I need to rotate bones of skeleton, i have already the quaterinion corresponding for each joints; and i am confused when it comes on rotating.
Skeleton to move is my opengl scene i need to move.
My problem is that i can't rotate the joint; Can anyone Help
Bellow is my code
//i evaluate each joint to get the translation and rotation.
void Node::EvaluatePoi(std::vector<POI> pois, const Vector &par_pos,
const Quaternion &par_rot, Vector Node::*world_pos,std::vector<Arti> joints)
{ Vector poi=this->PoiVec ;
Quaternion rot;
if (pois.empty()){
this->*world_pos= this->rest_position ;//OFFSET
rot= this-> rest_rotation ;//identity
}else{
if(this->name=="Hips")
{
this->*world_pos = eval_instant_positionPOI(poi);
rot= this-> rest_rotation ;// do not rotate
}else if(this->name=="LeftUpLeg")
{
this->*world_pos = this->rest_position;// set to OFFSET
rot= this-> rest_rotation ;// do not rotate
}else if(this->name=="RightUpLeg")
{
this->*world_pos = this->rest_position;
rot= this-> rest_rotation ;
}
else
{
this->*world_pos= this->rest_position;
rot= eval_instant_rotationPOI(joints);
}
}
//Applying transformation on the global position with rot =qparent * qchild
(this->*world_pos).rotate(par_rot);
this->*world_pos += par_pos;
rot = par_rot * rot;
// draw joint's subhierarchy
for (int i = 0; i < n_children; i++)
child[i]->EvaluatePoi(pois, this->*world_pos, rot, world_pos,joints);
}
EDIT:
//here i get the local rotation of each joint, after that create quaternions equivalent to individual Euler rotations and then compose to one rotation
Vector x_vector(1.0, 0.0, 0.0),
y_vector(0.0, 1.0, 0.0),
z_vector(0.0, 0.0, 1.0);
Quaternion Node::eval_instant_rotationPOI( std::vector<Arti> joints)
{
Quaternion roto;//= new Quaternion();
Quaternion sample;
double t= 0.02;
Vector v;
Vector Euler(0,0,0);;
string x =this->name;
if(x== "Head"){
Euler=GetEulers(joints,JOINT_HEAD);
}else if(x== "Neck"){
Euler=GetEulers(joints,JOINT_NECK);
}
else if(x== "LeftUpArm"){
Euler=GetEulers(joints,JOINT_LEFT_SHOULDER);
}
else if(x== "RightUpArm"){
Euler=GetEulers(joints,JOINT_RIGHT_SHOULDER);
}
else if(x== "LeftLowArm"){
Euler=GetEulers(joints,JOINT_LEFT_ELBOW);
}
else if(x== "LeftHand"){
Euler=GetEulers(joints,JOINT_LEFT_HAND);
}
else if(x== "RightLowArm"){
Euler=GetEulers(joints,JOINT_RIGHT_ELBOW);
}
else if(x== "RightHand"){
Euler=GetEulers(joints,JOINT_RIGHT_HAND);
}
else if(x== "Hips"){
Euler=GetEulers(joints,JOINT_TORSO);
}
else if(x== "LeftUpLeg"){
Euler=GetEulers(joints,JOINT_LEFT_HIP);
}
else if(x== "RightUpLeg"){
Euler=GetEulers(joints,JOINT_RIGHT_HIP);
}
else if(x== "LeftLowLeg"){
Euler=GetEulers(joints,JOINT_LEFT_KNEE);
}
else if(x== "LeftFoot"){
Euler=GetEulers(joints,JOINT_LEFT_FOOT);
}
else if(x== "RightLowLeg"){
Euler=GetEulers(joints,JOINT_RIGHT_KNEE);
}
else if(x== "RightFoot"){
Euler=GetEulers(joints,JOINT_RIGHT_FOOT);
}
Quaternion qx(x_vector, (Euler.x ));
Quaternion qy(y_vector, (Euler.y ));
Quaternion qz(z_vector, (Euler.z ));
sample = qz * qy * qx;
roto= slerp(qTemp, sample, t);
qTemp=roto;
return roto ;
}
/*here i multiply the joint and its parent to get the Euler Angle ; is it necessary to convert to
Euler Angle?/
Vector Node::GetEulers(std::vector<Arti> joints, const int idx) {
// Get the quaternion of its parent.
