Quantcast KVCD: Superbit Success! - digitalFAQ.com Forums [Archives]
  #1  
11-24-2003, 07:34 PM
audioslave audioslave is offline
Free Member
 
Join Date: Mar 2003
Location: Sweden
Posts: 725
Thanks: 0
Thanked 0 Times in 0 Posts
I've experimented with encoding the Superbit version of Bram Stoker's Dracula, and the results are simply awesome! It's the NTSC version of the movie (don't know if they do Superbit for PAL...?) and with CQMatic and a slightly modified MA script I got a CQ of 63,21. The audio is calculated as 128 kbps dual channel (of course! ). That's not bad at all since the movie is 127 minutes long. The script I'm using is this:
Quote:
### Ver. November 21, 2003 @07:45 GMT ###



## Main section and static filters ###
#
Mpeg2Source("D:\Rippat\Dracula\Dracula.d2v")
#
UnDot()
Limiter()
ASharp(1, 4)
BicubicResize(448, 352, 0, 0.6, 16, 0, 688, 480)
STMedianFilter(3, 3, 1, 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)+"UnFil ter(-(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100)) ).TemporalSoften(fmin(round(2/nf), 6), round(1/nf), round(3/nf), 1, 2)")

#
#
#

AddBorders(16, 64, 16, 64)
Limiter()

#
#
## Functions ###

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

#
####
The things I changed in the script:
1) Removed the MergeChroma/Luma lines.
2) Changed the last parameter for TemporalSoften to 2 (instead of 1).

I will try to post a sample of the movie (if I can find any webspace to upload to ) so you can see and judge for yourselves.

Don't know if this is of any interest to anyone, but hey...


@kwag
About my TempSoft parameter change... I don't kow if I remember correctly but I think I saw a post were you said it was a good idea to do this change? Any other changes perhaps...?
__________________
AudioSlave
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  
11-24-2003, 08:48 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 audioslave
@kwag
About my TempSoft parameter change... I don't kow if I remember correctly but I think I saw a post were you said it was a good idea to do this change? Any other changes perhaps...?
I don't think any other change is need ( for the moment )
I would leave the MergeChroma(blur(1.5)) line, and drop the MergeLuma(blur(0.1))
I have to update the script to reflect those changes.

-kwag
Reply With Quote
  #3  
11-24-2003, 08:59 PM
audioslave audioslave is offline
Free Member
 
Join Date: Mar 2003
Location: Sweden
Posts: 725
Thanks: 0
Thanked 0 Times in 0 Posts
Yeah, I usually use the MergeChroma line but this movie have a quite large amount of red in many scenes, and the movie looked better without the line. That's the only reason I removed it.

EDIT: Sorry... I made a new comparison with and without MergeChroma(blur(1.5)) and I can't see any difference between the two. Perhaps my lack of sleep have something to do with my previous statement... So I'm restarting my encode to gain those precious CQ units!
__________________
AudioSlave
Reply With Quote
  #4  
11-24-2003, 10:21 PM
audioslave audioslave is offline
Free Member
 
Join Date: Mar 2003
Location: Sweden
Posts: 725
Thanks: 0
Thanked 0 Times in 0 Posts
Huh?!
This is strange - with MergeChroma(blur(1.5)) I only gain 0.19 in CQ So the question is: How much compression is gained with MergeChroma and has it become obsolete (like MergeLuma)? Since the CQ is almost identical I think I'll skip MergeChroma. It should be faster to encode without it too, I guess... As they say - less is more.

It would be nice to get feedback from you all on this MergeChroma discovery. I don't know, but it seems to me it's no longer needed?
__________________
AudioSlave
Reply With Quote
  #5  
11-24-2003, 10:33 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
Hi audioslave,

I think that with the last changes I made to STMedian ( STMedianFilter(3, 3, 1, 1 ) ), the "Merge" lines are not really needed.
Also, with the current MA script which is now "true linear", it makes quite a big difference on final file size

-kwag
Reply With Quote
  #6  
11-24-2003, 10:37 PM
audioslave audioslave is offline
Free Member
 
Join Date: Mar 2003
Location: Sweden
Posts: 725
Thanks: 0
Thanked 0 Times in 0 Posts
Okay, kwag. Then it's goodbye and good riddance to MergeWhatever. At least for now. We never know what the future brings, eh? Thanks for clearing out my questions - once again!

