IETF
videocodec@jabber.ietf.org
Monday, 5 November 2012< ^ >
stpeter has set the subject to: Video Codec BoF, IETF 85 | agenda and logistics: https://datatracker.ietf.org/meeting/85/agenda/videocodec/
Room Configuration

GMT+0
[05:16:28] stpeter joins the room
[05:18:57] stpeter leaves the room
[10:09:23] thdavies joins the room
[10:10:17] thdavies leaves the room
[10:21:54] thdavies joins the room
[10:21:54] irc- leaves the room
[10:23:30] thdavies leaves the room
[14:16:48] stpeter joins the room
[15:51:10] thomas joins the room
[16:28:43] rillian joins the room
[16:33:54] stpeter leaves the room: Disconnected: connection closed
[16:42:14] thomas75732 joins the room
[16:42:14] thomas75732 leaves the room
[16:42:14] thomas11569 joins the room
[16:42:14] thomas11569 leaves the room
[16:42:14] thomas39065 joins the room
[16:42:14] thomas39065 leaves the room
[16:42:14] thomas19350 joins the room
[16:42:31] thomas leaves the room
[16:46:51] thomas19350 leaves the room
[16:58:55] irc- joins the room
[16:59:01] <irc-> * members: dD0T, derf, DuClare, gmaxwell, gnafu, rillian, terfran, wiking
[17:32:52] thomas joins the room
[17:32:52] irc- leaves the room
[17:34:12] <rillian> hmm
[18:15:19] louquillio joins the room
[18:15:49] louquillio leaves the room
[18:18:09] stpeter joins the room
[18:18:46] irc- joins the room
[18:26:12] gmaxwell joins the room
[18:27:09] <gmaxwell> I suspect the irc bridge may be less likely to time out if there is traffic.
[18:28:42] <irc-> [gnafu] One can only hope.
[18:29:49] <rillian> gnafu: indeed
[18:32:44] <irc-> [gnafu] )^_^(
[19:02:30] <gmaxwell> jxself: it's still a couple hours out now but feel free to invite other people!
[19:57:00] stpeter leaves the room: Disconnected: connection closed
[20:01:29] gmaxwell leaves the room
[20:11:11] gmaxwell joins the room
[20:45:30] stpeter joins the room
[21:09:17] <irc-> [onemanx] Greetings IETF netizens, I come in peace .. and surprisingly early
[21:34:54] danyork joins the room
[21:36:49] stpeter leaves the room: Disconnected: connection closed
[21:48:06] N.e.k.i.t joins the room
[22:01:55] louquillio joins the room
[22:02:42] katwalsh joins the room
[22:07:49] <rillian> hi katwalsh
[22:09:19] katwalsh waves
[22:13:18] j joins the room
[22:16:15] EHalls joins the room
[22:20:53] <rillian> hi j
[22:21:07] <irc-> [gmaxwell] Around the start of the meeting someone in IRC should /who and paste in the names of the IRC participants as they aren't visible in XMPP.
[22:21:30] <danyork> We have IRC, too?
[22:22:29] <rillian> danyork: irc- is providing a bridge with ##IETF_videocodec on irc.freenode.net
[22:22:46] <danyork> cool!
[22:22:56] <irc-> [rillian] it is :)
[22:23:03] <gmaxwell> There is an IRC bridge on irc.freenode.new ##IETF_videocodec (this is something we used for the past Audio codec working group sessions)
[22:24:20] <gmaxwell> Getting some of the assorted open source developers onto xmpp can be hard, especially at (locally to them) odd hours.
[22:24:28] <danyork> :-)
[22:24:43] tterribe joins the room
[22:25:00] jmspeex joins the room
[22:25:05] HEx joins the room
[22:25:11] jesup joins the room
[22:25:38] jmspeex leaves the room
[22:25:38] tterribe leaves the room
[22:25:38] <irc-> *** gnafu enjoys "listening" in when he can.
[22:29:26] <irc-> [Nekit1234007] it died?
