Re: Standardizing coordinate systems, units of measure

Linas Vepstas (linas@innerdoor.austin.ibm.com)
Thu, 4 Aug 1994 11:48:07 -0500


> Date: Wed, 3 Aug 1994 10:49:21 -0700
> From: "D. Owen Rowley" <owen@autodesk.com>
> Subject: Re: Standardizing coordinate systems, units of measure

[... vitriolic stuff deleted ...]
Geez, Owen, cool down!

> > We should also decide which way is up. I like the OpenGL coordinate
> > system, which has X increasing to the right, Y up, and Z coming out of
> > your monitor.
>
> I've been trying to argue for abstractions that would allow local users to
> use what they are used to in filling out the details of the abstracted values.
> whats wrong with up down left right forward and backward yaw pitch and roll.

The point is that vrml will have constructs like
{point 0.3 5.4 2.2}

Is the first number up, the second left, and the third backward?
This should not be a big deal. So propose:

0) vrml version 1.0 supports 3D cartesian coordinates.
(future versions of vrml might support cylindrical, spherical, parabolic,
hyperbolic, projective, spinorial, twistors, SO(3,1), whatever. but not
version 1.0)
1) first number is called x, second is called y, third is called z
2) the coordinate system is right-handed i.e. x cross y = z.
3) as in grade school, where your sheet of graph paper lies on your desk,
horizontally, parallel to the ground, x is to the right, y is forward
and z is up.

4) A "group" is an atomic vrtp (vr transfer protocol) "page". The vrtp
server delivers it in an atomic way. Its a single, indivisible chunk.

5) The size of the group is measured in meters. The orientation is as 0-3
above.
6) designers working with a different orientations/dimensions should
supply a matrix at the head of thier "group" that converts from thier
favorite coord system, to the above. Ditto for working in millimeters,
whatever.

I do not beleive that any of the above should be controversial. We haven't
gotten to any of the hard parts yet.

--linas