Re: [MEDIA_PIPELINE_TF] ISSUE-34: ViewPort-Support

Le 11/08/2011 12:34, Silvia Pfeiffer a écrit :
> On Thu, Aug 11, 2011 at 7:35 PM, Cyril Concolato
> <cyril.concolato@telecom-paristech.fr>  wrote:
>> Hi Sylvia,
>>
>>> Can you fill the TextTrack object and the cues from a SVG? As long as
>>> you can make a mapping, it's possible. If the format doesn't fit with
>>> the elements, then it's an orthogonal concept that won't fit the bill.
>>
>> Reading http://www.w3.org/TR/html5/video.html#text-track it says:
>> "A text track consists of:"
>> * "The kind of text track": would apply to SVG tracks
>> * "A label": would apply to SVG tracks
>> * "A language": would apply to SVG tracks
>> * "A readiness state": would apply as well
>> * "A mode": same
>> * "A list of zero or more cues"
>>
>> A Cue is defined as:
>> * "An identifier": applicable to SVG tracks
>>
>> * "A start time"/"An end time":
>> As such SVG does not define frame-based content as Flash would do or other
>> subtitling formats, but you can define frame-based content such as:
>> http://perso.telecom-paristech.fr/~concolat/SVG/flash8.svg
>
> This is one of the major requirements.
Do you have a document describing those requirements ?

> Trust me - SVG won't fit the
> bill as a time-aligned data format.
General purpose SVG no but SVG with some constraints yes. It's not the perfect time-aligned format but it works. I've demonstrated that it's possible to fragment some SVG into frames, store them in MP4, stream them ... [1]

> And browsers won't implement it.
That me be true for now but you never know. They might be interested in it.

> But you can always try a JavaScript implementation to prove me wrong.
"implementation of what" ?

Regards,

Cyril
[1] http://biblio.telecom-paristech.fr/cgi-bin/download.cgi?id=7129
>
> Cheers,
> Silvia.
>
>
>>
>> * A pause-on-exit flag:
>> not sure what it means but seems applicable
>>
>> * A writing direction: when restricted to unidirectional text content, it
>> would be applicable but I don't see it in the TextTrackCue object
>> (http://www.w3.org/TR/html5/video.html#texttrackcue)
>>
>> * A size:
>> not sure I understand why it's not 2 numbers but it could match the SVG
>> width (also not in the interface)
>>
>> * The text of the cue:
>> if the SVG animation contains text, it could be passed, if not, could be an
>> empty string (or the content of a<desc>  element) if required.
>>
>> The active flag: applicable
>> The display state: applicable
>>
>> What I'm trying to say is that it seems very restrictive to limit the track
>> element to pure text tracks. It disables interesting use cases (synchronized
>> graphics overlay) while options to do it are there with SVG, already
>> implemented in browsers.
>>
>>>
>>> In my understanding, SVG has a complex DOM that goes far beyond what
>>> TextTrack is capable of representing.
>>
>> I don't get this point. The definition of a TextTrack element should not
>> care about the expressiveness of the language used to represent the text
>> content as long as this language can be mapped to the TextTrack features: if
>> it has graphics or not, if it can do bold/italic or not, if it can do bidi
>> text or not ...
>>
>>> So, I think it's too rich a
>>> format for the feature.
>>
>> "Oh I don't want your color TV, I just need a black-and-white one".
>>
>>>
>>> Of course you can always throw any format at a HTML element.However,
>>> if browsers don't support it, you can only deal with it through
>>> JavaScript - so it's not a standardised feature and not really
>>> relevant to the W3C.
>>
>> SVG is not any format. It's a W3C format. I think it's relevant.
>>
>> Cyril
>>
>> --
>> Cyril Concolato
>> Maître de Conférences/Associate Professor
>> Groupe Multimedia/Multimedia Group
>> Telecom ParisTech
>> 46 rue Barrault
>> 75 013 Paris, France
>> http://concolato.wp.institut-telecom.fr/
>>


-- 
Cyril Concolato
Maître de Conférences/Associate Professor
Groupe Multimedia/Multimedia Group
Telecom ParisTech
46 rue Barrault
75 013 Paris, France
http://concolato.wp.institut-telecom.fr/

Received on Thursday, 11 August 2011 20:35:22 UTC