[22:30:34] EHalls leaves the room
[22:30:48] wwcv joins the room
[22:30:52] Audio videocodec joins the room
[22:31:10] tterribe joins the room
[22:31:29] wwcv leaves the room
[22:31:30] Lorenzo Miniero joins the room
[22:31:44] Anant Narayanan joins the room
[22:31:52] burn joins the room
[22:32:01] Alessandro Amirante joins the room
[22:32:08] hardie@jabber.psg.com joins the room
[22:32:49] <irc-> [JelloLag] has the audio stream begun?
[22:33:12] <gmaxwell> jellolag: What do you hear?
[22:33:20] <irc-> [JelloLag] nothing
[22:33:36] <rillian> http://atlanta.conf.meetecho.com:8000 <http://atlanta.conf.meetecho.com:8000/>/zr0o3mr.opus
[22:33:36] <Lorenzo Miniero> we have started the streaming in the Meetecho room
[22:33:48] <Lorenzo Miniero> even though the session hasnt started yet
[22:33:55] <Lorenzo Miniero> so all you'd hear is background noise
[22:34:01] <rillian> thanks Lorenzo!
[22:34:14] dd0t joins the room
[22:34:28] <irc-> [JelloLag] thanks rillian
[22:34:36] <N.e.k.i.t> http://nagasaki.bogus.com:8000/stream02 this one doesn’t work
[22:35:02] <rillian> the mp3 streams seem to still be offline
[22:35:02] <irc-> [JelloLag] thanks lorenzo
[22:35:06] <burn> FYI, either Ted Hardie or I will be relay for remote folks. We'll let you know :)
[22:35:26] Tobia Castaldi joins the room
[22:35:29] <rillian> burn: great, thanks. We had questions about whether there would be a scribe
[22:35:31] <Lorenzo Miniero> to get all media in a synchronized view, there's our Meetecho room available http://www.meetecho.com/ietf85/videocodec
[22:35:34] <irc-> *** gnafu wgets the Opus stream to listen to later.
[22:35:39] <Lorenzo Miniero> slides+audio/video+jabber
[22:35:44] hta joins the room
[22:35:53] <Lorenzo Miniero> or if you're not interested in jabber http://atlanta.conf.meetecho.com/WebLite/event.jsp?w=videocodec
[22:36:02] <gmaxwell> gnafu: thanks for thinking to fetch the stream, I always remember late.
[22:36:10] <Lorenzo Miniero> both options have several streaming alternatives to get what's going on
[22:36:19] Thomas joins the room
[22:36:25] katwalsh listening to the background noise on the opus stream.
[22:36:39] jmspeex joins the room
[22:37:02] <gmaxwell> I wonder if this is the first IETF bof to be streamed in Opus?
[22:37:02] kyle joins the room
[22:37:12] <Lorenzo Miniero> might be :)
[22:37:17] Ralph Giles joins the room
[22:37:20] John Koleszar joins the room
[22:37:24] <Lorenzo Miniero> this is the first meeting we've added the Opus streaming to
[22:37:38] <Lorenzo Miniero> you all have to thanks Ralph Giles for that ;)
[22:37:47] Edward Halls joins the room
[22:37:58] <hta> :-)
[22:38:05] Jan Skoglund joins the room
[22:38:08] <irc-> [gnafu] \o/
[22:38:48] Edward Halls leaves the room
[22:39:00] Jonathan Lennox joins the room
[22:39:01] David Flynn joins the room
[22:39:08] tomoyuki tsuichihara joins the room
[22:39:09] John Leslie joins the room
[22:39:30] <gmaxwell> Unfortunately it's narrowband. But hey, good start.
[22:39:49] <Lorenzo Miniero> wideband = next step on the list :)
[22:40:03] <tterribe> Thank you rillian.
[22:40:07] Martin Thomson joins the room
[22:40:16] Simon Romano joins the room
[22:40:21] plehegar joins the room
[22:40:25] <irc-> [JelloLag] thanks
[22:40:34] <irc-> [JelloLag] got streaming using html5
[22:40:53] tsuichi joins the room
[22:41:04] dd0t leaves the room: offline
[22:41:10] Stefan Hacker joins the room
[22:41:37] David Flynn leaves the room
[22:41:38] <Lorenzo Miniero> Slide 1: Video Codec BoF IETF 85
[22:41:50] m&m joins the room
[22:42:02] wwcv joins the room
[22:42:08] <gmaxwell> Are people getting some audio on the streams now?
