0
\$\begingroup\$

I was implementing some quaternion rotations in too my infant 3d game engine and the rotations were really strange? Excuses my cluelessness.

Anyways this is what I have:

static XMVECTOR XMConvertToQuaternion(XMFLOAT3 axis, float radian)
{
    axis = XMNormalize3(axis);
    XMVECTOR tempAxis = XMVectorSet(axis.x, axis.y, axis.z, 1);
    XMVECTOR q = XMQuaternionRotationAxis(tempAxis, radian);
    return q;
}

Then later on down when I apply the quaternion to the world matrix I have:

XMMATRIX scaling = XMMatrixScaling(scale.x, scale.y, scale.z);
XMMATRIX rotateX = XMMatrixRotationX(radianRotation.x);
XMMATRIX rotateY = XMMatrixRotationY(radianRotation.y);
XMMATRIX rotateZ = XMMatrixRotationZ(radianRotation.z);

XMMATRIX rotateQX = XMMatrixRotationQuaternion(MathHelper::XMConvertToQuaternion(axis, radianRotation.x));
XMMATRIX rotateQY = XMMatrixRotationQuaternion(MathHelper::XMConvertToQuaternion(axis, radianRotation.y));
XMMATRIX rotateQZ = XMMatrixRotationQuaternion(MathHelper::XMConvertToQuaternion(axis, radianRotation.z));
XMMATRIX translate = XMMatrixTranslation(translation.x, translation.y, translation.z);
XMStoreFloat4x4(&mWorld, scaling*rotateQX*rotateQY*rotateQZ*translate);

I did get a rotation. However, the character is rotating very strangely.

\$\endgroup\$
0
\$\begingroup\$

the way to create your Quaternion rotations matrices (I assume from euler angles) look funny, I would expect you use unit vectors of each axis rather than axis

XMMATRIX rotateQX = XMMatrixRotationQuaternion(MathHelper::XMConvertToQuaternion(XMFLOAT3(1, 0, 0), radianRotation.x));
XMMATRIX rotateQY = XMMatrixRotationQuaternion(MathHelper::XMConvertToQuaternion(XMFLOAT3(0, 1, 0), radianRotation.y));
XMMATRIX rotateQZ = XMMatrixRotationQuaternion(MathHelper::XMConvertToQuaternion(XMFLOAT3(0, 0, 1), radianRotation.z));
\$\endgroup\$
0
\$\begingroup\$

On this line:

XMStoreFloat4x4(&mWorld, scaling*rotateQX*rotateQY*rotateQZ*translate);

does changing the order to:

XMStoreFloat4x4(&mWorld, translate*rotateQX*rotateQY*rotateQZ*scaling);

make a difference? If you meant to rotate about the origin then translate outwards then you need to be careful with your ordering, though I'm making assumptions about the XM library.

\$\endgroup\$
  • \$\begingroup\$ No, that just made it worse. Haha \$\endgroup\$ – Andrew Wilson Sep 30 '14 at 18:36
  • \$\begingroup\$ Try taking out the translation and scaling completely. Then test each axis separately if necessary. Results? \$\endgroup\$ – parar Sep 30 '14 at 19:27
  • \$\begingroup\$ I just got the same results. Must be how I got the quaternion. \$\endgroup\$ – Andrew Wilson Sep 30 '14 at 21:57
  • \$\begingroup\$ I don't know if you just copied your code wrong, but you are currently using the same axis for each quaternion rotation, which is not what you want to do. \$\endgroup\$ – parar Oct 1 '14 at 7:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.