Re: slicing coordinates for multimedia

Hello Al, all,

Thank you for informing the WAI x-tech mailing list our discussions on 
URI fragments for MPEG-21.

At the last MPEG meeting, I showed your email to my working group.  
Subsequently, an Ad-Hoc Group was set up. Two of its mandates are:

        * Investigating solutions for an authoritative URI fragment 
identifier scheme for MPEG media types
        * jointly exploring conventions for describing slices of 
composites with MPEG-21, SMIL, X3D

As for the authoritative URI fragment identifier scheme for MPEG media 
types, the audiovisual addressing scheme proposed by CISRA that I 
mentioned to you earlier will be considered.

This AdHoc Group will work up to next MPEG meeting in October and I hope 
we will be able to have fruitful discussions for examining whether -- if 
I can borrow your expression-- an "agreement among multiple 
constituencies" is realistic and achievable.

I noted that the terminology you used in your email differs 
substantially from the terminology we use in MPEG.  Therefore, I would 
like to clarify two important points in your message.

As you know, we are particularly interested in a common convention for 
fragment identifier syntax and have been using XPath-like axes for 
expressing spatio-temporal fragments.  Is that similar to your view of a 
common convention?  Can I equate our view of generic axes for 
spatio-temporal fragments to one slice of composites as you put it?

I think your treatment of the virtual world as a model and breaking "the 
composite down, either componentwise or by slicing" is quite close to 
the notion of logical model MPEG-21 is currently debating.  A Logical 
Model of a resource provides an abstraction of the resource that is 
meaningful to the User of the resource.  Logical Models for audiovisual 
resources usually consist of a hierarchy of Logical Units and do not 
depend on how the audiovisual resource is physically stored.  A Logical 
Unit is a unit of an audiovisual resource such as a chapter of a DVD, 
that is semantically meaningful to the User of the resource.  Logical 
Units can be atomic or composed of other Logical Units.  MPEG has 
recognized that location schemes for fragments could profitably address 
Logical Units of audiovisual content.  Again, do you think this is in 
similar to your view?

Best regards
Myriam


Al Gilman wrote:

>
> There has been some discussion[1] on the uri list of "fragment"
> syntax for slices of multimedia composites, which the MPEG 21 working
> team had been considering expressing in #fragment syntax as with
> applications of XPointer.
>
> They may feel that the social side of establishing a community
> consensus across language proponent groups may be too daunting and
> just do something in the name of a specification to be used inside
> MPEG integration documents. If they do this, it could be a setback
> for the vision of the Web as one great commons, and at the very least
> a setback for disability access to information represented as
> multimedia composites[2].
>
> The alternative would be to have some sort of agreement among
> multiple constituencies such as MPEG 21, SMIL, X3D, and the like on a
> convention for expressing slices of composites (including reaching
> into constituents) in terms of the coordinates of the composite. And
> possibly what those default composite-frame coordinates are, but the
> latter might vary by integration format.
>
> This applies to multimedia canvases as contemplated by SMIL and CSS
> and more generally to virtual worlds[2] as contemplated by X3D. I
> believe that we can treat the virtual world as an absorbing model
> that covers the multimedia canvas. MMI and screen-reader worlds are
> possibly yet more general involving a sheaf of sub-worlds (view
> tuple, deck of WAP1 cards, for example).
>
> The accessibility argument is roughly that AT for VR and multimedia
> needs the capability to break the composite down, either
> componentwise or by slicing, until a slice is obtained that is
> conformable to an interaction world in which the user is comfortable
> and their motion within this world is securely under their control.
> Then they need view controls to sweep or navigate this slice around
> to cover the entire problems-space world. AT doesn't have the kind of
> resources to do this for multiple arbitrary conventions as to slicing
> coordinates in the composite experience world. But AT might be
> economic if there were a reasonable convention such that navigation
> and slicing in standard coordinates would be effective across all
> formats in this class.
>
> Al
>
> [1] Thread on uri at w3.org list
> http://lists.w3.org/Archives/Public/uri/2004Jul/thread.html#2
>
> [2] Workshop on access to visualization - please see my contributions
> http://www.w3.org/WAI/RD/2004/06/28-agenda.html
>
>
>
>

Received on Friday, 30 July 2004 01:10:27 UTC