Daala project

Action items:

  • Over the next year will be more opportunity for broader community involvement and industry support.


Action Items:

  • Get Siliva involved;
  • Form group to push forward on programmatic validation!

Second Screen

Action items:

  • Support HTML5? on big screens / reciver apps.
  • Review creation of consolidated casting library.
  • Propose dial protocol includes support for feature detection.( what codecs it can play etc )

Adaptive Streaming

  • Stage of dash.js adoption;
  • performance version of dash.js? ( small )
  • making the core smaller.
  • Mozilla focused on MSE to ship ( less focused on video metrics )
  • https://wiki.whatwg.org/wiki/Video_Metrics
  • would be great to consolidate these.
  • video analytics:
  • what happens until the data gets delivered and what happens in the player itself.
  • Buffer ranges; MSE rough bandwidth calculations “good enough” ?
  • Video tag vs browser based implementation.
  • Multi-track audio apple story introduced in iOS 7.2 and broke in 8.0
  • is Dash.js sufficient??
  • google shaka player -- https://github.com/google/shaka-player
  • bitdash -- http://www.dash-player.com/
  • Rate adaptation for dynamic adaptive streaming over HTTP - academic paper describing proposed adaptive metrics
  • Interchangeable adaptive component.
  • What is the current status of IPR ? ( intellectual property rights )
  • 85% of ads still going through flash.

Action Item

  • Video_Metrics could use a refresh. very relevant for dash.js work today.

Live Streaming

  • p2p and
  • icecast pure http streaming server.
  • isn’t recommended for low-latency (<10s) due to http.
  • Limitations mostly attributed to client buffers
  • wowza http streaming server.
  • Live vs WebRTC?
  • sporting events vs video conferencing .. or hangout broadcast.
  • Push HLS s3 lower latency; akamai has larger latency.
  • yes you can go low latency with icecast
  • but you need to control pipeline down to the player ( network )
  • Define “low latency”
  • You can skip to latest by having a 404 segment.
  • people trying to use dash.js for live streaming ;)
  • Reach to devices
  • Europe about the specs
  • Left with both DASH and HLS.
  • ‘not that bad’
  • Lower quality audio streams help with low latency by having low network bandwidth fallback.
  • Dash.js don’t play out of sync content.
  • Some references:
  • SNTP considered harmful
  • Causes time discontinuity, breaks streams
  • Use proper NTP
  • Server side Ad stitching ( events cuePoints, id3 tags, VAMP xml events )
    • 3 headed: streaming server catch the cueTones
    • convert cueTones to respective packing live protocols
    • include service or player integration to pick up on packaged VAST definitions or consume out-of band service if player does not make cuePoints available to the player ( id3 tag ) for safari OSX. or iOS safari.

Action Items:

  • wowza to implement webRTC ingest.
  • investigate “low latency mode” for HTTP streaming.
  • interoperability for ad timing effort. ( VAST / VMAP / Live Ad Stitching )
  • follow up offline on ad timing -- kaltura, wowza, jw, icecast etc.
  • VAST updates -- visibility metrics? should add events ?
    • being discussed in IAB standard groups.
  • encourage.

JavaScript? based delivery

https://www.theoplayer.com - JS HLS

  • Theo Player hacks to swap video tags, which is supported on many platforms
  • Peer5 - p2p wrapper around XHR, or an “aggressive” replacer
  • Standard bridge between Flash and JavaScript?
  • Progressive MP4? via MSE
  • Control of buffer
  • QoS? metrics
  • mp4box.js - waiting for implementations and requests for improvement
  • MSE+HLS by video.js
  • video.js to move to MSE by default instead of Flash
  • Mobile implementations - questionable if JS will dominate, although there is preference to maintain js logic across different platforms (JWPlayer?, videojs)


  • CENC common encryption.
  • castlabs impressive player
  • wowza working on it; shares common encryption logic with playReady effort.

mixed content issues ( fixed? )