[Matroska-general] Re: Perspective

Steve Lhomme steve.lhomme at free.fr
Thu Jun 24 10:35:07 CEST 2004

 > The complexity is... Well... B frames / native MPEG4 will be
 > interesting. Native MPEG4 involves removing some parts of the
 > bitstream(in AVI each MPEG4 is preprended with some bits (dunno what
 > they do). In MP4 this is not the case, and Matroska's native mode

Maybe you should ask some Xvid people or FFMPEG people what are these 
bits, they probably know.

 > should not contain this stuff either), and this seems to be easy
 > enough. Getting the players (I'll definitely fix mplayer and try to
 > fix vlc & xine) do accept that should be easy if they can already
 > handle MPEG4 in MP4 which all three do. B frames with proper
 > timestamping is more difficult, though, and I have no idea how long
 > it'll take me.

I see. Well, once it's done... It's done ;)

 > The complexity for TTA/Wavpack is unknown to me so far because I
 > haven't looked at how much source and documentation is available for
 > both projects. I'll take a look at that this weekend.

I had a look at TTA and it's *very* simple. All frames have the same 
duration (except the last one) and a very small header. Dunno about 
Wavpack but I guess it's something like that too.

Both are open source and working on Linux and working on libraries. So 
it should not be long to have MPLayer read these files too.

 > About the 'editor'. This is of course the biggest project. But before
 > I start working one bit for it we'll have to be clear about the
 > license to use. Maybe you've already come to a conclusion, but I don't
 > know of any decisions. The thing is that I won't work on it if the
 > editor will not be Open Source. I don't want a crippled version and a
 > full version. That's not Open Source, and that's not something I want
 > to work for.

Nothing is defined yet. And this is actually a sensitive issue.
IMO if we start it from scratch we'll build something powerful out of 
nowhere and it will be long, will need a lot of input, etc.
If we build it over GStreamer it will be mostly a GUI over an engine. So 
it will take less time (even though GStreamer is probably not mature 
enough on all platforms and we'll spend some time imrpoving it) and less 
people. So I think we should go this way. And there would be no point in 
wanting to get money for such a thing. It could become the GIMP of video 
:D (hopefully nicer)

robUx4 on blog <http://robux4.blogspot.com/>

More information about the Matroska-general mailing list