[22:42:18] <irc-> [oneman_] yes
[22:42:18] <irc-> [oneman_] works for me
[22:42:19] David Flynn joins the room
[22:42:22] <irc-> [oneman_] the opus stream
[22:42:26] <irc-> [oneman_] http://atlanta.conf.meetecho.com:8000/zr0o3mr.opus
[22:42:29] <John Koleszar> rtsp stream is down
[22:42:45] <burn> Okay, I am relay for remote folks. If you want me to bring your comment/question to the mic, prefix with 'mic:'
[22:42:46] <Alessandro Amirante> I'm listening to it, it seems to work
[22:43:02] <Lorenzo Miniero> rtsp://atlanta.conf.meetecho.com/zr0o3mr.sdp
[22:43:05] barryleiba joins the room
[22:43:06] <irc-> [rillian] participants from IRC broos derf JelloLag Nekit1234007 ron_ DuClare jk8 oggkoggk terfran danyork_ gmaxwell jxself oneman_ wiking dD0T gnafu louquillio_ rillian wwcv
[22:43:07] <Lorenzo Miniero> did you open this one?
[22:43:23] <Lorenzo Miniero> Slide 2: Note Well
[22:43:32] <irc-> [JelloLag] im a silent participant ok ?
[22:43:37] <irc-> [JelloLag] just listening
[22:43:38] <Lorenzo Miniero> Slide 3: Logistics
[22:43:49] <rillian> JelloLag: all are welcome
[22:43:55] <irc-> [JelloLag] thanks, :D
[22:43:57] <Lorenzo Miniero> Slide 4: Progress to Date
[22:44:09] sftcd joins the room
[22:44:14] <John Koleszar> still 404, but I've got the video stream up now, I'm good.
[22:44:18] Thomas Davies joins the room
[22:44:21] <Lorenzo Miniero> ok :)
[22:44:37] <Lorenzo Miniero> Slide 5: Goals of this Meeting
[22:44:39] <irc-> [JelloLag] John, log into meetecho and select html5 option
[22:44:47] David Flynn leaves the room
[22:45:03] David Flynn joins the room
[22:45:14] <irc-> [JelloLag] or one ofthe other options, that works for me
[22:45:34] <Lorenzo Miniero> Slide 6: Agenda Bashing (1)
[22:45:59] <Alessandro Amirante> @John Koleszar: if you want to listen to the RTSP stream, you need to open the link with an external player (e.g., VLC or QuickTime)
[22:46:26] <rillian> G.711 stream at rtsp://atlanta.conf.meetecho.com/zr0o3mr.sdp which is lower delay than the opus stream
[22:46:36] <Lorenzo Miniero> Slide 7: Agenda Bashing (2)
[22:46:39] <Lorenzo Miniero> Slide 8: Agenda Bashing (3)
[22:46:51] <Lorenzo Miniero> Presentazione interrotta
[22:46:51] cgriffiths joins the room
[22:46:54] <burn> Cullen Jennings from the front
[22:47:06] <Lorenzo Miniero> Slide 1: Lessons to learn from the Codec WG Chart
[22:47:18] <Lorenzo Miniero> Slide 2: If I could do it all over again ...
[22:48:07] thdavies joins the room
[22:49:02] <David Flynn> Can the presenter speak a little more slowly, it is very difficult to hear over VoIP
[22:49:06] cgriffiths leaves the room
[22:49:27] <irc-> [oneman_] Opus stream is very clear
[22:49:28] Yana Stamcheva joins the room
[22:49:44] <David Flynn> This is the telephone dialin
[22:49:53] <irc-> [oneman_] I am listening to it and watching the flash video, which sounds poor
[22:50:04] <rillian> oneman_ you must be listen to a different stream!
