<cpn> scribenick: cpn
<kaz> Agenda
Giri: We have a change to the agenda today, Rob Smith could introduce WebVMT, related work.
<kaz> [fyi, the email address of the group list is public-web-and-tv@w3.org ]
<kaz> Media Timed Events use case document
Giri: We have 5 open issues and 3 pull requests
<kaz> issues on GitHub
<kaz> PRs
Giri: #6 for use cases, are there more to be added?
<kaz> issue 6
<kaz> scribenick: kaz
Chris: We don't have a good
description on the purpose, in the document
... There are several use cases we can add, e.g., from the 3GPP use
cases,
... such as dynamic quizzes, surveys, additional web content
related to a programme.
Giri: OK
Chris: I'd like to add some of them, if we're allowed to.
<cpn> scribenick: cpn
Mark: The most common use cases I hear about are: ad insertion messages, and enhanced TV (red button).
Giri: I'll assign this to myself
Nigel: There's an implication with
in-band events, that the event changes the media that the user is
getting.
... A model that changes the media stream that's being consumed.
Need to consider what the UX is for that, e.g, how the back button
works?
... We've talked about ad synchronization already. But also,
regional services, e.g., a nation-wide news programme with segments
that are regional.
... After those, you jump back to the main programme. From a
streaming point of view, you want the minimum number of encodings,
and smooth playback.
... Something we could consider as a use case here.
... The distinction is whether you are switching the main programme
stream, or accessing additional media.
Mark: Regarding meeting schedules,
when we started this we didn't want to go beyond TPAC for this
work.
... For example, emsg in CMAF drives some urgency to get this
in.
... I'm concerned that we don't have the browser representatives on
the call.
... I'd like to see us move to the WICG, we know the subject matter
and use cases, we don't want to miss the opportunity.
... We have to go to the browser people, in WICG, start a topic in
Discourse, then open a repo there and move the document over
there.
... We'll get more participation there, and things should move
faster.
Rob: I'm proposing a break-out session at TPAC, for video metadata cues including localization, I'm trying to get the browser people involved too, to support this.
<kaz> scribenick: kaz
Chris: We could pretty quickly submit
something to the Web Incubator CG.
... The emsg support is a concrete requirement that we could
progress,
... while the IG continues to look at some more broader use
cases.
Giri: OK
<cpn> scribenick: cpn
Giri: We're still on track to complete by TPAC
<tidoust> [I note the recommended steps to propose work to the WICG: https://github.com/wicg/admin/]
<tidoust> [Repo migration happens later on, and requires an Intent to migrate: https://wicg.github.io/admin/intent-to-migrate.html]
Chris: I could take an action item to open the Discourse discussion.
Mark: We just want to write a couple
of paragraphs, and reference the document as work in
progress.
... My impression is that the browser vendors will respond, as it's
on people's minds.
Giri: There's editors notes in the doc, so they'll see it's WIP, shouldn't be a barrier.
Chris: I'll draft something and share with you.
Giri: https://github.com/w3c/me-media-timed-events/issues/5 terminology
Chris: The issue mentions "media timeline" I think we can close.
<kaz> 2. Terminology
<kaz> scribenick: kaz
Giri: https://github.com/w3c/me-media-timed-events/issues/3
DAInty
... There are 2 pull requests relating to this
Chris: Yes, #12 adds a description of DAInty based on discussion from the last TF call.
https://www.w3.org/2018/08/20-me-minutes.html
Chris: And #13 summarises the requirements for an emsg / DASH event API.
Giri: OK. It's hard to preview using
the diff, so let's merge these and review.
... regarding issue #1, we don't really have to care about this
ourselves,
... this will be up to the spec editors, as things progress.
Chris: OK.
<cpn> scribenick: cpn
Rob: I'm Rob Smith from AwayTeam
Software, Invited Expert in the Spatial Data on the Web Interest
Group.
... WebVMT is aimed at the mobile video market, dashcams, drones,
etc.
... Proprietary formats in use for dash-cam data, speed,
acceleration, and geolocation.
... These are embedded into the MPEG files, and needs a proprietary
video player.
... Standards exist within MPEG, there's format specific data in
MPEG.
... I want something for the Web, that can be used with WebM and
Ogg video.
... We have a format based on WebVTT. It's an annotated map, can
share location in a web friendly format.
... Can rapidly index video archives: does this geolocation appear
in this video archive? Useful for web crawlers and search
engines.
... Drones as sensor platforms.
... What browser support is required? Pass the track data, generate
metadata cues. I'm currently puzzling over this.
... Is WebVMT data a special case, or is there a generic
approach?
... Pass as text or as a structured object with a specific
interface?
... This is the main topic for my proposed TPAC discussion.
... http://webvmt.org gives a clear
overview of the concept and technical demos.
... We're currently in incubation in the W3C strategy funnel.
... It's monitored by the Spatial Data on the Web IG at W3C, and
OGC.
... Next steps are evaluation and then chartering a group for Rec
track.
... I'm interested in identifying communities who can use this.
Write the format more formally for TPAC.
... Want to find key players to standardise, discuss applications
for it.
... Paritipating in the OGC testbed activities.
<kaz> scribenick: kaz
Chris: With the DataCue API for emsg
events, the message has blob for payload,
... so it's up to the application to know how to extract the
data.
... Is there any schema description?
Rob: It's on my list to do.
<cpn> scribenick: cpn
Kaz: Thank you for presenting. I used to work for the geolocation WG. Could this handle acceleration and orientation as well?
Rob: Yes. Orientation is on the list
of features to include, from an augmented reality point of
view.
... Direction, pan, tilt, etc. Superimposing information on top of
video requires orientation.
... For acceleration, the dashcam market is a good example.
... The format is XML-like.
Kaz: The geolocation work is now in the Device and Sensors WG, you might want to look at that.
Nigel: I suggest it might be more
useful to look at the API side of things than the serialization
format.
... There are other formats than WebVTT, others more XML-like
... Lack of a schema makes WebVTT is difficult to validate.
... Having a model you can express as an API may be more
useful.
... For example, extending the TextTrackCue with additional
metadata.
Mark: I want to underline what Nigel said, the W3C's main concern is the API between applications and the browser.
Rob: That's one of the drivers for WebVMT, to get away from encodings into specific media formats.
Nigel: You may want to choose a data model and API rather than a serialization format.
<MarkVickers> The other thing I'd add is you should add a section on privacy & security. Geolocation is considered very private data.
Giri: Thank you Rob, and all for participating. If we feel this should be added to our to document, please add an issue.
Chris: Next call is October 15th, the week before TPAC.
<kaz> [adjourned]