See also: IRC log
<trackbot> Date: 14 February 2013
<scribe> scribenick: mmartin3
<scribe> Agenda:
http://lists.w3.org/Archives/Public/public-tt/2013Feb/0037.html
Netflix presentation:
http://lists.w3.org/Archives/Public/public-tt/2013Feb/0036.html
anthony: Positioning and aspect
ratio are becoming a problem because of the content of what we
receive from providers.
... Slide 3
... Slide 4 authoring intent
... Slide 5
... Captions can be delivered and displayed on a different
aspect ratio.
... Slide 7 what is rendered isn't what the author
intended
... Rendering interpretations
... Slide 10 Disregard the 320x240px note
... We favor #4.
... Slide 10.
... Aligns to Proposal #1 in Slide 11.
sean: Would need to intend that there is centering in the display area as well.
anthony: Neither proposal clarifies this.
glenn: We don't have a model on
how to express authoring intent for registering the root
container extent with a related media object for display or
video.
... If we choose to handle, we have to look to this general
problem.
... Singer asked if transforming video resolution, why couldn't
you do captions?
... Origins match root container extent in examples. This is
different than we defined regions. Regions are designed
differently.
... Percentages to position regions according to a 4x3 area no
matter what the video resolution was. What happens in that
case?
anthony: If we knew aspect ratio, we could position regions on 4x3 aspect ratio. Issue is we don't know of the aspect ratio of the video at authoring.
glenn: Could have 4x3 and 16x9.
david: Captioning tooling have limitations. They typically don't use master video when generating captions.
c/genereating/generating
scribe: Use a proxy and the proxy may not be known and the video for delivery.
sean: Are they making decisions with what they want to align with video pixels or a general area.
david: Most of what we see is
intended for 4x3.
... Interpret intent.
... CEA-608 data 4x3
... SMPTE
... Conformance is widely varied. Give guidance to content
houses.
mike: CEA-608 didn't solve this
problem because it was 4x3 only. This is a hole in the end to
end processing.
... External data with aspect ratio relayed separately.
... DECE whatever you have done, stretch to width and height of
video. This is a deterministic option. CEA-708 SMPTE new
release sets aspect ratio as metadata. Don't know how to
generalize style element to handle.
sean: Pixels is based on XSL
which is the wrong measure and need something else for
consistency. Issues are related.
... No video object in TTML
... SDP-US does and maps to Proposal #1.
... TTML could be positioned to HTML5; many videos could be
present. There are different interpretations given the assets
involved.
... Add notes. Complete fix isn't found in this proposal.
anthony: We align to video.
sean: Unless you understand the
application and the pipeline, you can't express in TTML.
... Constraints could be applied in your implementation or in a
profile.
david: Our client may not be
doing presentation. We may deliver to a game console which
renders the captions.
... We have to be able to tell a client here is an asset and
know how to display.
... Working with device partners how to resolve how to render.
Prefer a standard guide.
... We can specify what is intended. Otherwise, we also need to
give client guidance on what to do. Different entities in the
pipeline.
mike: Make it better even if not
entirely.
... Use metadata.
... Skeptical to satisfy for all cases in TTML.
sean: Solved in SDP-US profile already.
mike: Clearly unambiguous in
DECE.
... Use SMPTE metadata that is defined?
anthony: Working on our own captions. Caption studios are creating content. Small % are CEA-608, -708.
david: Seeing a lot of CEA-608
and old EBU.
... A mess of caption world. Not designed for scale. Government
requirements are a stress on ecosystem.
... Large effort to author and manage.
glenn: If SMPTE asked for metadata, satisfy?
sean: Can use metadata
today.
... Has no semantic interpretation. There is no guarantee it
will be displayed in that way.
pal: Define metadata and author intent. Figure out how profiles use metadata is separate.
glenn: Won't define semantics for how a tool uses metadata.
sean: General solution is to have
a mechanism to reference where the existence of the region is.
Top Left isn't the same in 4x3 and 16x9.
... We don't have tools to accommodate. That would be part of
TTML 1.1.
... Immediately, if we take Proposal #1, we can't guarantee
players do the same thing.
anthony: Need long term
soluton.
... Netflix specific metadata tag.
sean: Use SMPTE metadata
tag.
... Doesn't have rendering model.
... Fold into Issue #179.
... Resolve in TTML 1.1.
Technical report: http://lists.w3.org/Archives/Public/public-tt/2013Jan/0006.html
mike: Doesn't include aspect ratio for CEA-708.
sean: Mike provide relevant data.
http://www.w3.org/AudioVideo/TT/tracker/issues/179
Issue Anthony How to specify aspect ratio to understand positioning that may apply for display or video.
Create Issue Anthony How to specify aspect ratio to understand positioning that may apply for display or video
pal: SMPTE when will we walk through SMPTE request? How do we handle from a process perspective?
sean: Put in semantic buckets to handle in TTML 1.1.
mike: Encourage vendors to look at a profile to implement captions. Look at SDP-US.
david: Ingest is challenging. We use a constrainted TTML set to deploy.
mike: Evaluate SDP-US.
david: Have own profile.
pal: No complete TTML
implementation everywhere.
... In practice is to follow a documented subset.
... Look at SDP-US and body of generated documents to see
gaps.
... Extended profile or SDP-US or other.
sean: Use for your use case.
david: Work with PLH on ensuring
Netflix are fully part of TTWG.
... Netflix has own profile for required and optional features.
We're not using pubic SDP-US.
pal: Stated requirements are that
the generated documents had to be able to play on clients not
under Netflix control. No implementation will do all of TTML.
Follow constrained profiles only. By not following one of
those, there will be clients that can't play Netflix
documents.
... Desire to use more standard one if we can get display issue
resolved.
anthony: Our profile is exactly a subset of that profile.
sean: Where are you in
conflict?
... Your content should render as well.
... Defined rendering.
trackbot, end meeting
<plh> Chair: Sean Hayes
<plh> Scribe: Monica Martin
This is scribe.perl Revision: 1.137 of Date: 2012/09/20 20:19:01 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/1/3/ Succeeded: s/generate/genereating/ Succeeded: s/generate/generating/ Succeeded: s/arae/area/ Succeeded: s/wya/way/ Found ScribeNick: mmartin3 Found Scribe: Monica Martin WARNING: Replacing list of attendees. Old list: +1.408.216.aaaa +44.118.900.aabb +1.408.335.aacc +1.858.847.aadd pal +1.417.671.aaee [Microsoft] glenn New list: +1.408.216.aaaa +44.118.900.aabb +1.408.335.aacc +1.858.847.aadd pal +1.417.671.aaee glenn Sean mmartin3 mike Anthony +1.408.515.aaff Default Present: +1.408.216.aaaa, +44.118.900.aabb, +1.408.335.aacc, +1.858.847.aadd, pal, +1.417.671.aaee, glenn, Sean, mmartin3, mike, Anthony, +1.408.515.aaff, David_Ronca, Plh, mijordan Present: +1.408.216.aaaa +44.118.900.aabb +1.408.335.aacc +1.858.847.aadd pal +1.417.671.aaee glenn Sean mmartin3 mike Anthony +1.408.515.aaff David_Ronca Plh mijordan Regrets: None Found Date: 14 Feb 2013 Guessing minutes URL: http://www.w3.org/2013/02/14-tt-minutes.html People with action items:[End of scribe.perl diagnostic output]