Re: [media] Navigation of media by content structure (Issue-163)

Hi John,

Some replies inline.

On Thu, Jun 30, 2011 at 5:20 AM, John Foliot <jfoliot@stanford.edu> wrote:
> Silvia Pfeiffer wrote:
>>
>> This is to address Issue-163, which is due by 1st July, see
>> http://lists.w3.org/Archives/Public/public-html/2011May/0428.html.
>>
>> I've put together some thoughts in the wiki page at
>> http://www.w3.org/WAI/PF/HTML/wiki/Media_Navigation on how navigation
>> of media could be addressed.
>>
>> There are basically two approaches, each with their own advantages and
>> disadvantages. They together provide quite some flexibility and
>> neither really needs any changes made to HTML5, unless I'm missing
>> something. They may, however, need changes to other technologies along
>> the solution chain, including screen readers and file formats.
>>
>> Please share any ideas for different solutions that I may have missed.
>
> Hi Silvia,
>
> Thank you for putting together this wiki page.
>
> I'm looking at this, and I'm perhaps just not catching this right, but
> your proposed code sample seems to only include one level of navigation.
> In the first example, could it also be replicated like this?:
>
> ...

[.. removed irrelevant stuff..]
This is the html-mapped cue text content:

>  getCueAsHTML(): "Talk on WebVTT
>      <ul>
>        <li><? target='timestamp' data='00:00:47.600'>Impact of Captions
> on the Web</li>
>        <li><? target='timestamp' data='00:01:50.100'>Requirements of a
> Video text format</li>
>        <li><? Target='timestamp' data='00:02:30.000'>The Web VTT Format
>                <ul>
>                        <li><? target='timestamp'
> data='00:03:33.000'>Simple WebVTT file</li>
>                        <li><? target='timestamp'
> data='00:04:57.766'>Styled WebVTT file</li>
>                        <li><? target='timestamp'
> data='00:06:16.666'>Internationalized WebVTT file</li>
>                </ul>
>          </li>
>      </ul>"
> ...
>
> To create a multi-level navigation scheme? (Realizing that indentation is
> of course for reading clarity only)

Yes, indeed, this is indeed the way in which a multi-level navigation
scheme would work.


> How would this map back to TTML? The code example for TTML does not show
> any Human readable chapter/subchapter names.

Yes it does: "Index point 1.1.1" etc are the subchapter names. I will
create an example that has chapter names, too, and add it to the wiki.


> *************
>
> As I continue through your proposals, you then suggest spreading the
> chapter and timings over 2 files:
>
>  <video src="video.ogv" controls>
>    <track src="webvtt1.vtt" kind="chapter" label="level 1 navigation">
>    <track src="webvtt2.vtt" kind="chapter" label="level 2 navigation">
>  </video>
>
> ... where you illustrate chapter <nav> in WebVTT, but not for TTML. I'd
> like to hear what Sean might suggest.

TTML would work the same way. I can add an example there, too.


> But then, I got to thinking...
>
> If you can suggest mixing both <nav> time markers and <ul><li> structure
> in one WebVTT file,


There are no <ul><li> markers in the WebVTT file - they are only in
the HTML mapped version the cue text.


> and/or reference 2 WebVTT files to carry different
> levels of navigation, then why not just go the distance and create a
> navigation file using getCueAsHTML() and list markup,

Yes, that's exactly what I did.

> at which point you
> have abstracted the human readable hierarchy to a document that maps back
> it's time-stamps to the relevant and reported (albeit different reporting
> scheme) time markers in the transcript file(s).

Yup, that's what I've done, with the time markers being presented as
ProcessingInstructions.


> In other words, it's the getCueAsHTML() that is the key (I think).

Yes.


> Simply
> put something like this (not sure of syntax):
>
> <!DOCTYPE html>
>  getCueAsHTML(): "Talk on WebVTT
>      <ul>
>        <li><? target='timestamp' data='00:00:47.600'>Impact of Captions
> on the Web</li>
>        <li><? target='timestamp' data='00:01:50.100'>Requirements of a
> Video text format</li>
>        <li><? Target='timestamp' data='00:02:30.000'>The Web VTT Format
>                <ul>
>                        <li><? target='timestamp'
> data='00:03:33.000'>Simple WebVTT file</li>
>                        <li><? target='timestamp'
> data='00:04:57.766'>Styled WebVTT file</li>
>                        <li><? target='timestamp'
> data='00:06:16.666'>Internationalized WebVTT file</li>
>                </ul>
>          </li>
>      </ul>
> </html>


No, you're mixing the content of a single cue with the concept of a
document. I only showed how a SINGLE cue's content is mapped into
HTML. Multiple cues go into different TextTrackCue [1] objects, which
are collect in the browser in a TextTrackCueList [2] and hooked into
the video element through a TextTrack [3] as part of the list of
TextTrack elements in a MediaElement [4]

[1] http://www.whatwg.org/specs/web-apps/current-work/multipage/the-iframe-element.html#texttrackcue
[2] http://www.whatwg.org/specs/web-apps/current-work/multipage/the-iframe-element.html#texttrackcuelist
[3] http://www.whatwg.org/specs/web-apps/current-work/multipage/the-iframe-element.html#texttrack
[4] http://www.whatwg.org/specs/web-apps/current-work/multipage/the-iframe-element.html#htmlmediaelement

We still need the concept of cues in the browser to deal with navigation.



> ...save as nav.html, then write:
>
>  <video src="video.ogv" controls>
>    <track src="webvtt1.vtt" kind="chapter">
>    <track src="nav.html" kind="navigation" label="Chapter Navigation">
>  </video>

Ah, no, you don't need that second nav.html - it's all already in the
webvtt1.vtt file. I'll put a bit more into the wiki page to explain.


> I know, I know, I hear somebody screaming SMIL.... (Argghhh), but...

SMIL would not help here.

Cheers,
Silvia.

Received on Wednesday, 29 June 2011 23:12:21 UTC