Quaternion q_parent;
Quaternion q_current;
if (idx == JOINT_TORSO) {
q_parent.identity();
}
/////
{
q_parent = Quaternion(joints[parent_joint_map[idx]].quat.x,
joints[parent_joint_map[idx]].quat.y,
joints[parent_joint_map[idx]].quat.z,
joints[parent_joint_map[idx]].quat.w);
}
// Get the quaternion of the joint.
q_current = Quaternion(joints[idx].quat.x, joints[idx].quat.y,
joints[idx].quat.z, joints[idx].quat.w);
// Calculate the relative quaternion.
Quaternion q_delta = quat_left_multiply(q_current , quat_inverse(q_parent));
Vector angle = euler_from_quat(q_delta);
// cout<<this->name<<" "<<angle<<" ";
return angle;
}
Quaternion quat_left_multiply(Quaternion l, Quaternion r) {
Quaternion q = {r.w * l.x + r.x * l.w + r.y * l.z - r.z * l.y,
r.w * l.y + r.y * l.w + r.z * l.x - r.x * l.z,
r.w * l.z + r.z * l.w + r.x * l.y - r.y * l.x,
r.w * l.w - r.x * l.x - r.y * l.y - r.z * l.z};
return q;
}
Vector& Vector::rotate(const Quaternion& q)
{
Quaternion p(x, y, z, 0.0f);
Quaternion qc(q);
qc.conjugate();
Quaternion pp(q * p * qc);
x = pp.x;
y = pp.y;
z = pp.z;
return *this;
}
Rotating a quaternion is actually multiplying a quaternion by another. Given the quaternion qA representing the current rotation of an object and qB the quaternion representing the amount of rotation to apply (to add) to this object, the resulting new rotation of this object is computed as follow (pseudocode):
qA = qA * qB;
Alternatively, you can apply (add) this rotation in what is called "object" or "local" transformation space by swapping operands:
qA = qB * qA
Each joint should hold (usualy as class member) a quaternion representing its current rotation in local space. This is probably what you already done. If you want to apply a rotation to that joint, then you simply need multiply the joint quaternion, by another quaternion representing the amount of rotation to apply. A quaterion rotation method can be like this (pseudocode):
Joint::Rotate(const quaterion& amount, bool local)
{
if(local) {
this->rotation = amount * this->rotation;
} else {
this->rotation = this->rotation * amount;
}
this->rotation.normalize();
}
That is all you need for the rotation part, nothing else. After that, you will need to convert the joint quaternion to a rotation matrix, so to be combined with the other joint transformations (translation, scale, whatever). Here is one implementation of the quaternion to rotation matrix conversion (pseudocode):
Matrix3 QuaternionToMatrix(const quaternion& q)
{
float x2 = q.x + q.x;
float y2 = q.y + q.y;
float z2 = q.z + q.z;
float xx = q.x * x2;
float xy = q.x * y2;
float xz = q.x * z2;
float yy = q.y * y2;
float yz = q.y * z2;
float zz = q.z * z2;
float wx = q.w * x2;
float wy = q.w * y2;
float wz = q.w * z2;
Matrix3 m; //< 3x3 matrix
m[0] = (1.0f - (yy + zz));
m[1] = (xy - wz);
m[2] = (xz + wy);
m[3] = (xy + wz);
m[4] = (1.0f - (xx + zz));
m[5] = (yz - wx);
m[6] = (xz - wy);
m[7] = (yz + wx);
m[8] = (1.0f - (xx + yy));
return m;
}
What you may finally need is to input rotation using Euler angles instead of quaternion values. Indeed, Euler angles are easier to handle and understand when it come to apply a rotation in a human point of view. In this case, you'll need to convert the input Euler angles to a quaternion. Here is one possible implementation of Euler angle to quaternion conversion:
Quaternion EulerToQuaternion(float x, float y, float z)
{
float sx = sinf(x * -0.5f);
float cx = cosf(x * -0.5f);
float sy = sinf(y * -0.5f);
float cy = cosf(y * -0.5f);
float sz = sinf(z * -0.5f);
float cz = cosf(z * -0.5f);
Quaternion q;
q.x = sx * cy * cz + cx * sy * sz;
q.y = cx * sy * cz - sx * cy * sz;
q.z = cx * cy * sz + sx * sy * cz;
q.w = cx * cy * cz - sx * sy * sz;
return q;
}
Related
