[whatwg] on codecs in a 'video' tag.

Maik Merten maikmerten at gmx.net
Thu Mar 29 02:00:35 PDT 2007

Dave Singer schrieb:
>> - You do want a baseline codec in the spec, and want it to be one
>> everyone can implement - i.e. you are happy for Ogg Theora (or another
>> codec with a similar IP position, such as Dirac) to be it
> Until someone starts using the Ogg family to make money, and in such a
> way that any possible IPR owners consider it in their business interests
> to start enforcing their IPR, the situation remains in question.  We
> have nothing against these codecs, but we are not currently feeling like
> being the guinea-pig...

Ogg Vorbis is widely used for commercial applications. There are
hardware players out there (I have one sitting on my desk) and basically
all major game studios are using it (take a look on
http://wiki.xiph.org/index.php/Games_that_use_Vorbis ) - even Microsoft
uses it for games!

According to http://www.mp3licensing.com/royalty/games.html each of
those released titles would have led to licensing income of at least
$2500. The total amount "lost" by Vorbis is in the hundreds of thousands
by now. If e.g. Fraunhofer of Thomson have any patents that apply to
Vorbis they have a pretty good reason to try to kill it.

Nullsoft's Winamp, which is (or at least used to be) the number one "MP3
player" for the Windows platform is shipping a Vorbis decoder since ca.

As for Theora: It is based on VP3, which is patended. On2 granted an
irrevocable free patent license for anything concerning VP3. Theora is
using the very same coding, just that the new superset of bitstream
features allows a way more flexible handling of quantization matrices
(every quality level may have its own set of tables, tables can be
switched per-block etc. so Theora can be more efficient - VP3 had
hardwired quantization tables) and adds support for additonal color
spaces. The changes increase compression efficiency without turning
Theora into a completely different thing. You can convert VP3 video to
Theora with a simple utility (Theora to VP3 obviously doesn't work
because VP3 can't store the additional quantization tables and doesn't
support all color formats Theora supports).

Some players in fact use a VP3 decoder to decode Theora (albeit that
means these players are out of the spec and will fail once Theora is
using the additional flexibility specified).

VP3 was born as a commercial codec. Apple licensed it.


VP3 was also used for Nullsoft's Winamp to receive Shoutcast video streams.

The successors of VP3, which evolved from it, are in commercial use
today. VP6 is part of Flash Video. VP5 can be found in some video
conferencing solutions.

If Theora was using patented coding methods not covered by On2's own
patents commercial entities would have had reason enough to kill it by now.

Now, of course there *may* still be submarine patents out there. But
that possibility applies to all codecs that are younger than 20 years.
Microsoft was hit by a MP3 submarine patents despite licensing the usual
MP3 patents. Same could happen to AAC and H.264 as licensing from
MPEG-LA gives zero security against submarine patents.

Maik Merten

More information about the whatwg mailing list