Am I Allowed to Post KVCD Optimal Scripts Here?

Edlund · #1 09-30-2003, 01:58 AM

Because I myself am using a script who produces less filesize and the same quality than the current "optimal script v3" and I think you could find it useful.

bigggt · #2 09-30-2003, 05:27 PM

Hi Edlund, i 'm nobody important here but i say just post your script here and if people like it i'm sure it can go to the optimal script page

I say the more people trying to get the best script,the better

vmesquita · #3 09-30-2003, 07:37 PM

Yes, Edlung, please post your script and let everyone test! We're all after the best picture possible after all.

[]'s
VMesquita

Edlund · #4 10-01-2003, 03:19 AM

Quote:

AviSource("---.avi",false)
GripCrop(---,---, overscan=1, source_anamorphic=false)
Blockbuster(method="noise",detail_min=1,detail_max =3,variance=0.1,seed=1)
GripSize(resizer="BicubicResize")
Undot()
Limiter()
ASharp(1, 4)

SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 2
ScriptClip("nf = YDifferenceToNext()"+chr(13)+ "nf >= SwitchThreshold ? \
Unfilter(-(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100))) : \
TemporalCleaner(round(3+nf), round(8+nf))")

DCTFilter(1, 1, 1, 1, 1, 1, 0.5, 0)

GripBorders()
Limiter()

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

When I use this script with "Motion estimate search" the encoding is much faster than the current optimal script version 3 and the filesize is smaller, but with very little difference. If I encode with "High quality (slow)", the encoding time is a little bit slower and the the difference in the filesize is bigger, but still not so much. I must admit that the current version 3 is good, but you can use my script for way faster encoding with the same quality.

Edlund · #5 10-01-2003, 11:05 AM

Here are the results that I got by encoding a music videoclip today :
movie time - 4:30 minutes
audio - encoding with no audio
Template - KVCDx3-mpg1-pal
resolution - 352x288
CQ = 70

using version 3 with high quality(slow) - encoding time 13:13 minutes, 49.8MB
using Edlund otimal script version 1 (

) with high quality - encoding time 12:08 , 48.1MB
using version 3 motion estimate search - encoding time 8:05, 51.9MB
using my script with motion estimate search - encoding time 7:56, 50.4MB

Dialhot · #6 10-04-2003, 03:29 PM

Quote:

Originally Posted by Edlund

Because I myself am using a script who produces less filesize and the same quality than the current "optimal script v3" and I think you could find it useful.

Others already answered but I want to clarify things : I locked the "optimal script for avi" thread just to keep a small, clear and easy to read page containing only the script. Like Kwag did in the "optimal script" section of the forum.

But of course everyone can give suggestion ! The optimal script suggested is the "latest" script, not the "last" script !

Dialhot · #7 10-04-2003, 03:33 PM

Quote:

Originally Posted by Edlund

Code:

AviSource("---.avi",false) 
GripCrop(---,---, overscan=1, source_anamorphic=false) 
Blockbuster(method="noise",detail_min=1,detail_max=3,variance=0.1,seed=1) 
GripSize(resizer="BicubicResize") 
Undot() 
Limiter() 
ASharp(1, 4) 

SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 2 
ScriptClip("nf = YDifferenceToNext()"+chr(13)+ "nf >= SwitchThreshold ? \ 
Unfilter(-(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100))) : \ 
TemporalCleaner(round(3+nf), round(8+nf))") 

DCTFilter(1, 1, 1, 1, 1, 1, 0.5, 0) 

GripBorders() 
Limiter() 

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

In few words : you adapt a little the MA script for using with MPEG4 sources ? Hum.. I have to test this.
But what about the file size prediction accuracy ?

And I'm also surprised this is faster as I remember that MA script on avi sources take a very long time to make a single prediction step. But perhaps that is just for the prediction and after everything run faster ?

