[jugglemaster] Dynamic scaling?

[Chunky Kibbles]

On Sat, Sep 02, 2006 at 11:58:24AM +0900, Per Johan Groland wrote:
> On 9/2/06, Chunky Kibbles <chunky at icculus.org> wrote:
> >Heya, I saw something about dynamic scaling added to jmlib, and, uh,
> >this sounds like exactly what I wanted for torque.
> >
> >Is there now an easy way to get an absolute position of balls on a known
> >scale, rather than translating them to fit into a window?
> 
> Perhaps the name I chose was a bit misleading. The new "dynamic"
> scaling works by keeping jmlibs internal window size at its native
> 480x400 at all times, scaling the pattern to match the actual window
> size. This works much better than the old method.

Ah, that makes sense.

<snip>

> If you need float coordinates, you won't be able to use the new
> scaling method. I was thinking of using the now defunct Juggler class
> for adding float coordinates.  This class could support various ways
> of scaling the coordinates to fit into different coordinate systems
> (for instance, flipping the Y axis).

That's kinda what I'm after...

> It should also be quite easy to scale the coordinates so that the ball
> radius will always be the same, is it this kind of absolute
> coordinates you were thinking of?

Kinda.

What I want is to know that the hands and head are always on the same
effective scale and in the same exact place, so I can apply a constant
transform matrix to my X,Y coordinates (ie, the head, hands, ball) to
render them in a different [3d] world.

What I'm looking for is kind of a normalised pattern, where the hands are
always moving over the same area, and the height of the balls just goes
up as there are more of them. The head would *always* be in the same
place and the same size.

I guess the simplest way to look at it is this; currently, I have a
little guy in a 3d world. I can't just scale the pattern and slap it in
front of him - and I'm lazy, and want to be able to :-)

Gary (-;