Hey Kwag,
I remember not long ago I asked the same thing about the "detect scene change" feature and you also told me to go ahead and use it.
I just forgot to tell you that I had also seen it in the download/usage intructions.
Shame on me :oops: ...That's why once in a while some guys still ask this one.

Cheer.

Calcumatic 1.0.6
 

I've updated CalcuMatic to version 1.0.6.
The only change is that audio now defaults to 128Kbps, instead of 112Kbps.
So I suggest always predicting with audio set to 128Kbps, and if the final MPEG file size is over or under the wanted size, you can encode the audio at either 112Kbps or 160Kbps, whichever is needed to compensate for the video size difference.

-kwag

If I end up with a filesize witch are ex. 50 Mb to big, are there any way to correct the filesize a litle bit by doing an other encoding with a different avg. bitrate?? Which formula should I use :?:

Hi azel,

The best way, as described above, is do your prediction with 128Kbps audio.
Then if your file is larger, as in your case, re-encode the audio at 112Kbps and you might also want to cut off your end credits when your doing your final muxing ( with BBMpeg ).
This way, you'll get a file size WAY lower than the 50MB in excess you had.
CQMatic does a very good job on most movies, but it can fail miserably on some films, depending on the distributed action of the movie. The only way to increase precision would be to increase the sampling, but at the cost of longer prediction which would then void the original purpose of CQMatic.

-kwag

I'd take acurate prediction over time anyday :D

Ok, then maybe I will put a "Long Prediction" option in CQMatic, so that it takes about 5 times more sampling then the "standard" prediction. This way, it should be far more accurate, but it will also take 5X times to find the correct CQ.
I'll look into that :) when I have some time 8O

-kwag

Not needed ... as the last days Krassi and I figured that a prediction using same sample size but with offset intervals set at every prediction turn will give a very acurate result. Even without pre-2pass run.

Hi incredible,

I'm not so sure about your theory, or maybe I just don't follow it.
How can you calculate a final size without prediction :?:
If you are using a fixed sample size, based on the size of the movie, it will not work :!:
For example: A one minute clip at a fixed CQ value, taken in the center of the movie, will have a different encoded file size if you encode another one minute clip from another part of the movie.
So you can't take an arbitrary sample based on the "run time" of a movie, calculate the supposed sample size based on movie time, and expect the complete encode to come out on target. I hope you understand what I meant :roll:
It's similar to what I tried months ago, taking the full VOBs ( demuxed video streams size ) of a movie, divided by the wanted size given by MovieStacker ( gives you the ratio ) and then cut an arbitrary part of the VOB and divide it by the ratio, and encode a KVCD sample matching that size. It just doesn't work :!:

-kwag

New version
 

CQMatic 1.2.03 is currently cooking in the oven ;)
I'm currently testing it. Will release it later today.
New function: Prediction X1 or Prediction X3 (selectable)
Defaults to X1, which is just as the current version of CQMatic.
Prediction X3, as it says, takes X3 (times 3) longer snapshots ( 15 seconds per sample, instead of 5 seconds as it's currently implemented)
This should yield much higher accuracy, but will also take three times longer to encode each sample :!:
End result: Higher precision and quality, and still faster than a 2-pass VBR.

-kwag

Hi, Kwag.

I assume I'll still need to allow 25MB or so for SVCD.

For muxing :?:
I assume you mean subtracting 25MB to allow for SVCD overhead, right :?:
If that's the case, probably yes, because with the new CQMatic, the final file size should be much more closer to the calculated by CalcuMatic.
I'm currently doing a final prediction in both modes (X1, X3) to verify the final CQ differences.

-kwag

Can you make a "autoclose" version of CQMatic 1.2.03????

Sure :!: :D
Let me open the oven, and I'll cook it in a couple of minutes :cool:

-kwag

Here you go ;)
http://www.kvcd.net/CQMatic-1.2.03_AutoClose.exe

-kwag

Thanks :D
I was just wondering. Could I include CQMatic into DVD2KVCD, like the AVS-filters????

Sure :D
As long as DVD2KVCD is free. Otherwise, we talk ;)

-kwag

Yes, of course :)

could someone explain the buttons to me? how do I know which mode I'm in with each button? so say if the button said Prediction X1 does that mean its in that prediction mode or if u press the button it will be in X1 prediction mode?

What is displayed, is the current encoding mode.
Default is "X1". If you press it, it changes to "X3", and that will be the mode if you start encoding.

-kwag

how come then when I choose prediction mode it does a full encode too after prediction? I had prediciton mode selected but it still seemed to 1. do a full encode and 2. produce a large final m1v file which caused a final mpg file size of nearly 900mb after muxin with 112 audio even when I used 128 audio in predicition aswell as 112 in calcumatic.