From: "V�ctor" <[EMAIL PROTECTED]>
Sent: Monday, May 14, 2001 5:15 AM
> Here you can find a nice article about quaternions:
>
>
http://www.gamasutra.com/features/19980703/quaternions_01.htm
>
> Bruno Janvier ha escrito:
>
> >
> > Hi all, I want to use J3D to visualize the results of a
physics
> simulation about the attitude control of a satellite.I
compute the
> positions and the orientations of the satellite in a
geocentric
> coordinate systemwhere the z-axis is vertical (it is the
y-axis in
> the 3D space). I solved that problems with the positions by
using
> the matrix : 1 0 00 0 10 1 0 How can I do the same thing
with
> quaternions ?I tried the following matrix but it does'nt
look
> good: 1 0 0 00 0 1 00 1 0 00 0 0 1 THanks for your help.
The Bobick article remains one of the best and most concise
treatments on the basics on the web (with a nit or two to pick),
but (IMHO) it doesn't go far enough in its cautions about Euler
angles.
According to me, the first and best advice is express rotations
some (any?) other way. If you're stuck with them because of an
outside interface or If you're convinced that they are the
proper approach for your application, stay away from integral
multiples of 90 degrees, for any rotation _especially_ when
you're making up sample problems to explore whether your
understanding lines up with the math and with the geometry.
Start with small angles to verify how they work in your
application, then you can work out ways to trap the exceptions
they generate after you get the normal cases running.
Someday, there'll be dual quaternion transformations in
hardware, and old-timers will sit around saying things like,
"Remember transformation matrices?"
Cheers,
Fred
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
