FilmEd application / ZAnnot discrepencies

Hi,

As a part of DSTC's FilmEd project, we've been working on a collaborative
video discussion tool.  One of the requirements has been subjective
annotations at various degrees of granularity (shots, frames, sub-frames).
And as usage alongside conferencing tools is imagined, users should be
able to annotate with video or audio clips, captured from the session.

Fortunately Annotea fit the bill.  In the process we've built our own
client library (in C#) and taken advantage of the context field to locate
regions of the source video.  Video/Audio is handled via URIs and letting
an embedded browser interpret them.

The need to run easily on Windows led us to ZAnnot, and for the most part
things have been very succesful - thanks Brent.  However, we found a few
places where 0.4rc1 seemed divergent from the spec:

  - it looks for the argument w3c_replyTree, instead of w3c_reply_tree

  - Requests for w3c_replyTree=X appear to ignore X and return
    all replies reguardless

  - a small patch to AnnotRequestParser.py was required to correctly
    record plain text (non-xhtml / non-uri) annotations.  This problem was
    reproducable with Amaya.  My patch is not ideal but I can provide it
    to help with a better solution (or disprove that there is a bug)

Also some things I would like to note about the spec:

  - Appendix A does not describe how Algae Queries are to be transmitted.
    I gathered from annotest.w3.org that they are to be URL-encoded into
    a w3c_algaeQuery argument.

  - Section 2.2 demonstrates a query response using <r:Description>,
    while servers tend to return <a:Comment> or <a:Annotation>.  While
    either would seem valid, this was initially confusing.

FilmEd will be on display next week at APAC '03 and we hope to make it
publicly available shortly after that.

thanks,
michael


Relevant URLs:

http://staff.dstc.edu.au/mjh/images/FilmEdDemo.jpg
http://metadata.net/filmed
http://www.apac.edu.au

Received on Wednesday, 24 September 2003 01:51:15 UTC