Re: Standardizing coordinate systems, units of measure

Eric Van Hensbergen (hensberg@escmail.orl.mmc.com)
Thu, 04 Aug 1994 15:36:37 -0400


In message <9408041201.ZM6121@krypton.engr.sgi.com>, "Gavin Bell" writes:
>On Aug 4, 11:48am, Linas Vepstas wrote:
>

Well, personally I like the idea of multiple coordinate systems and units
of scale, and just have a reference to which type the model uses in the model
description (this reference could be a URL-type reference all to itself).
Of course there would still have to be a standard 'conversion' unit of
scale that every unit must be able to convert to and from. Meters would
probably be cool for this.

As far as coordinate systems goes, the reference could designate which
axis is horizontal to the screen (left to right), which axis is verticle
to the screen (top to bottom), and which axis is depth (front to back). A
negative in front of the axis would reverse the direction.

Example:

coordinate_system {
units: 1000.0; // scale: 1 unit = 1000 meters (or a kilometer)
horizontal: +x // x goes from left to right
verticle: -y // y goes from bottom to top
depth: +z // z goes from out to in monitor wise
}

Then its just a matter of conversion.

Or, if we MUST go to a single coordinate system, my vote goes for
(Quoting from "Computer Graphics: Principles & Practice, Van Dam, Foley, etc.):
"The left handed system is used so that, with the eye or camera at the origin
looking toward +z, increasing values of z are farther away from the eye, x is
to the right, and y is up."

However, I still think coordinate systems should be definable for maximum
flexibility.

-Eric