well....
things are getting mopped up over the next few days,
so bring your waders with you...
indeed. we're proud of the work, it looks good, the content is there and
files are too large. improvements, however, are under way.
we've written some scripts that reduce the file sizes by as much as 75% (a
good indication of the amount of datafat on these things) but the rendering
time is still a primary concern of ours. if we trim the last 6 floating
point units this _will_ reduce rendering time (less so than download time),
but i think we're pretty well mired in some 1.8fps rate for now... stay
tuned.
a few offthecuff thoughts on what we could have done *better*:
1. not used as much architecture
(like walls, for example).
2. avoided the use of as many polygons
(the fourth floor is 14,000 (!@#$!) poly).
3. focused more on light convey the feeling of the environment.
(interest is elicited with mystery, not additional models)
4. concentrated more on interface than appearance -
put more thought into the *informational hierarchies*
of what we are doing. it's frustrating to have this MIMEtype
degrade - moving from vrml to html, or from vrml to gif is
not the best way to approach this. what is the best way to
house the volumes of HTML in the libraries of VRML...?
it's still consuming our brains.
what are the qualitative relations the two protocols have..?
5. made everything link. ...we tried it on the fourth floor....
this is not myst, and it's not html - we wanted there to be
positive feedback from wherever you pointed your attention.
unfortunately, webspace lights EVERYTHING up as you move the
cursor over it. it looks like crap. so much for that.
i, at least, hope that this will change in the browsers.
the easy solution is for the _cursor_ to change - leave the
models themselves alone.
a few offthecuff thoughts on what we could have done *worse*:
1. used texture maps.
we avoided this since it takes so long to
get the final picture and it turns rendering into an
accelerated aging process for the poor sap that's navigating.
2. split the floors into multiple files.
this was tricky since we wanted each floor to have its own
navigation mode. splitting things would have presented
severe interface problems.
3. see 5, above. pro and con, this one was a stickler.
the interface of an almost entirely iconic world is pretty tricky.
this has been the most difficult hurdle.
this is still panning out.
we're only now ready to start on this project.
the doing was the learning.
the lack of sleep was our penance.
================
please review the files, take _anything_ you want off of arc, use it to
your own twisted ends, and let us know what you think of it all.
this was a highly collaborative and relatively complex process that has
vacuumed the lifeforce of 15 of us. the people that have dealt with this
are listed on vrml.arc.org.
they each deserve massive kudos and a hotel in tahiti.
the files aren't finished yet - we're trying to clean it all up, trim it
down, install a live video feed (fishcam style) and other bells and
whistles. give us a few more days to get all the links in place. the
festival opens june 4 in Los Angeles at the Variety Arts center on 940
Figueroa street. since this project is largely about the interface between
the virtual and the real it would be worth your time to drop in and see the
final work (imo). almost all of us that built this will be there.
bye.
6
/-^^^^^^^- /\--\ Mark S. Meadows
| [email protected]
@ @
) OO
| .l__I i_.. http://www.arc.org/
i ./ \ | i ! http://vrml.arc.org/
_ ... UJ . UJ UJ ._ http://www.arc.org/who/pighed/