On Jan 28, 2008 9:46 AM, David Gerard <dgerard(a)gmail.com> wrote:
The Dirac video codec is nearly fixed and part of it
is set to become
an SMPTE standard. The BBC has tried hard to ensure it is as
unencumbered as possible.
As an aside, what is up for SMPTE is "Dirac pro", an simplified
i-frame only version of Dirac. It is the dirac-equivalent of MJPEG.
As such it's not suitable for streaming but is a really good high
bandwidth archival format.
http://dirac.sourceforge.net/specification.html
Apart from the coder and decoder both being somewhat inefficient at
present, I understand there isn't much that plays it back as yet (VLC
in experimental builds).
What would we need before allowing Dirac-encoded video on Wikimedia?
Allowing? Well since there Dirac-in-Ogg is well defined and audio is
needed for many uses, arguably we allow it already! \0/
(
http://wiki.xiph.org/index.php/OggDirac)
More seriously, before dirac pro would useful as anything other than a
"source/archival format" we'd need infrastructure for transcoding.
For real use of Dirac (not Pro) on the sites getting real-time
decoding in Java would be needed to get anywhere parity with Theora.
This may well be really hard, but there has been some work done on
this (
http://dirac.cvs.sourceforge.net/dirac/soc_javadec_luis/). (In
fact, it is quite possibly not possible on current hardware :( )
In general Dirac is about two years behind on tools but it should
catch up quickly now that the specifications are done.
None of these are reasons that it couldn't be allowed for upload, but
they are reasons why no one may care yet. :)