I've been working on a project for some time and Needed something that could from a Vector3 representing rotation in the XYZ axis make Forward, Right and Up vectors. I was looking through a lot of stuff and after some time I figured out I had to implement Quaternions (I have my own Math Libary but this same thing happened with glm) and Here is the code for calculating the Forward Vector: (quaternion is the rotation Quaternion member in my class and Quaternion::Euler is a static function that returns a Quaternion from Euler Angles)
quaternion = Quaternion::Euler(rotation);
Vector3 ret = quaternion * Vector3(0.0f, 0.0f, 1.0f);
when the rotation is 0, 0, 0 the function returns 0, 0, 1 as it should, but if I try something like 0, 180, 0 it should return 0, 0, -1, but instead I get -8.74228e-08, 0, -1. After some investigation I figured out that the Quaternion::Euler function returns a Quaternion where the w part is messed up. In the case where the rotation is 0, 180, 0 the Quaternion the Quaternion::Euler function returns is 0, 1, 0, -4.37113883e-08 which is almost exactly half of the random number the Forward functions returns. Here is Quaternion::Euler:
float x = Radians(euler.x);
float y = Radians(euler.y);
float z = Radians(euler.z);
x = x / 2;
y = y / 2;
z = z / 2;
return Quaternion(cos(z) * cos(y) * sin(x) - sin(z) * sin(y) * cos(x), //X
cos(z) * sin(y) * cos(x) + sin(z) * cos(y) * sin(x), //Y
sin(z) * cos(y) * cos(x) - cos(z) * sin(y) * sin(x), //Z
cos(z) * cos(y) * cos(x) + sin(z) * sin(y) * sin(x));//W
and Honestly, I stole this function from an article of a guy that was making his own Math Engine, in his case this seemed to work. Here is the Quaternion Vector Multiplication function, that I "borrowed" from the Unity Implementation: (in this case it's inside the Quaternion struct so this is a pointer to the quaternion from the Quaternion Vector multiplication)
inline Vector3 operator*(const Vector3& other) {
float x = this->x * 2.0f;
float y = this->y * 2.0f;
float z = this->z * 2.0f;
float xx = this->x * x;
float yy = this->y * y;
float zz = this->z * z;
float xy = this->x * y;
float xz = this->x * z;
float yz = this->y * z;
float wx = this->w * x;
float wy = this->w * y;
float wz = this->w * z;
Vector3 ret;
ret.x = (1.0f - (yy + zz)) * other.x + (xy - wz) * other.y + (xz + wy) * other.z;
ret.y = (xy + wz) * other.x + (1.0f - (xx + zz)) * other.y + (yz - wx) * other.z;
ret.z = (xz - wy) * other.x + (yz + wx) * other.y + (1.0f - (xx + yy)) * other.z;
return ret;
}
Does anyone know what might be wrong ? I tried to do this with glm:
glm::quat quat(glm::vec3(glm::radians(rotation.x), glm::radians(rotation.y), glm::radians(rotation.z)));
glm::vec3 v = quat * glm::vec3(0.0f, 0.0f, 1.0f);
but it's the same thing, the vector is the same and the quaternion is the same too, I've been reading into things a lot about this and couldn't find a fix, always when I tried to search implementation for the Quaternio::Euler function it just came up with how to use a math library. It would be best if the solution wouldn't require me to use glm, because I have to use my own Math Library, but honestly I will try anything to at least understand what is wrong.
I am writing a program with OpenGL/GLUT using the fixed function pipeline (I know, I know, it's university). I've written the Quaternion class from scratch with help of other implementations and the internet and it essentially works fine, but it's rotating along every axis in the opposite direction to what I thought it would.
I thought about posting this on the Math stack exchange, but given it's OpenGL/GLUT I thought it would be better understood here.
The axis below are: green -> Y, red -> X, blue -> Z. The darker sides are the positive directions. I've rotated it a little to show the positive Z axis. The axes are the world coordinate axes.
I have defined pressing "a" as a positive rotation in Y. The right hand rule states that this is a counterclockwise rotation in the Y axis. The image shows the rotation of the cube after pressing "a". As you can see, it has rotated clockwise. This occurs for every axis.
