[theora-dev] Analogue artifact estimation

Daniel B. Miller dan at on2.com
Mon Sep 2 20:38:47 PDT 2002

On Mon, 2 Sep 2002, Henry Mason  wrote:
> What I guess I'm getting around to is this: Is this kind of work
> (psycho-visual models, filtering, etc.) on the video codec itself
> outside the scope of Theora? Is Theora really 2 projects, one for the
> container, one for the video codec? Do we want to make drastic changes
> to the codec, or should we hold off until Theora 1.0 is out the door?

There is definitely room for work on both the container/stream issues and
the codec proper.  Further, there is a distinction to be made between the
bitstream definition, which it is not practical to change frequently, and
the encoder, which can be completely reworked whenever anyone feels like
it, as long as it produces compatible streams.  This is referred to in
MPEG work as 'normative' (bitstream-related) vs. 'non-normative' (may
affect quality or performance but does not change the playback
specification).  In general, the work of a group such as MPEG or Theora
must always prioritize the normative issues as they directly affect the
specification, which is what we absolutely must get as close to right as
possible out of the box.  The theory is that multiple interested parties
will take up the challenge of improving the encoder once the bitstream is
standardized and starts to get distribution.  Of course this can't happen
unless the initial encoder is at least good enough to be commercially
interesting, so there is some need to build a high-quality encoder sooner
rather than later.

We may want to make an MPEG-style distinction between isses that affect
the specification, and issues that affect quality or optimize for various
input conditions (high resolution, low framerates, etc).  Essentially this
breaks down to work on the decoder and bitstream spec on the one hand, and
a 'reference' encoder on the other. I assume this issue has been addressed
in Vorbis work, so perhaps Monty will chime in here on what Xiph's core
philosophy is on this subject.

> -Henry
 ___  Dan Miller
(++,) CTO and founder, On2 Technologies

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'theora-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.

More information about the Theora-dev mailing list