Re: Interest for quaternions (and octonions)?

From: Hubert.Holin_at_[hidden]
Date: 2001-01-02 18:59:16

Paris (U.E.), le 03/01/2001

--- In boost_at_[hidden], "Reid Sweatman" <borderland_at_u...> wrote:
> Same problem we need to solve in games.  Is it possible (assuming you don't
> end up licensing it) to generalize the mechanism handling your orientation
> constraints?  That would really be useful.

        Well, actually, something like five or six years ago, I tried to 
sell myself using that among other things, to Alias/Wawefront. That did 
not work out. So I tried to give it for *free* to MetaCreations (if 
they agreed to keep it available to the mac, and bar it to M$-stuff ;-) 
). Since my political requests were impractical, they just said they 
might licence it for all platforms, for money, and then they, as a 
company, commited what I believe is corporate suicide... I did that bit 
of research with a colleague who now would really like to make a paper 
out of it, so I guess what you want will come to pass in the not too 
distant future, unless some company want to keep an edge over the 
competition for a few years (hint hint...).

        The original code is in C (the only common programming language 
between my colleague and me), and fairly ad'hoc as representations go. 
Before building a purely C++ solution, I am looking for some reasonably 
"standard" representation of rotations (and other geometry elements) 
which would work well with packages such as TNT. I would also use this 
representation to add a few things to the quaternions I submited here.

        That being said, what generalisation would you be interested in?

                Hubert Holin
                Hubert.Holin_at_[hidden]

> Reid Sweatman
> Software Engineer
> 
> > -----Original Message-----
> > From: Hubert.Holin_at_B... [mailto:Hubert.Holin_at_B...]
> > Sent: Thursday, December 21, 2000 3:26 PM
> > To: boost_at_[hidden]
> > Subject: [boost] Re: Interest for quaternions (and octonions)?
> >
> >
> > Paris (U.E.), le 21/12/2000
> >
> > 	Actually, the file I uploaded is one end product of soome work I
> > did with a colleague about orientation interpolation (quaternions are
> > superior to Euler angles, among other reasons because they are not
> > succeptible to "gimbal lock", which is a degeneracy in the
> > representation for special values of the orientation).
> >
> > 	The problem we solved (though this is *not* included in the public
> > file), is that of orientation under constraint, i.e. if we want to
> > emulate a real film camera, the "up" must not "wobble" between
> > keyframes. I tried to licence that part, so far with no result :-<  .
> >
> > 	One thing which *is* in that file is the exponential of the
> > quaternions, which I believe, is a result which I invented (more
> > precisely, the closed formula giving the result is my invention, not
> > the existence of the exponential (of one stripe or another) on the
> > quaternions...). This can be used to achieve interpolation of
> > orientations (in 3d-space), but with the "woble"...
> >
> > 	A last note: I have some ideas about cryptography and quaternions,
> > but the implementation I proposed is not the best suited to this
> > problem.
> >
> > 		Hubert Holin
> > 		Hubert.Holin_at_B...
> >
> > --- In boost_at_[hidden], Lutz Kettner <kettner_at_i...> wrote:
> > > Hi,
> > >
> > > there is interest in the graphics community for quaternions as
> > > a _good_ representation of 3d rotations and for the ArcBall user
> > > interface interaction technique for rotations. They are also very
> > > nice for interpolating between two different orienations of a
> > > 3d object. I don't know about numerics and quaternions.
> > >
> > > Lutz Kettner
> > >
> > > ----------------------------------------------------------------------
> > > UNC Computer Science                       email: kettner_at_c...
> > > CB 3175, Sitterson Hall                    phone: [phone] x7759
> > > Chapel Hill, NC 27599-3175, USA            fax:   [phone]
> > > ---------------------------------------------------------------------
> > -
> >
> >
> >
> >
> >
> >
> >