[22:50:31] <irc-> [oneman_] Indeed it is
[22:51:08] <irc-> [oneman_] the audio on the flashvideo is a throwback ;p
[22:51:13] Roy joins the room
[22:51:51] Jan Skoglund leaves the room
[22:52:23] Jan Skoglund joins the room
[22:52:42] fireglow joins the room
[22:52:56] Steven Sokol joins the room
[22:53:30] <burn> Jean-Marc Valin at mic
[22:54:29] Steven Sokol leaves the room
[22:54:37] hildjj joins the room
[22:55:38] <burn> Stefan Wenger at mic
[22:56:17] <burn> Christian Hoene at mic
[22:56:43] <irc-> [ron_] RFity through obscurity?
[22:56:54] <burn> Tim Terriberry at mic
[22:57:08] <Lorenzo Miniero> Presentazione interrotta
[22:57:23] <burn> Greg Maxwell up front
[22:57:27] <Lorenzo Miniero> Slide 2: ??
[22:57:44] <Lorenzo Miniero> Slide 2: Introduction
[22:58:47] richard.bares joins the room
[22:58:48] <Lorenzo Miniero> Slide 3: Applications
[22:58:48] richard.bares leaves the room
[23:00:15] Thomas Davies leaves the room
[23:00:27] <Lorenzo Miniero> Slide 4: Constraints of the Internet
[23:01:06] Thomas Davies joins the room
[23:02:09] Alex Converse joins the room
[23:02:50] <Lorenzo Miniero> Slide 5: Requirement details
[23:03:46] mreavy joins the room
[23:03:58] <Lorenzo Miniero> Slide 6: Comments so far
[23:04:48] Randall Gellens joins the room
[23:05:11] Jason Self joins the room
[23:05:19] <burn> Stefan Wenger at mic
[23:05:30] Erik Lagerway joins the room
[23:06:21] <burn> Mark Francisco at mic
[23:08:40] <rillian> multiviews? 3d?
[23:09:02] <burn> Alan Morton (sp?) at mic
[23:10:01] george oloo joins the room
[23:10:06] <burn> Kevin Gross at mic
[23:10:56] <kyle> Wireless networks tend to suffer more packet loss, as well as bandwidth instability.
[23:11:03] <burn> Tim Terriberry up front
[23:11:20] <Lorenzo Miniero> Slide 2: Introduction
[23:11:42] <Lorenzo Miniero> Slide 3: Proposed Process
[23:12:14] <burn> Reminder: if you are remote and want me to speak a comment or question at the mic, please prefix your message with 'mic:'
[23:12:33] krasimir kolarov joins the room
[23:12:33] <Lorenzo Miniero> Slide 4: Proposed Process (cont.)
[23:13:16] <Lorenzo Miniero> Slide 5: Intellectual Property
[23:14:45] <Lorenzo Miniero> Slide 6: Intellectual Property (cont.)
[23:16:24] <Lorenzo Miniero> Slide 7: Relationship with Other SDOs
[23:17:00] <Lorenzo Miniero> Slide 8: Relationship with Other SDOs (cotd.)
[23:17:21] <rillian> lol
[23:17:23] Adnan Saleem joins the room
[23:17:25] Piers O'Hanlon joins the room
[23:17:50] <Lorenzo Miniero> Slide 9: Testing and Characterization
[23:18:25] Piers O'Hanlon leaves the room
[23:18:29] Thomas Davies leaves the room
[23:18:52] Thomas Davies joins the room
[23:19:23] <Lorenzo Miniero> Slide 10: Specification and Conformance
[23:21:26] <burn> Nathan Egge up front
[23:21:38] <Lorenzo Miniero> Slide 2: Motivation
[23:21:44] george oloo leaves the room
[23:22:03] <gmaxwell> Unusually quiet in Jabber today. Too late in Europe I guess. :)
[23:22:26] <Lorenzo Miniero> it's just midnight in Italy, we usually go to sleep much later! :D
[23:22:30] <Lorenzo Miniero> Slide 3: Prefilter and Postfilter
[23:23:25] <Lorenzo Miniero> Slide 4: Lapped Transforms: Prefilter
[23:23:56] UBSggu joins the room
[23:24:13] <Lorenzo Miniero> Slide 5: Basis Functions
[23:24:34] <rillian> so the synthesis filter is the inverse of the analysis filter?