My Quaternion class:
#define _USE_MATH_DEFINES
#include <cmath>
#include "Quaternion.h"
#include "Utility.h"
Quaternion::Quaternion() : X(0), Y(0), Z(0), W(1) {}
Quaternion::Quaternion(Vector3D axis, float angle) {
float mag = Vector3D::magnitude(axis);
angle = utility::toRadians(angle);
float sine = sinf(angle * 0.5f);
// Divide by magnitude for pure quaternion
X = axis.X * sine / mag;
Y = axis.Y * sine / mag;
Z = axis.Z * sine / mag;
W = cosf(angle * 0.5f);
}
Quaternion::Quaternion(float x, float y, float z, float w) :
X(x), Y(y), Z(z), W(w) {}
float Quaternion::magnitude(const Quaternion &q) {
return sqrtf(q.X * q.X + q.Y * q.Y + q.Z * q.Z + q.W * q.W);
}
Quaternion Quaternion::normalise(const Quaternion &q) {
float mag = magnitude(q);
return Quaternion(q.X / mag, q.Y / mag, q.Z / mag, q.W / mag);
}
std::array<float, 16> Quaternion::toMatrix(const Quaternion& q) {
float X = q.X;
float Y = q.Y;
float Z = q.Z;
float W = q.W;
return {
1 - 2*(Z*Z + Y*Y), 2*(X*Y - W*Z), 2*(Z*X + W*Y), 0,
2*(X*Y + W*Z), 1 - 2*(X*X + Z*Z), 2*(Y*Z - W*X), 0,
2*(Z*X - W*Y), 2*(Y*Z + W*X), 1 - 2*(X*X + Y*Y), 0,
0, 0, 0, 1
};
}
Quaternion Quaternion::conjugate(const Quaternion& q) {
return Quaternion(-q.X, -q.Y, -q.Z, q.W);
}
Quaternion operator*(Quaternion lhs, Quaternion rhs) {
return lhs *= rhs;
}
Quaternion& Quaternion::operator*=(const Quaternion& rhs) {
float rhsX = rhs.getX(), rhsY = rhs.getY(),
rhsZ = rhs.getZ(), rhsW = rhs.getW();
Quaternion q;
q.X = W * rhsX + X * rhsW + Y * rhsZ - Z * rhsY;
q.Y = W * rhsY - X * rhsZ + Y * rhsW + Z * rhsX;
q.Z = W * rhsZ + X * rhsY - Y * rhsX + Z * rhsW;
q.W = W * rhsW - X * rhsX - Y * rhsY - Z * rhsZ;
*this = q;
return *this;
}
// Quaternion->Vector multiplication is not commutiative, must be Q*V
Vector3D operator*(Quaternion lhs, Vector3D rhs) {
Quaternion pure = Quaternion(rhs.X, rhs.Y, rhs.Z, 0);
Quaternion right = pure * Quaternion::conjugate(lhs); // v * q-1
Quaternion left = lhs * right; // q * (v * q-1)
return Vector3D(left.getX(), left.getY(), left.getZ());
}
float Quaternion::getX() const { return X; }
float Quaternion::getY() const { return Y; }
float Quaternion::getZ() const { return Z; }
float Quaternion::getW() const { return W; }
std::ostream& operator<<(std::ostream& ostream, Quaternion& q)
{
ostream << q.X << " " << q.Y << " "
<< q.Z << " " << q.W << " " << std::endl;
return ostream;
}
Pressing "a" (and similar) calls:
void GameManager::handleKeyboardInput() {
// ...
if (keyboard->isPressed('a')) {
ship->rotate(Axis::y, Direction::positive, dt);
}
// ...
glutPostRedisplay();
}
Which, when the object updates itself in the display function, calls:
void Ship::rotate(const Axis axis, const Direction direction, const float dt) {
int sign = direction == Direction::positive ? 1 : -1;
float speed = sign * ROTATION_SPEED;
if (axis == Axis::x) {
rotation = Quaternion(Vector3D::right(), speed * dt) * rotation;
}
else if (axis == Axis::y) {
rotation = Quaternion(Vector3D::up(), speed * dt) * rotation;
}
else if (axis == Axis::z) {
rotation = Quaternion(Vector3D::forward(), speed * dt) * rotation;
}
}
Where rotation is the object's current rotation stored as a Quaternion, left multiplied so it's local coordinates.
