Orientation = Quaternion( EulerAngle + MouseDeltas ); vs. Orientation = Quaternion( MouseDeltas )*Orientation; Shouldn''t these two be the same? Damn if i didn''t forget in the 4 months since I was replying to these types of posts... Thank you for your bandwidth. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~Succinct Demos Online~

Gimble Block?

The big problem brewing that I see, is the mousedeltas are inherently 2D while the orientation is 3D. i.e. left/right mouse-view-scrolls aren't the same if you're look forward vs. up.

I use the second method.

Magmai Kai Holmlor
- The disgruntled & disillusioned


Edited by - Magmai Kai Holmlor on May 22, 2001 12:22:08 AM

Ok, that's what I was thinking... Problem is, when I used method 2, my direction normals end up pointing in opposite directions every time... It flashes 180 extra degrees every render... Doesn't make any sense... I've even tried using a few other substituting other people's quaternion math for mine, character for character, such as Jeff Lander, David Eberly, and Solbiet Void. Still, nothing is working correctly.

Here is what I'm doing...

// method 1: // find mouse deltas MouseDelta = s_MousePosition - s_MousePositionLast; s_MousePositionLast = s_MousePosition; // convert mouse deltas to euler angles Swap( MouseDelta.x,MouseDelta.y ); s_Euler += MouseDelta*Vector3d( -ROTATION_SPEED,ROTATION_SPEED,0 ); // calculate new orientation s_Orientation = Quaternion( s_Euler );// method 2 // find mouse deltas MouseDelta = s_MousePosition - s_MousePositionLast; s_MousePositionLast = s_MousePosition; // convert mouse deltas to euler angles Swap( MouseDelta.x,MouseDelta.y ); s_Euler = MouseDelta*Vector3d( -ROTATION_SPEED,ROTATION_SPEED,0 ); // calculate new orientation s_Orientation *= Quaternion( s_Euler );

the only difference is that in the first one the s_Euler has a += and the s_Orientation has an =, whereas in the 2nd s_Euler has an = and s_Orientation has a *=.

I'm 90% sure the implicit math here works... Like I said, i've substitued those other big named guys' multiplication/euler conversion/axis angle conversion, and everything was the same...
Furthermore, method #1 works just as would be expected... gimbal lock and all.

what the heck is up w/ #2?

arrgh.
thanks, mag.

--succinct

Edited by - Succinct on May 22, 2001 12:39:22 AM