Report on Television and the Web, Nice 29-30 June
Date: Tue, 07 Jul 1998 15:57:59 +0200
By: Lynda Hardman

The program for this event can be found at:
http://www.w3.org/Architecture/1998/06/Workshop/program.html


There were 2 groups of people at the workshop - those interested in 
hacking current standards to get web-content on a TV screen and those
interested in taking broadcast type material and making it web compatible.

What surprised me was that very little mention was made of SMIL - but I
guess it is still so new that not many people have looked at it.
One of the most substantial "HTML-hacks" was from Microsoft and co., so
they certainly wouldn't be recommending a SMIL solution. Their position
paper is:
http://www.w3.org/Architecture/1998/06/Workshop/paper30

There were, however, a handful of people, such as Warner, myself and the
W3C employees (also Glen Adams, who said he worked on Mosaic) who were
taking a higher-level view and thinking how the technological
infrastructure should progress further.


My talk (based on Lloyd's slides) went OK.  It followed on from the talk
Warner gave, 
http://www.w3.org/Architecture/1998/06/Workshop/paper38

but was even more esoteric for the audience comprising mainly hardcore
software engineers.  (I'm not inadvertently insulting Sjoerd or Jack am I??)

The favourite slide of my talk was the Lloyd/Jacco slide of HyTime, SGML,
Dexter, AHM, SMIL etc. and their placing on the two dimensions of hypertext 
functionality and presentation independence.  People
were clamouring for copies!  (OK - two people asked....)

One of the problems of relevance ofmy talk was that the examples were very
static (pictures and text).  With some video streams in there, the audience 
would have seen the connection better.  (Lloyd, maybe something to bear in
mind for the MM98 presentation.)

Apart from the plenary talk sessions there were two break-out sessions.

On the first day I took part in the workshop with Hakon-Lie(CSS2) and Dave
Raggett (HTML) and did my very best to be CSS2 friendly...
There were issues, however, which Hakon didn't quite grasp.
(Hmm, if only I could remember what they were - sorry.)
I think one thing I managed to get across is that CSS needs a temporal
layout section as well as a spatial one.

The second day I went to the "MPEG-4" workshop.  The MPEG4 guys had already 
given a talk, and explained that MPEG4 did everything you could want, and
their were tools to support it all.  The hour in the workshop was then
spent with the audience trying to establish just what exactly MPEG4 could
do.  Rob Koenen (KPN research, Leidschendam) and a guy from CTIT (France,
Oliver I believe his name was) wanted to get W3C to somehow "recommend"
MPEG4 as a standard.  Abramatic was very wary about it all, and wasn't even 
keen on having Philipp look into it (in fact, he suggested me as a
coordination person....).  There is now a web-mpeg4 mailing list that has
been brought into existence.  I should join it at some point....

http://garuda.imag.fr/MPEG4
http://drogo.cselt.it/ufv/leonardo/mpeg-4/mpeg-4.htm (doesn't work)
http://drogo.cselt.it/ufv/leonardo/

It is entirely unclear to me what MPEG4 has to offer to the Web world, but
perhaps in August/September we should find out. (E.g. pay a visit to Rob at
KPN.)  I had dinner with Jan van de Meer (Philips) and asked him what the
MPEG4 history was.  His summary was that the MPEG2 group had such fun
working together that they forgot to stop, but that the need for the
standard was not very clear.

MPEG4 has format for streaming content, a means of combining streams (and
making one stream hard sync'd with another), including VRML in a
presentation e.g. including 3D objects within 2D video (?).
They have a scene description in BIFF, which includes synchronization. It
is VRML like.
They can also do runtime selection of streams (a la CMIFed channels).
I was unclear about their timing issues and how you synchronized to
(named?) streams.
I think their only clock is the stream clock.  You then sync. to a stream.
I don't think they have a "document clock" as SMIL/CMIFed does.
The standardised MPEG4 format is binary (ugh).
They have linking, but what or how I do not know.
The MPEG-4 people (Rob and Olivier) wondered about defining a web-based
profile, so that MPEG4 for the Web would not use _everything_ that is
defined in MPEG4, thus allowing for lighter-weight applications.


So, I think I represented the group pretty well, and the direction our
(Lloyd/Jacco) work is going in is very much in parallel with a number of
the people there.  If anything, I was a little disappointed with the
inward-lookingness of the vision of the W3C people.  Philipp, our fearless
SYMM leader, is endowed with more vision than the others (was my feeling).
Not sure about Abramatic, he didn't say very much.

One of my summaries is that combining TV with the Web requires: streaming
media, temporal synchronization and atemporal composition. W3C has no
streaming.  RealNetworks has it, but is proprietory (this is where MPEG4
could play a role).  SMIL has temporal synchronization.  Atemporal
composition has still to be added.

Philipp (Hoschka) is now defining a list of areas which W3C should
address. (You should have received this already.)  We should give
constructive feedback.


Some things we could do/follow up:

*   We could have GRiNS generate MPEG4 (whatever MPEG4 is).

*   There were a number of interesting partners for an ESPRIT project, in
    particular Reuters, who have content and are interested in "quick and
    dirty" authoring to get relevant content to clients fast.

    The BBC were there too, but they were far more protective of their
    content.  And they are not interested in producing web pages for anyone
    outside of the UK.  If they were to get huge numbers of hits on their
    pages but none from within the UK then they wouldn't get support for
    the web site.

*   A potentially straightforward journal publication to produce
    would be a three-way comparison of SMIL, MHEG-5 and MPEG4 (assuming
    that finding out about MPEG4 is sufficiently interesting to make it
    worthwhile).

*   Warner and I promised each other to get together to define some future
    work for one or two AIOs.  (Warner has already sent me a mail....)


Should I send this to Warner too?


     Lynda