[23:24:45] <rillian> P and P^-1 from the earlier slide?
[23:24:56] <Lorenzo Miniero> Slide 6: Regularized Transforms
[23:24:57] UBSggu leaves the room
[23:25:39] <Lorenzo Miniero> Slide 7: Coding Gain
[23:25:52] <burn> rillian: believe so
[23:25:54] sftcd leaves the room
[23:26:07] <rillian> thanks
[23:26:35] <Lorenzo Miniero> Slide 8: Future Work
[23:26:53] sal joins the room
[23:27:00] <gmaxwell> rillian: The transform is technically bi-orthorgonal, not orthogonal. So you get perfect reconstruction but its not a simple inverse.
[23:27:38] <gmaxwell> (the fades out to nothing behavior requires it not be orthogonal)
[23:27:56] cgriffiths joins the room
[23:28:42] <burn> Jean-Marc Valin up front
[23:29:03] sal leaves the room
[23:29:19] <Lorenzo Miniero> Slide 2: Motivations
[23:30:15] devnull joins the room
[23:30:42] <Lorenzo Miniero> Slide 3: Transform Codec
[23:31:02] <Lorenzo Miniero> Slide 4: Gain-Shape Quantization
[23:32:33] <Lorenzo Miniero> Slide 5: 2-D Projection
[23:32:44] <Lorenzo Miniero> Slide 6: 2-D Projection
[23:33:03] <Lorenzo Miniero> Slide 7: 2-D Projection
[23:33:22] <Lorenzo Miniero> Slide 8: 2-D Projection
[23:33:48] <Lorenzo Miniero> Slide 9: 2-D Projection
[23:34:06] <Lorenzo Miniero> Slide 10: 2-D Projection
[23:34:29] <hardie@jabber.psg.com> So, in a screen sharing application, where you may have an application that "pops" to the front or is raised from an icon with some funky animation, do you have to have specific predictions for these activities? Or does this just lose efficiency massively in this approach?
[23:34:41] <Lorenzo Miniero> Slide 11: Results
[23:34:45] Pat Jensen joins the room
[23:35:09] plehegar leaves the room
[23:35:10] <rillian> mic: are those images at the same bitrate?
[23:35:52] plehegar joins the room
[23:36:00] <hildjj> he said that they *are* at the same bitrate.
[23:36:07] <rillian> hildjj: thanks
[23:36:39] <Lorenzo Miniero> Slide 12: IPR
[23:36:39] thdavies leaves the room
[23:36:45] <David Flynn> how about compared to something a little more modern?
[23:37:03] <rillian> David Flynn: like jpeg-xr?
[23:37:27] <burn> Adrian Grange up front
[23:37:28] <David Flynn> jpex-xr, AVC-I, etc.,
[23:37:39] <N.e.k.i.t> webp possibly? (VP8 i-frame coding)
[23:37:44] <irc-> [oneman_] I just wanted to say I'm enjoying the show
[23:38:00] <gmaxwell> The real point of the image isn't that it looks good. It's _how_ it looks good: there is no blocking, and the contrast and texture is preserved in flat areas, etc... I wish JM could have pointed to that a bit better.
[23:38:07] <Lorenzo Miniero> Slide 2: The WebM Project
[23:38:08] <irc-> [jxself] Me too.
