In a possible past, Eric Kimminau said:
> Another nod towards the need for some type of "HUD" to determine
> orientation. We are adding complexity and bandwidth with this idea. I
> think it may have to be said up from on your VRW entry point whether or
> not the world is a single horizon, fixed path progression or it is a
> "fly-through" world and require specific browser capabilities in order to
> navigate it.
I don't see the necessity. Whether the world is planar-based or
relatively arbitrary can be abstracted by rule of thumb from scene
data (the user can always designate a reference plane to the browser
if it's confused or can't guesstimate one for itself).
The "glass helicopter" model should be able to traverse _any_ form of
world. It doesn't even need a reference plane - that's handy for
orientation, but the "ladders" combined with a translation-only
movement model give enough orientation information to make do.
There are no immediate implications for VRML that I can see (arguably
therefore the discussion should be taken elsewhere; it's just that
there are browser writers reading this) - it all happens in the
browser itself. It would be nice to have reference plane or COG
information in the world, but not necessary. Maybe something to
consider for 2.0.
The extra overhead involved in calculating and drawing the "ladders"
is, I would suggest, insignificant compared to the overall scene
generation. I would hope that any browser would be able to offer a
selection of traversal models - if the chopper _is_ too
compute-intensive for the platform, the user should pick a simpler
one.
-- [email protected] Hyphen home page: http://www.hyphen.com/ [email protected] And mine: http://www.hyphen.com/html/jonsg/ PGP key available on request Opinions here are mine, not Hyphen's