The Vectors are:
Vector3D Vector3D::up() { return Vector3D(0, 1, 0); }
Vector3D Vector3D::right() { return Vector3D(1, 0, 0); }
Vector3D Vector3D::forward() { return Vector3D(0, 0, 1); }
And finally the object is drawn in the main display loop:
glPushMatrix();
glMultMatrixf(Quaternion::toMatrix(rotation).data());
glColor3f(1.0, 1.0, 1.0);
glutWireCube(8);
glPopMatrix();
The easy solution is to just flip the signs of my rotations, but I'd rather know what's wrong. Thank you.
Edit: For what it's worth, I can confirm that if I hold "a" such that the cube is rotated 45 degrees (so a little more rotated than the second image), my cube's rotation (x, y, z, w) = (0, 0.389125, 0, 0.918102), which agrees with this page if I set y = 1 and the angle (radians) to 0.785398 (45 degrees). So the final rotation quaternion is correct, but my cube still rotates in the wrong direction. This makes me think there is something wrong with my code.
This is the final quaternion rotation matrix after holding "a" until it rotates 45 degrees clockwise:
{0.705871, 0, 0.708341, 0, }
{0, 1, 0, 0, }
{-0.708341, 0, 0.705871, 0, }
{0, 0, 0, 1, }
Which according to the same site is { [ 0, 1, 0 ], 45.1000806 }, which is what I want, but still the rotation is clockwise, not anticlockwise.
You build your matrix in row-major order, but glMultMatrixf expects a column-major matrix. Transposing a rotation matrix is equivalent to inverting it, i.e. rotating in the opposite direction.
To fix it, either build your matrix in column-major order, transpose it, or use glMultTransposeMatrixf.
I cannot understand the math behind this problem, I am trying to create an FPS camera where I can look freely with my mouse input.
I am trying to rotate and position my lookat point with 180 degrees of freedom. I understand the easier solution is to glRotate the world to fit my perspective, but I do not want this approach. I am fairly unfamiliar with the trigonometry involved here and cannot figure out how to solve this problem the way I want to...
here is my attempt to do this so far...
code to get mouse coordinates relative to the center of the window, then process it in my camera object
#define DEG2RAD(a) (a * (M_PI / 180.0f))//convert to radians
static void glutPassiveMotionHandler(int x, int y) {
glf centerX = WinWidth / 2; glf centerY = WinHeight / 2;//get windows origin point
f speed = 0.2f;
f oldX = mouseX; f oldY = mouseY;
mouseX = DEG2RAD(-((x - centerX)));//get distance from 0 and convert to radians
mouseY = DEG2RAD(-((y - centerY)));//get distance from 0 and convert to radians
f diffX = mouseX - oldX; f diffY = mouseY - oldY;//get difference from last frame to this frame
if (mouseX != 0 || mouseY != 0) {
mainCamera->Rotate(diffX, diffY);
}
Code to rotate the camera
void Camera::Rotate(f angleX, f angleY) {
Camera::refrence = Vector3D::NormalizeVector(Camera::refrence * cos(angleX)) + (Camera::upVector * sin(angleY));//rot up
Camera::refrence = Vector3D::NormalizeVector((Camera::refrence * cos(angleY)) - (Camera::rightVector * sin(angleX)));//rot side to side
};
Camera::refrence is our lookat point, processing the lookat point is handled as follows
void Camera::LookAt(void) {
gluLookAt(
Camera::position.x, Camera::position.y, Camera::position.z,
Camera::refrence.x, Camera::refrence.y, Camera::refrence.z,
Camera::upVector.x, Camera::upVector.y, Camera::upVector.z
);
};
The camera is defined by a position point (position) a target point (refrence) and a up-vector upVector. If you want to change the orientation of the camera, then you've to rotate the direction vector from the position (position) to the target (refrence) rather then the target point by a Rotation matrix.
Note, since the 2 angles are angles which should change an already rotated view, you've to use a rotation matrix, to rotate the vectors which point in an arbitrary direction.
