I think I understand what you are saying - click on the object some 
distance from the center and it starts rotating in the direction, perhaps 
at a rate proportional to distance from the center? I haven't done this, 
but it should be straightforward. I like the virtual track ball because the 
original selection point stays under the pointer no matter where it moves 
(unless it goes outside of the sphere radius) and when the cursor returns 
to its original position, the object is back to its original pose. The 
problem with it, and this would not be a problem with what you propose, is 
that the object drifts in the axis around the vector defined by the 
selected point. It isn't bad, but it sometimes requires a reorienting of 
the object. Note that I don't expect to use this much, much because of this 
problem, and because I think the edit cube described later is far more 
elegant and controllable.

Regards,

David


At 03:12 PM 2/12/2004, you wrote:
>Hi David,
>
>Thanks for the link.  Quick question about your description of Virtus
>trackball... you described a method of rotating the object around an
>axis perpendicular to the screen.  What about a similar circular area
>perpendicular to the axis from the sphere center to the mouse pointer?
>This would allow the object to be rotated without moving the last
>point on the sphere selected by the mouse.  Have you tried this?  If
>so, how well did it work?
>
>Joshua
>
>
>
>On Wed, Feb 11, 2004 at 01:19:32PM -0500, David A. Smith wrote:
> > So don't give up yet. We are still working hard on this. In the meantime,
> > here is a link to the paper we presented in Kyoto at C5:
> >
> > <http://www.reed.com/tealand/lightsmith/04.01.01%20CroquetUIPaperV3.pdf>
> > The first half of the paper is a rehash of previous work with a bit more
> > depth. I think you will find the rest of it worth reading.
> >
> > Regards,
> >
> > David