
  #8  
09-13-2003, 01:55 PM
vmesquita vmesquita is offline
Invalid Email / Banned / Spammer
 
Join Date: May 2003
Posts: 3,726
Thanks: 0
Thanked 0 Times in 0 Posts
Dialhot and boulder,

This is very interesting! I'll do some tests about it, I always thought post-processing was not used when decoding but I never tested... So maybe I can replace undot by leaving Post-Processing on.

[]'s
VMesquita
Reply With Quote
  #9  
09-13-2003, 02:08 PM
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 wouldn't do that. Post-processing in the decompression process means removing blocks whereas UnDot deals with general noise. Besides, UnDot doesn't slow down the encoding process nearly as much as the PP settings.
Reply With Quote
  #10  
09-13-2003, 02:19 PM
Dialhot Dialhot is offline
Free Member
 
Join Date: May 2003
Posts: 10,463
Thanks: 0
Thanked 0 Times in 0 Posts
What is amusing is that XVID decoder has the deblocking/dering features since a long time a noone ever ask a question about it.

The same : the "film effect" of Divx decoder that is present since version 5 is the equivalent of Blockbuster(noise), and again noone ever thought about this.

But that is a good thing : do not rely on anything that is in the codec else you will need a script for each CODEC TYPE avi : xvid, divx 5.05, divx 3.11, divx 4.01...

Note : Didn't you ever notice that mpeg2dec3 has a deblocking/deringing feature also ?
Reply With Quote
  #11  
09-13-2003, 02:38 PM
Boulder Boulder is offline
Free Member
 
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by Dialhot
Note : Didn't you ever notice that mpeg2dec3 has a deblocking/deringing feature also ?
I actually use it sometimes, deblocking luma (cpu=2,moderate_h=30,moderate_v=50). I've used it with some very poor quality DVDs and had good results. I think that those moderate_h and v values were suggested a long time ago when MarcFD added Nic's deblocking features to MPEG2DEC3. They don't soften much and get rid of the worst visible blocks on the DVD.
Reply With Quote
  #12  
09-13-2003, 10:57 PM
vmesquita vmesquita is offline
Invalid Email / Banned / Spammer
 
Join Date: May 2003
Posts: 3,726
Thanks: 0
Thanked 0 Times in 0 Posts
@Dialhot
DIvX "film effect" is not the same thing as blockbuster. It adds noise to the hole picture (not only the DCT areas), making compression very hard (I tried with DirectShowsource).
About the codec problem: I agree, but most people is doing only DivX AND XVid and it DivX 5 can handle DivX3 content. It looks like a good way to improve the source. BlindPP() produces good results but emulates a quantisizer
"BlindPP(clip, int "quant", int "cpu", str "cpu2", bool "iPP", int "moderate_h", int "moderate_v")

To Deblock and Dering on any kind of DCT-encoded source.
Of course, less accurate than decoder intergrated PP, but still very efficient
need YV12 input."

What means that integrated PP would be even better. Real PP works by knowing the quantisizer and softening the blocks and rings. Maybe PP+blockbuster noise would be a great team...

@boulder
Good point about undot.
Reply With Quote
  #13  
09-13-2003, 11:37 PM
vmesquita vmesquita is offline
Invalid Email / Banned / Spammer
 
Join Date: May 2003
Posts: 3,726
Thanks: 0
Thanked 0 Times in 0 Posts
update:
Just did a 1000 frames xvid clip with and without post-processing using avisource and got exactly 8.125.412 bytes in both... Wired... I encoded in CCE Q40.

[]'s
VMesquita
Reply With Quote
  #14  
09-14-2003, 03:05 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
Quote:
Originally Posted by vmesquita
update:
Just did a 1000 frames xvid clip with and without post-processing using avisource and got exactly 8.125.412 bytes in both... Wired... I encoded in CCE Q40.

[]'s
VMesquita
There's two different places where you can set the postprocessing. One is in the codec settings in the screen you use when you're encoding something to XviD, the other's available when you play XviD material in some media player. I'm not sure which one is used with AVISource.
Reply With Quote
  #15  
09-14-2003, 05:50 AM
Dialhot Dialhot is offline
Free Member
 
Join Date: May 2003
Posts: 10,463
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by vmesquita
@Dialhot
DIvX "film effect" is not the same thing as blockbuster. It adds noise to the hole picture (not only the DCT areas),
But what do you think Blockbuster does ? Looking for DCT area before spaying its noise ? Of course not. The purpose of blockbuster is to help reducing the creation of DCT : they aren't supposed to be there when blockbuster starts its work !
Quote:
making compression very hard (I tried with DirectShowsource).
The problem can be here. Or simply the ramdomizer algorythm used in blockbuster is different from the one used in the codec. But the goal of both effects is the same.

