> > >> up when I first come in. BUT, my whole point is that if I want to look
> > >> at a world sideways, I should be free to do so.
> >
> > No way! We will not tolerate drunken navigators in Cyberspace. We've
> > cleaned up what we think is indecent*, now we'll clean up the drunk
> > and disorderly :-)
>
> I want to be the first to open a virtual bar. Coming soon to a site near you.
> Serious. I'm going to do it. I was looking for a test scene.
Since "up" is purely a human concept of orientation within a gravitational
field, the concept of up really only applies to virtual worlds which desire
to produce this conceptual effect on the viewer. Obviously two virtual
spaces which "border" each other, and which the authors desire to provide
the appearance of a seamless interface betwsen the two worlds should use the
same coordinate system and/or gravitational orientation. However if author
1 prefers to model in the +z is up system, and author 2 only has tools which
treat +y as up, then author 1 could use Gravity { 0.0, 0.0, -9.8 } while
author 2 could use Gravity { 0.0, -9.8, -0.0 } and the two worlds would
provide consistent orientation. If you are cesating a space station, you
could use Gravity { 0, 0, 0 }.
The Gravity characteristic could apply to the entire world, or be overridden
by a Gravity characteristic attached to a volumetric object like a cube, thus
providing a different orientation while the viewer (camera) is located within
the volume of the object. The issue of modeling spaces large enough to
produce their own gravitational field, such as the entire earth, where the
field can no longer be considered a simple line vector, could be solved by
allowing gravity vectors to contain functions of the spacial position.
I was hoping I would be able to set the gravity, in my bar (modeled with
z up) to ...
Gravity { 0, 0, sin(z)+sin(y) }
That should keep the place hoppin' ;)
>
> > >> All my worlds have Z as up, so I'm coming from the same side that says
> > >> that Y should not always be up.
> >
> > Seriously now, if you want to use other's objects and cesate objects for
> > use to others, then it is easier if we think Y is up.
>
> OK. I will go with the agreed convention. I will also adjust my modeller
> so that it resets up to be Y if VRML is the default target.
>
I agree that browsers for the current spec should follow the spec.
My proposition was for future versions of the spec.