W3C home > Mailing lists > Public > public-media-fragment@w3.org > May 2011

Re: a question about dimensions...

From: Silvia Pfeiffer <silviapfeiffer1@gmail.com>
Date: Sat, 7 May 2011 09:01:47 +1000
Message-ID: <BANLkTikeeSW5GOY8vfYNm=WJfUYtz9wppg@mail.gmail.com>
To: David Singer <singer@apple.com>
Cc: Davy Van Deursen <davy.vandeursen@ugent.be>, Media Fragment <public-media-fragment@w3.org>
On Sat, May 7, 2011 at 2:33 AM, David Singer <singer@apple.com> wrote:
>
> On May 5, 2011, at 21:50 , Silvia Pfeiffer wrote:
>
>> On Tue, Mar 22, 2011 at 6:25 PM, Davy Van Deursen
>> <davy.vandeursen@ugent.be> wrote:
>>> Hi David,
>>>
>>> On 21/03/2011 15:55, David Singer wrote:
>>>>
>>>> Guys
>>>>
>>>> I am not sure that 'named' is really a dimension.
>>>>
>>>> I understand focusing on time; focusing on a spatial area; and focusing on
>>>> tracks.  These are indeed orthogonal axes.
>>>>
>>>> But you use names to focus on a section, which seems to be a named
>>>> time-subpart.
>>>>
>>>> Specifically, can I use names for subsetting in any of the dimensions?  If
>>>> I have a track called 'fred', are these equivalent?
>>>>
>>>> http://www.example.com/thing#track=fred
>>>> http://www.exampke.com/thing#id=fred
>>>>
>>>> Would it not be cleaner to allow selection by name on any of the three
>>>> axes (i.e. named timed periods, named spacial regions, named tracks)?
>>>
>>> To me, 'named' can be seen as a kind of virtual dimension, which points to a
>>> combination of any of the three "real" dimensions (i.e., time, spatial, or
>>> track). The mapping between a named fragment and a combination of time,
>>> spatial, or track is left to the application.
>>>
>>> So for example: #id=fred_low could correspond to
>>> #t=10,20&track=low&xywh=20,20,30,30. It is of course not required to combine
>>> all dimensions: #t=30,40 could for example correspond to #id=blooper.
>>> Similarly, #track=fred could correspond to #id=fred if this is defined
>>> within the application; but as I said, the mapping is out of scope of the MF
>>> spec.
>>
>> I typically use #id to reference chapters, which are a subpart of the
>> time dimension.
>>
>> I'm currently dealing with a demo where I use WebVTT chapters to
>> navigate a video. It would be really nice if the browser would just
>> understand video.ogv#chapter2 and I wouldn't have to use script to set
>> the currentTime to the chapters2 start and catch the ontimeupdate
>> event to test if I've already gone beyond the end of the chapter2 to
>> stop playback.
>>
>> OTOH, this feature is not urgently needed compared to the temporal and
>> track dimensions. If I could just convert the label to a time range
>> through script and then do the media fragment addressing with #t, that
>> would help a lot, too.
>>
>> Silvia.
>
> Bring back cue ranges!  It would be so easy if one could set up cue ranges that matched the chapters, and then use the enter/exit events to set the labels and so on!

That's being done with TextTrackCues now:
http://www.whatwg.org/specs/web-apps/current-work/multipage/video.html#texttrackcue
. They have onenter and onexit events. The named fragment URIs are
still useful to further simplify that.

Cheers,
Silvia.
Received on Friday, 6 May 2011 23:02:53 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:52:46 UTC