theora - Jan 2008 - Re: [dyne:bolic] Concerning the (correct) use of Theora in FreeJ

On 1/23/08, jaromil <jaromil@dyne.org> wrote:
> nope. we have a freej specific mailinglist, see http://lists.dyne.org
Ah, so sorry.
> freej is statically including theora-mmx or offering to be dynamically
> compiled  to  system-wide  theora  libraries  (so the  issue  is  then
> delivered to the distribution)
Most distros should be already using the latest beta, so that should be fine.
> in  case theora-mmx  is obsoleted,  would you  suggest an  update?  is
> theora-oil up-to-date with changes? is  there a webpage we can monitor
> with precise updates about these developments?
As far as I am aware, Theora-mmx is obsoleted.  Newer MMX
optimizations were recently added to the main tree (aka "plain" theora
in SVN).

Theora-oil is neither up to date -- again, as far as I am aware.  All
recent development has been directed to main trunk, theora-exp, and
theora-thusnelda.  Theora-exp has pretty much been merged to main
trunk as of the 1.0 beta releases.  There's a new API, much
improvements to quality, and it runs much faster in older/less
powerful systems.

Thusnelda is a very experimental next-generation Theora, aiming to fix
many problems in Theora as well as achieving equal or superior quality
to h.264.  It is not production ready, though.

There is no exact webpage to monitor these changes, I'm afraid.  News
on //theora.org while good to get a general idea, never go much into
detail on what is going on exactly.  Following the mailing list
"theora-dev" and/or the IRC channel usually wield better results,
though it is not optimal for everyone.
> nope, we  didn't knew about this  new convention. we  use theora since
> quite a  while now, it  was simply .ogg  at the beginning,  then .ogm,
> then now  .ogv - i really hope  this is the final  change! people (and
> applications) can get easily confused.
Er, officially Xiph has only issued .ogg for everything, which has
changed last year for .ogv for all video in Ogg (Theora, Dirac,
OggMNG, etc).  .ogm has always been a non-official file extension to
identify the hacked version of Ogg (aka OGM) which usually contains
only MPEG-4 video (aka non-free video).  Please do not use or promote
.ogm.

So, yes, .ogv is final.
> > * a Skeleton bitstream for all output files?
>
> please detail: how can i verify this is produced, in encoded files and
> looking at the source? my current guess is: yes, we have that.
If you are looking for an implementation in code, you may wish to look
into ffmpeg2theora.

To verify that Skeleton is present, run the ogginfo command tool as
compiled from SVN.  The very first bitstream in the Ogg file should
say "skeleton" and not "theora".

Since it appears you have not heard of it here's just a brief
explanation: Skeleton is an extension of Ogg to go around certain
limitations of the container without breaking existing players.  Among
its features, it allows for instance a user to select either the
Korean audio track or the English audio track in a video file that has
two audio tracks in different languages.  Can do the same thing with
regards to subtitles too.

We recommend the use of Skeleton in all new Ogg video files.

-Ivo

P.S: CC'ing theora@xiph.org as well.

On Jan 23, 2008 5:09 AM, Ivo Emanuel Gon?alves <justivo@gmail.com>
wrote:
>
> Thusnelda is a very experimental next-generation Theora, aiming to fix
> many problems in Theora as well as achieving equal or superior quality
> to h.264.  It is not production ready, though.
Thusnelda is planned as a significant update/improvement to the
existing libtheora encoder. It is NOT a 'next-generation Theora' -
it's still Theora, bitstream compatible and all.

It's not at all reasonable or realistic to say it will compete with
H.264; it's intended to substantially improve the quality of the
libtheora encoder, but Theora is simply an older-generation codec than
H.264.

Mike

On 2/16/08, jaromil <jaromil@dyne.org> wrote:
> i hereby confirm we are  correctly producing a skeleton in ogg encoded
> with freej
Thanks!  It is certainly appreciated.
> very good self contained component! cheers to J
Cheers to j.

-Ivo

On 4/4/08, jaromil <jaromil at dyne.org> wrote:
> not in the theorautils.c that we include?
Oops.  I examined all the copies of oggtheora_encoder.cpp instead.
> should i update to the latest version of that file in ffmpeg2theora?
Yeah, if it doesn't break anything in FreeJ that would be a good solution.

-Ivo