Write a function which set 3x3 rotation matrix around an arbitrary axis:
void RotateMat(float m[], float angle_radians, float x, float y, float z)
{
float c = cos(angle_radians);
float s = sin(angle_radians);
m[0] = x*x*(1.0f-c)+c; m[1] = x*y*(1.0f-c)-z*s; m[2] = x*z*(1.0f-c)+y*s;
m[3] = y*x*(1.0f-c)+z*s; m[4] = y*y*(1.0f-c)+c; m[5] = y*z*(1.0f-c)-x*s;
m[6] = z*x*(1.0f-c)-y*s; m[7] = z*y*(1.0f-c)+x*s; m[8] = z*z*(1.0f-c)+c };
}
Write a function which rotates a 3 dimensional vector by the matrix:
Vector3D Rotate(float m[], const Vector3D &v)
{
Vector3D rv;
rv.x = m[0] * v.x + m[3] * v.y + m[6] * v.z;
rv.y = m[1] * v.x + m[4] * v.y + m[7] * v.z;
rv.z = m[2] * v.x + m[5] * v.y + m[8] * v.z;
return rv;
}
Calculate the vector form the position to the target:
Vector3D los = Vector3D(refrence.x - position.x, refrence.y - position.y, refrence.z - position.z);
Rotate all the vectors around the z axis of the world by angleX:
float rotX[9];
RotateMat(rotX, angleX, Vector3D(0, 0, 1));
los = Rotate(rotX, los);
upVector = Rotate(rotX, upVector);
Rotate all the vectors around the current y axis of the view by angleY:
float rotY[9];
RotateMat(rotY, angleY, Vector3D(los.x, los.y, 0.0));
los = Rotate(rotY, los);
upVector = Rotate(rotY, upVector);
Calculate the new target point:
refrence = Vector3D(position.x + los.x, position.y + los.y, position.z + los.z);
U_Cam_X_angle is left right rotation.. U_Cam_Y_angle is up down rotation.
view_radius is the view distance (zoom) to U_look_point_x, U_look_point_y and U_look_point_z.
This is ALWAYS a negative number! This is because you are always looking in positive direction. Deeper in the screen is more positive.
This is all in radians.
The last three.. eyeX, eyeY and eyeZ is where the camera is in 3D space.
This code is in VB.net. Find a converter online for VB to C++ or do it manually.
Public Sub set_eyes()
Dim sin_x, sin_y, cos_x, cos_y As Single
sin_x = Sin(U_Cam_X_angle + angle_offset)
cos_x = Cos(U_Cam_X_angle + angle_offset)
cos_y = Cos(U_Cam_Y_angle)
sin_y = Sin(U_Cam_Y_angle)
cam_y = Sin(U_Cam_Y_angle) * view_radius
cam_x = (sin_x - (1 - cos_y) * sin_x) * view_radius
cam_z = (cos_x - (1 - cos_y) * cos_x) * view_radius
Glu.gluLookAt(cam_x + U_look_point_x, cam_y + U_look_point_y, cam_z + U_look_point_z, _
U_look_point_x, U_look_point_y, U_look_point_z, 0.0F, 1.0F, 0.0F)
eyeX = cam_x + U_look_point_x
eyeY = cam_y + U_look_point_y
eyeZ = cam_z + U_look_point_z
End Sub
I am trying to display a 360 panorama using an IMU for head tracking.
Yaw works correctly but the roll and pitch are reverse. I also notice that the pitch contains some roll (and maybe vice-versa).
I am receiving (W, X, Y, Z) coordinate from the IMU that I am storing in an array as X, Y, Z, W.
