W3C home > Mailing lists > Public > public-web-and-tv@w3.org > February 2020

Re: Why WebVMT Branched From WebVTT

From: Gary Katsevman <gkatsevman@brightcove.com>
Date: Thu, 20 Feb 2020 17:51:04 -0500
Message-ID: <CA+kYEUeJ9RXiye694=vQ5SK=85DYtY2zTXZptka1ROuSL6LMJw@mail.gmail.com>
To: Rob Smith <rob.smith@awayteam.co.uk>
Cc: public-web-and-tv@w3.org, Silvia Pfeiffer <silviapfeiffer1@gmail.com>
Hi Rob,

That sounds like a reasonable choice. I wonder if there are any interesting
takes away that you've learned working with WebVMT that may apply to WebVTT
as well? Any features in WebVMT that may make sense to bring back to WebVTT?
Does it make sense to update WebVTT in such a way that WebVMT can be an
extension of it rather than a fork, as opposed to outright merging the two?
For example, mark stuff like regions as not required for metadata use cases?

Thanks,
Gary.

On Thu, Jan 23, 2020 at 12:27 PM Rob Smith <rob.smith@awayteam.co.uk> wrote:

> On 22 Jan 2020, at 20:07, Silvia Pfeiffer <silviapfeiffer1@gmail.com>
> wrote:
>
> On a side note, why did you have to define a completely new format for the
> mapping use cases? Want it possible to define a new metadata type for track
> and reuse webvtt? I'm asking because webvtt was built with the idea of
> being extensible for all use cases for time aligned data. If there's a
> limitation to the extensibility, it should get addressed.
>
>
> Hi Silvia,
>
> Good to hear from you, and thanks for your question.
>
> I initially wanted to make a WebVTT extension, however soon felt that I
> was trying to bend the format in a completely new direction for the map
> interface, and beyond its original scope. Text and geotagged metadata share
> common cue properties, e.g. start time, end time and content, though are
> presented and rendered in very different ways and places - text versus
> graphics, video overlay versus map overlay. There's a clear use case for
> text with video, i.e. subtitles, but metadata use cases are far less well
> defined, which is reflected in the relative maturities of HTML’s
> TextTrackCue and DataCue elements.
>
> From a design perspective, I needed new features, e.g. the map block, and
> found existing ones were unnecessary, e.g. the region block and text
> formatting features. Having worked on standards' development before, I’m
> wary of feature creep and was concerned that a combined solution would
> create an unnecessarily-bloated format which would be complex to implement
> with minimal overlap between text and geotagged metadata features. Hence, I
> borrowed the common elements, i.e. cues and the file structure, discarded
> the text components and added a command structure to make a lean format for
> WebVMT instead. It may be possible to merge the formats back at a future
> date, though I still have reservations about this currently.
>
> I’m happy to have a chat with you about this further if you’re interested,
> and would also welcome your input to the WICG DataCue activity [1] and
> metadata discussions in Media Timed Events TF [2]. More details about
> WebVMT can be found at our website [3] and in the Editor’s Draft [4].
>
> I look forward to hearing from you.
>
> Rob Smith
>
> Away Team
> www.awayteam.co.uk
>
> [1] https://github.com/WICG/datacue/
> [2] https://www.w3.org/2011/webtv/wiki/Main_Page/Media_Timed_Events_TF
> [3] https://webvmt.org/
> [4] https://w3c.github.io/sdw/proposals/geotagging/webvmt/
>
Received on Thursday, 20 February 2020 22:52:09 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 20 February 2020 22:52:09 UTC