[MPlayer-users] Re: Quicktime: "No 'moov' atom could be found"

[rcooley]

Moritz Bunkus wrote:
> Ogg does not have the potential to become a general container. It simply
> lacks a lot of important things. Most important: there are no specs!
First of all, it has a lot of potential, because it is so flexible that 
it can do just about anything, without everything it can do, having been 
defined ahead of time.  That gives it more potential, not less IMHO. 
Sure, there's no specs yes, but it's just a matter of somebody writing 
them.  All this software is open source.  Pretty easy to look up how 
everything is done, even without specs.

> The thing is that Ogg/Ogm development is only done by one person atm
> (Nic on doom9). He's a Windows guy using a Windows approach (like using
> WAVEFORMATEX and BITMAPINFOHEADER like structures). Not that HAS to be
> bad, but neither is xiph.org involved in all this. And being a one man
> development team he is not paying a lot of attention to cross platform
> issues.
ogmtools are based upon oggtools, which was, and still can be pulled 
from the xiph.org CVS server.  I don't really know what you mean that it 
isn't cross-platform (mainly because I haven't looked at the source yet) 
but the fact that ogmtools (and libogg/MPlayer) works perfectly on just 
about every platform, is a sign that it is, for all practical purposes, 
cross-platform.

> Let's take text subtitles. Ogg does not have a standard way of storing
> which charset was used for them. Neither do they have to be stored as
> UTF-8.
Just not popular yet.  Feel free to get involved and solve all the 
things that haven't been worked-on/standardized yet.

> Next point: No index whatsoever. Pretty big overhead (oh yes!). No way
The overhead is larger than something like avi, but it's relatively low.

> to know if an Ogg page is part of an Ogg packet spanning several pages
> (so I have to throw away the first Ogg page after seeking). Theora is
> stored differently in Ogg than all the other stuff is. No "real"
> timestamps (just the "granulepos" which is based on the "sample rate"
> for the stream - e.g. 1, 2, 3, 4... So no vfps video at the moment
> without braking compatibility). No aspect ratio. No standarized way of
> storing ANY additional information - be it pictures, text, comments,
> playlists...
Ogg is flexible.  As soon as someone puts in aspect tags (hasn't been 
necessary, due to the MPEG-4 aspect tag)  then they will exist, and will 
get adopted by the rest of the Ogg software.  It just hasn't been done yet.

And BTW, Theora isn't even to beta yet, so it's quite likely there will 
be significant changes.  It's open source, so you very well could get 
involved and solve everything that you consider to be a problem.


>>The main advantage Quicktime would have over Ogg and Matroska is that it 
>>is more standardized and widely implemented in the proprietary world.
Well, Ogg is just too new to be standard.  However, because it is 
BSD-licensed, it has a MUCH better chance of being adopted than 
Matroska.  (Sorry all you hard-core GPLers, but that's the real world 
for you)

> Yes. But Matroska has specs, Ogg doesn't (only very basic ones). I don't
> see Ogg winning the race for the Next General Purpose Container
> Format. Maybe Matroska doesn't win either.
Ogg is very good, and it is here, right now.  Quicktime container would 
be good, but so far I haven't seen too much support for it, outside of 
Quicktime itself.  Sure, there's plenty of playback support, but few 
programs that encode/edit quicktime files.