Forums [Archives] Forums [Archives] (
-   Video Encoding and Conversion (
-   -   What about a matrix varying based encoder? (

fabrice 03-29-2005 02:14 PM

What about a matrix varying based encoder?

As varying some value in a matrix can cause blurring, would it be posible to make an encoder which vary the matrix acording to vectors, using a blurrier version with high motion vectors, and using a normal version in less motion zone?
Or the mpeg2 specification don't allow to vary the matrix within a mpeg file?


Dialhot 03-29-2005 03:56 PM

This encoder has already a name : CCE 2.70 (adaptative Q-Matrix feature that must be unselected to do KDVD by the way).

fabrice 03-29-2005 09:35 PM


So someone already think about that before... I came too late! :-)
What about the quality? And why do you have to uncheck this to do KDVD?

I'll download the trial version to see it , and if it's a interesting feature, ask Hank if it will be posible to modify HC to do that, ... ( or have, at least, a proof of concept version).


kwag 03-29-2005 09:46 PM

I think that a dynamic matrix would be a geat idea, but I'm not sure it would work on all standalone players.
The reason is, that probably many players read the first header of the MPEG file, and set the decoder accordingly.
If a standalone decoder uses that first matrix as a reference for the complete movie, and the matrix in the movie varies on every header segment, the result will be a decoded mess :!:


Dialhot 03-29-2005 09:47 PM


Originally Posted by fabrice
What about the quality?

Don't know, never used.


And why do you have to uncheck this to do KDVD?
Because we want the encoder to use the KVCD-Notch and nothing else :)

fabrice 03-29-2005 10:17 PM


According to CCE user guide:


What is adaptive Q-matrix switching ?
There are some cases that Cinema Craft Encoder SP cannot keep
the minimum bitrate. It is considered to be caused by that the lowest
quantization scale cannot raise the bitrate under a single quantization
matrix. To solve the problem, Cinema Craft Encoder SP automatically
creates other matrices cutting the value of each element in the
matrices by half, quarter and so on, and applies another one to produce
a much lower distortion for much higher bitrate. This switching
of quantization matrix can occur at every GOP.
Thus, adaptive Q-matrix switching can improve the quality of image,
especially with lower bitrate sources or CG works or the edge of
fading out.
So it's not really what I was talking about. but if CCE give the posibility to change the matrix each GOP, it means that it's mpeg compliant. (because it's activated by default).
Is there a way to get a file, with multiple matrix, to do a DVD test? (without encoding each GOP in a different file, with a different matrix, of course).

The only problem I see, is that if the matrix can only be put at gop level, we will have the same problem with moving object in front of a static area we had with optimal script: loosing all static area details...


kwag 03-29-2005 10:35 PM


Originally Posted by fabrice
Is there a way to get a file, with multiple matrix, to do a DVD test? (without encoding each GOP in a different file, with a different matrix, of course).

I think that's exactly the easiest way to try it, at this moment :)
Just encode four or five 1 minute clips, each with a different matrix, and then join the MPEG files together as a single large file.
Then see if there are hickups on a standalone. :D


fabrice 03-30-2005 12:32 AM


I'm not sure it would do the trick, because it don't seem that the matriz is always present in the Sequence header ( ), so merging the files only would give me a first matrix, but no sequence matrixes...
Is there any tool I can use to confirm that? (o should I write my own mpeg header parser?)


EDITED: I found Mpeg parser, and I confirm that the "load Intra quantiser Matrix" flag is on with hc and cce. The only problem is that I can't see the matrix used in each Sequence Header, just to be sure that merging the 5 files won't erase this 'local' matrix.

All times are GMT -5. The time now is 01:55 AM  —  vBulletin Jelsoft Enterprises Ltd

Site design, images and content © 2002-2022 The Digital FAQ,
Forum Software by vBulletin · Copyright © 2022 Jelsoft Enterprises Ltd.