Good night... ZzZzZz
__________________
AudioSlave
Reply With Quote
  #7  
11-25-2003, 08:16 PM
audioslave audioslave is offline
Free Member
 
Join Date: Mar 2003
Location: Sweden
Posts: 725
Thanks: 0
Thanked 0 Times in 0 Posts
I got the wierdest idea this morning. To try to encode this movie - Bram Stoker's Dracula (Superbit) - with a static script. So after some experimenting I came up with a script that not only produced better output than the current MA script, but also gave me higher CQ
With the current MA script I got CQ63,21 and with my static script I got CQ64,43. Strange but true.
I used 300 for min. bitrate and 2500 for max, and Motion search precision set to High quality. The audio was 128 kbps, dual channel (downmixed to 44.1 kHz), and I used a final resolution of 480x480 (SVCD). I used ToK to predict. The script I came up with is this:
Quote:
## Static script by AudioSlave - November 26, 2003 ##

Mpeg2Source("D:\Rippat\Dracula\Dracula.d2v")
LanczosResize(448, 352, 16, 0, 688, 480)
Undot()
MergeChroma(blur(0.5))
MergeLuma(blur(0.1))
BlockBuster(method="noise", detail_min=1, detail_max=3, variance=0.1, seed=1)
DCTFilter(1, 1, 1, 1, 1, 1, 0.5, 0)
AddBorders(16, 64, 16, 64)
By using UnDot() right after the LanczosResize almost all gibbs were eliminated. Please, try it out. I don't think you will get disappointed!

EDIT: Damn, now I see I only used 0.5 for MergeChroma. Was supposed to be 1.5. Well, I totally missed that. Guess I'm gonna have to re-encode the movie - AGAIN! My, oh, my...

EDIT #2: I made a new prediction with the intended MergeChroma(blur(1.5)) and got CQ65,13. Another thing, when I encoded with the MA script I posted above the video got too large so I had to use 112 kbps audio. That means the CQ difference between the scripts are even higher than I first thought. Even more CQ with my latest script compared to the latest MA. I don't know if the script differences are the same with other movies since I've only encoded this one with my script - so far...
__________________
AudioSlave
Reply With Quote
  #8  
11-26-2003, 04:42 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 audioslave
I got the wierdest idea this morning. To try to encode this movie - Bram Stoker's Dracula (Superbit) - with a static script. So after some experimenting I came up with a script that not only produced better output than the current MA script, but also gave me higher CQ
I also do all my DVD with a variation on the optimal script I developped for avi->kvcd (as it seems you did, am I wrong ?). I did yesterday "The professional" in 480*576 MPEG1 / audio 128kbp/s and the CQ is... 78.82 ! (movie length 1h42)

I have to test the last version of the MA script on my next DVD as I dropped the former one because I didn't like the blur effect on the action scenes. The new script is perhaps better on that point.
Reply With Quote
  #9  
11-26-2003, 05:10 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 audioslave
With the current MA script I got CQ63,21 and with my static script I got CQ64,43. Strange but true.
I used 300 for min. bitrate and 2500 for max, and Motion search precision set to High quality. The audio was 128 kbps, dual channel (downmixed to 44.1 kHz), and I used a final resolution of 480x480 (SVCD). I used ToK to predict. The script I came up with is this:
Quote:
## Static script by AudioSlave - November 26, 2003 ##

Mpeg2Source("D:\Rippat\Dracula\Dracula.d2v")
LanczosResize(448, 352, 16, 0, 688, 480)
Undot()
MergeChroma(blur(0.5))
MergeLuma(blur(0.1))
BlockBuster(method="noise", detail_min=1, detail_max=3, variance=0.1, seed=1)
DCTFilter(1, 1, 1, 1, 1, 1, 0.5, 0)
AddBorders(16, 64, 16, 64)
The ... DCTFilter(1, 1, 1, 1, 1, 1, 0.5, 0) ...-Line should also be used at the end of MA in case of gaining the CQ. This could also be the reason why you obtain more CQ. Depending on the Movie and its "frequencies" it behaves every time different according to the resulting CQ advantage.
Quote:
By using UnDot() right after the LanczosResize almost all gibbs were eliminated. Please, try it out. I don't think you will get disappointed!
Another Point ... I also always use Undot() AFTER resizing so the filter has to perform on less size which does it make more effective.
Reply With Quote
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
KVCD: Thank you - success! zerotype Video Encoding and Conversion 3 11-17-2004 12:08 AM
Success! Thank you KVCD! gijluiaard Video Encoding and Conversion 4 09-04-2004 06:16 AM
Success with VCD to KVCD! J-Wo Video Encoding and Conversion 14 10-20-2003 06:45 PM
KVCD: Success! telemike Video Encoding and Conversion 0 02-26-2003 06:57 AM
Kvcd: Success! Paul0889 Video Encoding and Conversion 2 12-11-2002 05:48 PM

Thread Tools



 
All times are GMT -5. The time now is 08:06 AM  —  vBulletin © Jelsoft Enterprises Ltd