Re: Office Space ;) and coordinate systems

Kevin Goldsmith (unitcirc@netcom.com)
Fri, 16 Sep 1994 16:15:39 -0700 (PDT)


> ::
> :: You're map analogy is reasonable.
> ::
> :: But what if you get maps from 10 different people and you want to paste
> :: them together to form one big map?
> ::
> :: Your job will be much, much easier if you don't have to look at all of
> :: those scales, figure out how they differ, and spend time at your Xerox
> :: machine enlarging or reducing them so they all fit.
> ::
> :: That's what we're trying to avoid by deciding on common units.
> ::
I haven't seen gavin's post above, so I'm going to respond to it
here. Umm, maybe I didn't state my point very clearly, because you seem
to be restating what I said before. I think I confused things by using
the word "scale" when I should have meant units. ahh. that's probably it.

>
> The 10 maps problem will exist in any case also.
>
> It depends on what you are going to use the scale for. I dont think this
> has been fully determined for VRML at this point.
>
> With the 10 maps or the overlapping maps, comparisons would have to be
> made between two/many datasets presumably obtained from different places on
> the internet. I presently see two possible reasons for this.
>
> 1) >merging< datasets together to create a combined world/scene.
>
> 2) >linking< datasets to travel from one to the other.
>
> If you have a VRML-standard scale, as suggested in both cases above, then
> linking from one set to another(camera movement) would require comparing
> them to determine how they relate in scale. The same would be true of a
> merging of two datasets. Comparisons would have to take place, right?
>
Wrong. that's the whole point of having VRML-standard units and
coordinate systems. There is no comparison necessary, because you know
exactly what you are getting when you encounter an object.

> For example, suppose you move from an ant-scale world to a human-scale
> world. Suppose VRML-standard scale is the same as metric. Ant-scale might
> have all objects rendered in millimeters whereas human-scale might have
> all objects rendered in centimeters.
>
I don't see your point here. A CG camera does not have bulk or
mass like a real camera does. It is essentially a point with a cone
sticking out of it. You can change the size of the cone if you wish, but
there is no need. There is no reason why I can't have a camera travel
between those two spaces without any changes to it whatsoever.

> Or, perhaps you have a mix of objects. Some are large enough to be in
> meters, some are so small as to require millimeter scale. Why state an
> object is 1/10 of a meter when centimeters might be a better scale?
>
So make the scale of your object in centimeters and then put a
scaling node at the top of your group converting everything to meters.
That way you get to define everything in centimeters, but the rest of the
world sees your object in meters and the viewer doesn't have to do that
conversion (or scale) for you.

> In terms of programming algorithms, I see no way around a basic metric
> conversion process that converts centimeters to millimeters, kilometers
> to meters, etc. in order to place objects within a coordinate system.
>
see my reasoning above

> If this is the case then why not extend the conversion processing to
> allow for any existing unit of measure? All that would be required is
> that each VRML object description state the nature of its units in
> some known scale. I am thinking of each object being something like
> a cube or cone or a more sophisticated 3D object of a stated size.
>
That adds a lot of unecessary and posibly time-costly work for
the rendering/viewing engines.

Kevin