I think that in addressing the question about whether or not Z is up, the
central question has gotten confused.
We are talking about defining a space. VRML as a language will describe
the space. It is the client who will map this three dimensional space into
the perspective for the user. How a client interface deals with mapping
x,y,z into the perspective window shouldn't be defined by the space.
So, the idea of x being horizontally across the bottom IN THE SPACE, seems
to me to be fairly wacky. What if I turn ninety degrees to the right? How
does reality justify the previous x-axis declaration?
It is however a reasonable paradigm to expect that a large percentage of
the virtual worlds we build will loosely model an earth-like horizon-based
movement and awareness. That means we can expect to have an up and down
that are well defined. It's well accepted that up in such a space is the
positive z axis.
Which direction X and Y go is only relevant locally (who cares what the
other links define) or if you want 'rooms' to be related in a cartesian
fashion.
If walking forward (positive Y for instance) moves me into a link, There's
no inherent reason for me the space to guarantee walking backward would
bring me back from that link. I'm not talking about one way doors. I'm
saying that motion in the hyperlinked space doesn't necessarily map IN ANY
GUARANTEED FASHION to any other space, including the one I just left.
Everything is local, so x,y only matter within a space. Z is is only
important if you are using gravity to define up and down - consider how
you'd generically define Z in a weightless environment. You can't. It's
all arbitrary and local.
It's in the transitions from one link to another that different spaces will
be translated to the user coordinate system. Hence 'entry' points, which
need to define your entrance into the new coordinate system spatially and
rotationally.
In summary, the question about Y being up, or Z going into the monitor has
_nothing_ to do with the space. It's a client issue.
If we want to define some physical characteristics, then we could figure
out which way gravity goes, and would probably align the z axis to that.
Or we might want a "sun" which could give us east and west (x axis perhaps)
and north and south.
And lastly, we may even decide that magnetic north is an important concept
and allow people to align their x,y coordinates to magnetic north instead
of true north. These will be decided based on characteristics of a given
class of 'worlds' we might define and let people incorporate into their
pages. For instance, Earth would be described as above, but Joeland might
have a magnetic north aligned exactly with true south but no gravity. And
Places might be of the Joeland paradigm or the Earth paradigm, or anything
else, including maybe a world that has mapped a cartesion allocation for
rooms.
Since VRML is about the space, we shouldn't try to answer client questions
about how z maps for the monitor. X,Y,Z are arbitrary without physical
characeristics to define them.
Am I making any sense?
-joe
Joe Andrieu andrieu@alta.com
818-405-1812