[theora-dev] Scalable Extension

Timothy B. Terriberry tterribe at email.unc.edu
Thu Feb 28 05:18:29 PST 2008


Horacio Sanson wrote:
> With the newly released SVC (Scalable Video Coding) of H264 I was
> wondering if Ogg/Theora has any plans of following suit or are there
> plans to implement a better solution? the closest to SVC I have seen at
> Xiph is bitrate peeling but that never happened and to date does not
> seem to be in any TODO list.

The short answer is "no". MPEG4/AVC contains many features which, while
of interest to the academic research community and some specialty
applications, are not widely demanded by the community at large. We've
got plenty of things on the TODO list to tackle that _are_ widely
demanded, and only limit resources to devote to them.

Bitrate peeling was a neat theoretical idea, but in practice turned out
to be almost completely useless. The decode side has always been
implemented, as it is part of the normal Vorbis spec. The actual peeling
process itself was trivial: simply truncate packets. However, it would
have required a major restructuring of the Vorbis encoder to produce
streams suitable for peeling in this manner, and it was unclear if this
would even work. Theora uses a much simpler scalar quantization scheme
that makes even this reorganization impossible without a major redesign
that would require a new decoder as well as a new encoder (a la MPEG4's
SVC).

If this is a feature that's truly important to you, feel free to attack
the problem yourself.


More information about the theora-dev mailing list