[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Linphone-users] VP9 experimental release
From: |
Karl Schmidt |
Subject: |
[Linphone-users] VP9 experimental release |
Date: |
Sat, 22 Jun 2013 17:07:46 -0500 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:10.0.12) Gecko/20130116 Icedove/10.0.12 |
I'm thinking that VP9 (open and royalty free) might be a game changer (similar
performance to H265):
Imminent release? :
From
https://groups.google.com/a/webmproject.org/forum/#!topic/webm-discuss/UzoX7owhwB0
(see Jun 12)
Paul Wilkins Jun 12
Here is the latest VP9 status update.
As of last night the VP9 (profile 0) bitstream is FINAL and the process of
validating it on
multiple platforms and in different build environments and testing for
submission into Chrome is
under way.
The last few days have been pretty hectic as we had to make several last minute
changes. As
always seems to be the case, these were more complex than we had hoped and I
want to thank the
whole team for working long hours to bring about closure. I also want to thank
partners and
hardware vendors who have submitted comments and suggestions over the past few
weeks.
Details of the final issues resolved over the last 7 days are as follows:-
Changes to the loop filter. Changes to make the loop filter more HW friendly
had to go through
several iterations and gave rise to a couple of late bugs. As is often the
case, initial attempts
to make it more hardware friendly turned out to make things more difficult in
SIMD software
implementations. Hopefully the final solution is a comfortable compromise for
both.
Changes to the way compound prediction modes are coded in the bitstream. Though
not conceptually
complex and necessary to correctly constrain the allowable compound prediction
modes, this change
touched a lot of files and getting it done on time and debugged was a Herculean
effort by those
involved.
Definition and implementation of behaviour at the edge of images. This relates
to what prediction
modes/sizes and transforms are allowed near the edge of an image that does not
break down nicely
into 64x64 blocks. I mentioned this at the summit and while it sounds simple,
like all the
simplest changes, it was actually quite hard to get right. Again a big thank
you to all involved
in working through the issues.
Some minor changes to the frame header (E.g. coding of frame size not needed
for all frames)
Some minor changes to the entropy coding of some symbols and some updates to
the default tables.
Paul Wilkins
--------------------------------------------------------------------------------
Karl Schmidt EMail address@hidden
Transtronics, Inc. WEB
http://secure.transtronics.com
3209 West 9th Street Ph (785) 841-3089
Lawrence, KS 66049 FAX (785) 841-0434
Truth is mighty and will prevail.
There is nothing wrong with this,
except that it ain't so.
--Mark Twain
--------------------------------------------------------------------------------