[Matroska-devel] Overhead comparison FLAC in MKA, created with jcsston's FLAC encoder filter

Jory jcsston at toughguy.net
Mon Nov 10 17:21:17 CET 2003


----- Original Message -----
From: "Christian HJ Wiesner" <chris at matroska.org>
To: "Discussion about the current and future development of Matroska"
<matroska-devel at lists.matroska.org>; <jcsston at matroska.org>;
<gabest at matroska.org>
Sent: Monday, November 10, 2003 8:50 AM
Subject: [Matroska-devel] Overhead comparison FLAC in MKA, created with
jcsston's FLAC encoder filter


>
> Hi,
>
> here a few overhead comparisons on a MKA file made with Jory's FLAC
> encoder filter and matroskamuxer filter on Graphedit :
>
> Source file :                    682 MB
>
> FLAC ( level 5 ) :           409 MB
> OggFLAC ( level 5 ) :    413 MB
> MKA Flac ( level ? ) :    438 MB
The DShow encoder currently defaults to a compression level of 5.

>
> Any explanations why overhead is so much worse than for OggFLAC ? Do we
> use lacing in matroskamuxer on DirectShow ? Any news from Gabest about
> using the new lacing sheme in an updated version of matroska muxer filter
?
>

If you remux the file with mkvmerge with lacing enabled you shold get much
lower overhead. The DShow filter wisely does not use any lacing. It could
also have a larger overhead because for some reason each frame has a
reference to it's self.

|+ Cluster
| + Cluster timecode: 0.000s
| + Block group
|  + Block (track number 1, 1 frame(s), timecode 0.000s)
|   + Frame with size 14
| + Block group
|  + Reference block: 0.000ms
|  + Block (track number 1, 1 frame(s), timecode 0.000s)
|   + Frame with size 14
| + Block group
|  + Reference block: 0.000ms
|  + Block (track number 1, 1 frame(s), timecode 0.000s)
|   + Frame with size 14

I think it maybe that the syncpoint flag on the sample in DShow is not being
set.

Later,
Jory





More information about the Matroska-devel mailing list