[whatwg] Codecs (was Re: Apple Proposal for Timed Media Elements)

[Maik Merten]

vladimir at pobox.com schrieb:
> I actually agree with this -- I think that MPEG-4 already has lots of heavy weight behind it and is quite a good format with lots of existing implementations.  Theora/Vorbis are definitely the upstarts in this; they should live and die on their technical merits and adoption, not because of philosophical (i.e. open source) reasons.  Personally, I think that Theora is quite strong quality-wise, but it's severely lacking on the adoption front.

Well, personally I don't see why adoption elsewhere would matter for web
video. If the spec was about a plugin based video tag then it may matter
as it would rely on whatever codecs there are deployed, but we're
talking about shipping a decoder of whatever base format happens to get
chosen. So it'll work out of the box, which is what matters.

MPEG4 adoption to the web has been poor from my point of view. Today I'd
guess the absolute king in marketshare is Flash video, then following
Windows Media, then followed by QuickTime (that may carry MPEG4... but
the container is not MPEG) and perhaps a bit of RealVideo in between.

As for Ogg: It's predominant on Linux just like Windows Media is
predominant on Windows. There's a whole ecosystem relying on the Ogg set
of codecs - and it seems they're doing fine.


> To that end, I'd suggest that the spec not specifically require Theora support, but instead /suggest/ that implementations support Theora, MPEG-4, or both.

The current spec says that UAs may implement whatever video format they
like, but that they should implement Ogg Vorbis and Ogg Theora. If Apple
wants to adopt MPEG4 (they can afford this and are member of the MPEG
Industry Forum, so their will to adopt MPEG should be rather clear) they
can do so.

As long as nobody proposes another format that can be distributed freely
without licensing issues I'd strongly propose keeping the Ogg codecs in
a "SHOULD" state.


> I don't agree with the earlier comment that Theora would be good for 'everyone' -- there are far more content producers out there with MPEG-4 software, hardware, and knowledge than there are Theora content producers.  Specifying Theora as the baseline could just as easily have the opposite effect than intended: content authors could simply say 'thanks, but no thanks' and continue using their plugin based solutions.  I think that is a far, far worse alternative.

I don't really see how the current production chain happens to be a
problem. If you're happening to produce content in MPEG you nowadays
have to convert your content to Flash Video anyway. It's no problem to
instead convert to Ogg Theora (there are existing and easy to use tools
for that). Your existing content most likely isn't suitable for web
streaming anyway due to bitrate concerns - so in most cases a conversion
stage is inevitable.