[Plib-users] RE: Problem with rotations

[McEvoy, N. <nic...@ds...>]

Ben Woodhead wrote:
>As you can see in pitch when it hits 90 on pitch, role also
>goes to 90. The one thing that I just noticed is that non of
>the other rotations get to exactly 90.0.
>Any ideas, thanks for all your help steve.

You really gotta take Steve's advice on this ... it is gymbal lock ... read
and understand those links he mentioned. :-)

I had the same trouble when I started with matrices ... it drove me crazy
(and still does) ... my model flipped all over the place when I hit 90 pitch
... I even tried Quats (eg. sgHPRToQuat) ... which (I thought) should avoid
the problem ... but unfortunately I couldn't get them to work either ... I'm
still not sure why !?

I now use ODE (a physics engine) and PLIB for the graphics.  ODE does all
the rotation work for me and avoids the 'gymbal lock' problem ... but before
I found ODE I did fix the problem you are having by doing the following
(notice I never let the pitch hit exactly 90) ...

// If the eyepoint it pitched by 90 degrees - then roll and heading do the
same
// thing and the whole thing goes crazy ... so ensure we never point exactly
90.
if (mPosition.hpr[1] == 90.0f)
	mPosition.hpr[1] -= 0.2f;
if (mPosition.hpr[1] == -90.0f)
	mPosition.hpr[1] += 0.2f;

// Update player hpr
sgMat4 POS, ROT;
sgMakeCoordMat4(POS, &mPosition);
sgMakeRotMat4(ROT, mDirChange);
sgPreMultMat4(POS, ROT);
sgSetCoord(&mPosition, POS);

I hope that helps.

Good luck !

Nick
http://members.ozemail.com.au/~ndmcevoy/