Quote:
About the codec problem: I agree, but most people is doing only DivX AND XVid and it DivX 5 can handle DivX3 content.
You know that, I know that, a lot of people don't. They install codec via pack (nimda, ace...) that contain and install Divx 3.11, Divx 4.01 and Divx5.05 together (I really don't know why). But we can for sure give them hints about this.
Quote:
It looks like a good way to improve the source.
That's a way I tried weeks ago and decided to drop. Too blurry, whatever you do. The solution would be to have a control on the strength of each filter the codec applies. But we don't. Tell me if you find a way.
Quote:
BlindPP() produces good results but emulates a quantisizer
"BlindPP(clip, int "quant", int "cpu", str "cpu2", bool "iPP", int "moderate_h", int "moderate_v")
I will look at this.
Reply With Quote
  #16  
09-14-2003, 06:30 AM
zes zes is offline
Free Member
 
Join Date: May 2003
Location: Antwerp, Belgium
Posts: 43
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to zes
i'm thinking off letting the codec's PP do it's job and be done with it. no filters in avisynth. gonna do a few encodes, see what it's worth.
Reply With Quote
  #17  
09-14-2003, 07:36 AM
zes zes is offline
Free Member
 
Join Date: May 2003
Location: Antwerp, Belgium
Posts: 43
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to zes
i did 2 quick encodes to see if there's a clear difference .
first one: divx max deblocking, no deringing, and a little bit film effect, no script
second : divx PP disabled, and dialhot script.

ok i must admit i was doubting this whole scripting bit for a sec, but i'm back on track now though they both had practically the same bitrate, dialhot's script does a nicer job on the source than divx' own PP options. more artifacts were seen with divx PP, dialhot's was crisper.
but for the deblocking part: both were blocky as hell (used a low quality divx) in the same parts.

seems that there's no real solution to this yet
Reply With Quote
  #18  
09-14-2003, 08:47 AM
Dialhot Dialhot is offline
Free Member
 
Join Date: May 2003
Posts: 10,463
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by zes
but for the deblocking part: both were blocky as hell (used a low quality divx) in the same parts.

seems that there's no real solution to this yet
Did you use the script we both blockbuster lines or only the fist one ? Because I removed the second line for my own encoding because I found one was enought, but perhaps on your very bad divx, you should try to put it again.
Reply With Quote
  #19  
09-15-2003, 01:22 AM
zes zes is offline
Free Member
 
Join Date: May 2003
Location: Antwerp, Belgium
Posts: 43
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to zes
i've been using this script for a while now

Code:
#Post-Process 

Blockbuster(method="noise",detail_min=1,detail_max=8,variance=0.3,seed=5823) 
Deen() 

GripCrop(528, 576, overscan=2, 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=0.5,seed=5623) 

GripBorders()
i see you've been working on a new script, so i'm gonna give that a try now
Reply With Quote
  #20  
09-15-2003, 03:54 PM
vmesquita vmesquita is offline
Invalid Email / Banned / Spammer
 
Join Date: May 2003
Posts: 3,726
Thanks: 0
Thanked 0 Times in 0 Posts
@Dialhot

How do I enable post-processing during decoding in XVid? I was trying to do that changing XVid settings during playback, but I guess I was only changing DirectShow Decoder properties... Maybe changing in other place I can get it to work. I really would like to compare this to blindpp()...

[]'s
VMesquita
Reply With Quote
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
TMPGEnc: Recommended mpeg settings scorch Video Encoding and Conversion 1 06-21-2007 01:35 PM
TMPGEnc: CQ and CQ_VBR Mpeg Settings mjclark Video Encoding and Conversion 1 10-26-2005 09:54 PM
TMPGEnc: way to specify number of discs to output the mpeg files? ezysk Video Encoding and Conversion 1 05-17-2003 04:09 PM
BBMpeg: Larger output mpeg? Wolfi Video Encoding and Conversion 4 04-28-2003 11:32 AM
final mpeg output has tiny black bar on top and bottom? syk2c11 Video Encoding and Conversion 4 09-07-2002 07:20 PM

Thread Tools



 
All times are GMT -5. The time now is 03:09 AM  —  vBulletin © Jelsoft Enterprises Ltd