[23:38:16] <Lorenzo Miniero> Slide 2: VP-Next
[23:38:21] <Lorenzo Miniero> Slide 3: VP-Next
[23:38:33] <rillian> in fact, they both look terrible. :) but the point is the improvement in the artefacts
[23:38:59] <rillian> but comparing against newer image codecs is a good point for the ongoing metrics Tim mentioned
[23:39:17] <Lorenzo Miniero> Slide 4: List of Techniques
[23:39:30] <N.e.k.i.t> might be livestream artifacts as well
[23:39:30] <Lorenzo Miniero> Slide 5: Techniques
[23:40:06] <irc-> [oneman_] I enjoy why the fruit looked good quite a bit, looking forward to the end of blocking
[23:40:48] <Lorenzo Miniero> Slide 6: Techniques
[23:40:59] <rillian> N.e.k.i.t: there's a cleaner version of the figures at http://www.ietf.org/proceedings/85/slides/slides-85-videocodec-3.pdf
[23:41:12] <gmaxwell> oneman: you can avoid blocking with only a post-filter like the inloop deblocking used in theora, vp8, and h264. But it also makes the image blurry. .. and then you have to do stuff like adapt the deblocking intensity.
[23:41:37] <N.e.k.i.t> rillian, thanks
[23:42:01] <N.e.k.i.t> much better
[23:42:30] <gmaxwell> oneman: or by using some non-block oriented wavelet techniques. (but there have been perceptual challenges with those, at least a lower rates)
[23:42:44] <Lorenzo Miniero> Slide 8: Techniques
[23:44:10] <Lorenzo Miniero> Slide 9: Techniques
[23:44:46] <Lorenzo Miniero> Slide 10: Techniques
[23:45:32] Joshua Colp joins the room
[23:45:53] <Lorenzo Miniero> Slide 11: Techniques
[23:46:53] <rillian> encoding the integer and fractional parts separately? that's surprising.
[23:46:55] <Lorenzo Miniero> Slide 12: Conclusion
[23:47:27] <rillian> does that help by letting one distribute the error to the less-correlated component?
[23:47:34] <David Flynn> -7% worse or better?
[23:48:00] <gmaxwell> rillian: I don't know if it's their motivation but it lets you use a fancy codebook (makes the number of possible values small)
[23:48:06] <rillian> David Flynn: I assume that's worse than HEVC
[23:48:14] DanY joins the room
[23:48:21] <rillian> mic: I'm happy to hear about lossless scaling. Have you looked at alpha-channel or 3D coding modes?
[23:48:24] <David Flynn> guess so, as "move ahead of HEVC"
[23:48:37] <gmaxwell> It would be interesting to also know how much faster their encoder is than the JM HEVC they're comparing too.
[23:48:57] <burn> christian hoene at mic
[23:49:45] <David Flynn> its not difficult to make a faster encoder than JM or HM
[23:50:40] <gmaxwell> David Flynn: indeed.
[23:51:54] <rillian> thanks
[23:51:55] <David Flynn> who was asking the alpha/3d question at the mic?
[23:52:22] <hardie@jabber.psg.com> He was channelling rilian, but it was Dan speaking
[23:52:26] <burn> David, it's Dan Burnett (burn), jabber relay
[23:52:38] <Lorenzo Miniero> Slide 1: Closing Questions (cf. RFC 5434)
[23:52:48] <David Flynn> Ah okay, i missed the very start
[23:53:04] <gmaxwell> Christian was incorrect about the patentablity of material made available to the public. Operating in the public is not incompatible with filing patents. (Under US law (both current and upcoming) publication starts a 1-year clock for filing applications.)
[23:53:13] <burn> Now we have Peter St. Andre from the front again
[23:53:40] <Lorenzo Miniero> Proposed charter projected right now: http://www.ietf.org/mail-archive/web/video-codec/current/msg00002.html
[23:53:42] <burn> Tim Terriberry at mic
[23:54:10] <burn> Cullen Jennings at mic
[23:54:15] Joshua Colp leaves the room
[23:56:17] <burn> Harald Alvestrand at mic
[23:56:26] <danyork> Thank you, Dan (burn), for relaying and saying who is at the mic.
[23:56:59] <burn> Dan, my pleasure.
[23:57:15] <burn> Stefan Wenger at mic
[23:58:09] <burn> Tim Terriberry at mic
[23:58:29] <burn> Christian Hoene at mic
[23:59:07] <burn> Tim Terriberry at mic
[23:59:23] <burn> (looking at the problem statement on screen)
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!