Again, I have to test this

Edlund · #8 10-04-2003, 04:17 PM

Yes, the prediction seems really slow to me. But usually I don't use prediction so this is not a problem for me. (I calculate the CQ manually

heiloku · #9 10-04-2003, 04:54 PM

Sorry to answer but, how do you exactly calculate CQ manually?

kwag · #10 10-04-2003, 05:34 PM

Quote:

Originally Posted by heiloku

Sorry to answer but, how do you exactly calculate CQ manually?

Good question

Because you really can't calculate CQ, unless he means manually doing prediction

Anything else, would be a guesstimate

-kwag

incredible · #11 10-04-2003, 06:20 PM

Quote:

Originally Posted by Dialhot

In few words : you adapt a little the MA script for using with MPEG4 sources ? Hum.. I have to test this.

Well is the MA script routine only useful to use with m2v/d2v sources as they come from DVDs??
He only changed the routine in the part where he changed temporalsoften to temporalcleaner. Maybe its useful to insert the Filters you put into Vers.3 into the MA routine.

It makes really sense just to enjoy the advantage of MA Filtering for example at fast motion scenes ... thats why he gets more CQ ....

... We'll see

Edlund · #12 10-05-2003, 04:00 AM

Quote:

Originally Posted by kwag

Quote:

Originally Posted by heiloku

Sorry to answer but, how do you exactly calculate CQ manually?

Good question

Because you really can't calculate CQ, unless he means manually doing prediction

Anything else, would be a guesstimate

-kwag

Yes, guesstimate, a good word to describe my prediction

What I do is see what length the .avi is, how many high motion scenes there are, remember what CQ's I have used before and than set the CQ for this encoding

I also encode 1 or 2 minutes, (preferably with hifh motion scenes), and than I:
1.calculate the movie time in seconds.
2.divide the movie time in seconds by the 2-minutes-sample-time in seconds.
3.multiply the result with the filesize of the 2-minutes-sample.
The final result is the possible filesize of the encoded .m1v
I know that this is not accurate but actually I have more success doing this type of "prediction" instead of using ToK or CQmatic. For me personally this programs can't work precisely with avi files.

Dialhot · #13 10-05-2003, 07:09 AM

Quote:

Originally Posted by Edlund

For me personally this programs can't work precisely with avi files.

Because you use a MA script. CQMatic work perfectly with avi and static scripts (as the optimal one I have suggested).

Last attempt : file size requested 807 Mb, file size reached with CQMatic : 815 Mb !

Edlund · #14 10-05-2003, 07:46 AM

Then I must try with some other script

Edlund · #15 10-06-2003, 01:22 AM

Actually yesterday I put 6 files which I encoded with "my prediction methods" on one cd and the filesize of them together was exactly 800MB

heiloku · #16 10-07-2003, 04:31 PM

Thanks edlund, it seems we both use the same system to predict final size. It only my computer didn't give me error messages when using CQMatic.

Edlund · #17 10-11-2003, 03:00 AM

Ok, has someone tried my script? There are comments that the optimal script version 3 is brilliant but my script is faster and produces smaller filesize (at least for me

I used my script for encoding a rock concert with a whole lot of high motion scenes and the results are great! I encoded it with the v3 script first so I can compare.
If this script is not so good at least someone explain me why.

Siku · #18 10-11-2003, 03:44 AM

@Edlund

I tried your script and it worked pretty good. There is still problem: lack of details! Every DivX-script has this problem, if you want to get rid of the blocks you need to filter the picture and you'll lose details. That's so sad...

Anyway I'm just wondering if there is a workout for this thing. Usually XviD/DivX heavily compresses red and orange areas. It would be great if there were a script/filter for detecting these areas and thus could filter those areas more than 'normal' areas. By the way, there is a script which is able to detect low, medium and high motion scenes and with this script you can apply different filtering for each motion scene. For more information see this. I just though this would be useful when developing these Divx-scripts.

Regards,
Siku

Edlund · #19 10-11-2003, 04:32 AM

Thanks for the link, I'm trying this method right now