Re: [AVT] Breakfast meeting to discuss timestamp alignment issues

["Jonathan Lennox" <jonathan at vidyo.com>]

Roni Even, Sam Ganesan, Colin Perkins, Thomas Schierl and I (Jonathan
Lennox) met over breakfast this morning.


The issue we discussed is that RFC 3550's definition of multi-session
transmission requires that the session group have a common base session,
which every session member joins and sends RTCP to, so that SSRC
collision detection can be performed.

In principle, however, H.264-SVC's dependency structure doesn't require
a common base layer.  Though in practice we expect most SVC usages to
have one, future codecs using multi-session transmission (notably H.264
Multiview Video Coding [MVC]) may actually have independently-decodable
sessions be useful and common.  My timestamp alignment draft assumed
3550 rules, so there is a conflict.

Several different possible resolutions to this problem are possible, all
with advantages and disadvantages.  Thomas Schierl and I will be writing
an I-D describing the requirements and these possible resolutions, to
make working group discussion concrete.

-- 
Jonathan Lennox
Vidyo, Inc
jonathan at vidyo.com

_______________________________________________
Audio/Video Transport Working Group
avt at ietf.org
https://www.ietf.org/mailman/listinfo/avt