Re: [Celestia-developers] New center command

[Christophe T. <ch...@te...>]

Le Jeudi 28 Ao=FBt 2003 19:26, Chris Laurel a =E9crit :
> >     // rotation1 along OP brings u to the SOP plane.
> >     // nSOP normal to SOP plane
> >     // nuOP normal to uOP plane
> >     Vec3f nSOP =3D v ^ w;
> >     Vec3f nuOP =3D u ^ w;
>
> I think that you need to normalize nSOP and nuOP . . .  If they're not
> unit vectors, the length of their cross product won't be the sine of th=
e
> angle between them (as you assume a few lines down.)

Of course!
But anyway even the axis of rotation is wrong...

> I'm off to work in a few minutes so I don't have time to think this
> through thoroughly, but . . .  I'd recommend using slerp to interpolate
> between the quaternions directly rather than converting to axis/angle a=
nd
> then back.

I'm not really used to quaternions yet. I thought of using slerp to=20
interpolate, but prefered to keep it simple until I got it working with t=
wo=20
stages.

> If you're going to use orbit, I think what you want to do is compute tw=
o
> orientations: q1 at t, and q2 at t + dt.  Then find the rotation q such
> that:
>
> q1 * q =3D q2
>
> q =3D ~q1 q2
>
> (where ~ indicates the conjugate, as in Celestia)

Quaternions obviously make things simpler...

> Then, call orbit with q . . .  I think that should work, though there m=
ay
> be an easier way to do all this that doesn't involve using orbit.

I wanted the movement to be the same than when using the RMB, that's why =
I=20
used orbit.

--
Christophe