These are the notes from a meeting held on Mondy 19 June 2000 to discuss
the group's research direction(s), (just before FrankN's house warming
party and the birth of Julian Pemberton Kerssens...)

with Craig Lindley, FrankC, FrankN, Lloyd, Lynda, Steven

It is mainly a list of topics in no particular order and with no committments.

Content management
------------------

One of the major problems for multimedia content creators is how they keep
track of all their pieces of media, and how those pieces are used within a
presentation.  Many authors end up with files with names like
"scene1pic2.jpg", since any better way of storing media requires too large
an investment in effort - not to mention its use with the system in
question.

One serious option to look at is whether we can come up with a front end
for MMBase that would allow content creators to effortlessly put things
into the system, attach as many or as few labels/attributes/keywords to
them as they wish and still be able to access them from the standard tools
as if they were files.

If we were to do this properly it could fit in nicely with MMBase and score
on developing an interface for media aquisition and maintenance.  It should 
also have lots of links (excuse the word) with MPEG7 annotations and ways
of annotating (but we provide an interface that hides all the nasties) and
the types of annotations we could be providing.  The latter two directions
could be done without building any sort of tool.

If we approach it wrongly, then we will end up programming a bad version of
a picture database, with no added value.  This is something that no-one (in
the group) is even thinking about yet, but may be a fruitful direction.

Reasons for making the annotation tool are that we get a much better feel
for how much work it is and how one should be expected to annotate.
(Marc Davis did lots of video annotation - and however cute it looked it
must have been lots of hard work.)

Narrative authoring
-------------------

Idea two would be to provide authoring support at a higher level.
Authoring support for storyboard.  (A sort of "Storyspace" for multimedia.)

My (personal) approach (and not that of Lloyd's I may add) is that we are
trying to automate the creation of hypermedia presentations.  Lloyd's view
has always been that we are supporting the author, but at a different
level. I think I am bending to Lloyd's point of view.

So, an approach would be to build a an envrionment where the author can
create and view, a possibly empty, storyboard, and then fill it in at
his/her leisure. (E.g. see ACM MM9x system.) Where our system should differ 
is in the creation of non-linear stories and manipulation of the storybord
at a high level.  Something beyond GRiNS, beyond individual templates.  And 
preferably also allowing multiple "what if's" in the storyboard.  (A CVS
for hypermedia presentations  ;-)

FrankN mentioned lingo scripting in Director.  One thing we might like to
look at is what functionality is provided in new lingo and determine
whether we can replace it with higher level creative tools or declarative
models.

FrankN suggested he could find an example of a storyboard from Braunschweig
university. I want to carry out an anaylsis of what video/film producers 
put into their storyboards, either explicitly or implicitly, and then
whether we could provide concrete support for this -  first in an authoring
system and then later in a more automated version.  (There must be some
form of narrative hidden among the characters.)

Because I started talking about rhetoric, and because Craig thought I meant
RST, we had a quick clarification about the "meta concept"
(syntacomatic?!).  There is the narrative, and the narration of the
narrative.  The first is ethereal (fabula?  See FrankN's thesis?) and the
second is what the author presents to you. Either of these may be
interactive or not (i.e. there could be branches, different points of view,)
and either may be dynamic (when the reader starts reading it changes as
they go - so the presentation may be a linear multimedia presentation, but
the end is determined during the session).  This gives then 4 different
cases. 

Craig pointed out that a "real" film storyboard looks very like a comic.
Documentaries and news tend not to have storyboards.


Inspiration sources
-------------------

Craig asked me what I thought was inspiring for the research.

Maja - GoTo0 would be great, if it were not VR, but more multimedia
(i.e. flat screen, and more easily translatable into XML, so to speak).
http://www.cwi.nl/~maja/GoTo0/cave.html

Bruno Felix - The End of TV as we know it.  The fact that they are
producing content for both TV and the Web - right from the beginning.
http://www.vpro.nl/data/2166028/aflevering.shtml?2468171

Marc Davis's thesis. FrankN's thesis.

Scott McCloud's "Understanding Comics"


Notes from typing up time:
--------------------------

One topic we didn't discuss was streaming semantic
annotations. Over to FrankN:

    Streaming is still an important issue and we have to investigate it at
    some point in time. The problem here is twofold: - what needs to be
    done with respect to the size and handling of metadata being streamed
    with the data in real-time applications (e.g. live broadcasting)? One
    question here is how a setup [set-top?] box reacts on the metadata.
    All that is clearly an application oriented research field.

    - what does it mean to have a varying amount of metadata in interactive
    environments (very systems oriented questions, such as quality of
    service, bandwidth, etc.). Not that we are very much interested in
    these questions, but they will have an impact on our understanding of
    how metadata should be represented to allow our high level models to be
    mapped to low level features and to keep the overall architecture still
    manageable.

    In conclusion, I think it is not one of our major research concerns but
    surely an aspect we will be confronted with again and again. At least
    we should have an idea about the available literature.


Another thing not mentioned is my affinity to modelling.  For example, the
AHM never incorporated interaction in the hypermedia document, and SMIL
Boston does include interaction, but I'm not convinced about the underlying
model.  This will no doubt distract from the main line, but my fingers are
itching....


     Lynda