The next step is converting the quaternion to a rotation matrix. I have looked at many examples, and can't seem to find anything wrong with the following code:
static GLfloat rotation[16];
// Quaternion (x, y, z, w)
static void quaternionToRotation(float* quaternion)
{
// Normalize quaternion
float magnitude = sqrt(quaternion[0] * quaternion[0] +
quaternion[1] * quaternion[1] +
quaternion[2] * quaternion[2] +
quaternion[3] * quaternion[3]);
for (int i = 0; i < 4; ++i)
{
quaternion[i] /= magnitude;
}
double xx = quaternion[0] * quaternion[0], xy = quaternion[0] * quaternion[1],
xz = quaternion[0] * quaternion[2], xw = quaternion[0] * quaternion[3];
double yy = quaternion[1] * quaternion[1], yz = quaternion[1] * quaternion[2],
yw = quaternion[1] * quaternion[3];
double zz = quaternion[2] * quaternion[2], zw = quaternion[2] * quaternion[3];
// Column major order
rotation[0] = 1.0f - 2.0f * (yy + zz);
rotation[1] = 2.0f * (xy - zw);
rotation[2] = 2.0f * (xz + yw);
rotation[3] = 0;
rotation[4] = 2.0f * (xy + zw);
rotation[5] = 1.0f - 2.0f * (xx + zz);
rotation[6] = 2.0f * (yz - xw);
rotation[7] = 0;
rotation[8] = 2.0f * (xz - yw);
rotation[9] = 2.0f * (yz + xw);
rotation[10] = 1.0f - 2.0f * (xx + yy);
rotation[11] = 0;
rotation[12] = 0;
rotation[13] = 0;
rotation[14] = 0;
rotation[15] = 1;
}
The rotation matrix is then used in the draw call as such:
static void draw()
{
// Get IMU quaternion
float* quaternion = tracker.getTrackingData();
if (quaternion != NULL)
{
quaternionToRotation(quaternion);
}
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
glLoadIdentity();
glPushMatrix();
// TODO: Multiply initialRotation quaternion with IMU quaternion
glMultMatrixf(initialRotation); // Initial rotation to point forward
glMultMatrixf(rotation); // Rotation based on IMU
glEnable(GL_TEXTURE_2D);
glBindTexture(GL_TEXTURE_2D, texture);
gluSphere(quad, 0.1, 50, 50);
glBindTexture(GL_TEXTURE_2D, 0);
glPopMatrix();
glFlush();
glutSwapBuffers();
}
I tried to set all but one fields in the quaternion to 0, and I notice that they all work individually, except roll and pitch is swapped around. I tried swapping X and Y but this does not seem to help.
Any help would be really appreciated. Please let me know as well if you have any steps that can let me debug my issue. Thanks!
The problem is when I face my camera down the z axis for example and pitch this works fine however, after I have finished the pitch and would like to yaw on this new axis it begins to roll for some unknown reason =s.
void FrustumCamera::xAxisRotation(float angle)
{
// angle = angle * (double)degToRad;
Vector3<float> x = m_orientation.getXAxis();
Vector3<float> y = m_orientation.getYAxis();
Vector3<float> z = m_orientation.getZAxis();
y.rotateAroundAxis(x,angle);
x = m_orientation.getXAxis();
z.rotateAroundAxis(x,angle);
m_orientation.setYAxis(y);
m_orientation.setZAxis(z);
}
void FrustumCamera::yAxisRotation(float angle)
{
// angle = angle * (double)degToRad;
Vector3<float> x = m_orientation.getXAxis();
Vector3<float> y = m_orientation.getYAxis();
Vector3<float> z = m_orientation.getZAxis();
x.rotateAroundAxis(y,angle);
y = m_orientation.getYAxis();
z.rotateAroundAxis(y,angle);
m_orientation.setXAxis(x);
m_orientation.setZAxis(z);
}
void FrustumCamera::zAxisRotation(float angle)
{
Vector3<float> x = m_orientation.getXAxis();
Vector3<float> y = m_orientation.getYAxis();
Vector3<float> z = m_orientation.getZAxis();
x.rotateAroundAxis(z,angle);
z = m_orientation.getYAxis();
y.rotateAroundAxis(z,angle);
m_orientation.setXAxis(x);
m_orientation.setYAxis(y);
}
template <class Type>
void Vector3<Type>::rotateAroundAxis(Vector3<Type> axis, const float angle)
{
float radians = static_cast<Type>(angle * degToRad);
Type sinAngle = static_cast<Type>(sin(radians));
Type cosAngle = 0.0;
if (angle == 90 || angle == -90)
cosAngle = 0.0;
else
cosAngle = cos(radians);
normalise(axis); // normalise the axis
Type oneMinusCos = 1 - cosAngle; // (1 - cos(theta))
// construct the rotation matrix
Type tempMatrix[3][3];
tempMatrix[0][0] = (axis.x * axis.x) * oneMinusCos + cosAngle;
tempMatrix[0][1] = (axis.x * axis.y) * oneMinusCos + axis.z * sinAngle;
tempMatrix[0][2] = (axis.x * axis.z) * oneMinusCos - axis.y * sinAngle;
tempMatrix[1][0] = (axis.x * axis.y) * oneMinusCos - axis.z * sinAngle;
tempMatrix[1][1] = (axis.y * axis.y) * oneMinusCos + cosAngle;
tempMatrix[1][2] = (axis.y * axis.z) * oneMinusCos + axis.x * sinAngle;
tempMatrix[2][0] = (axis.x * axis.z) * oneMinusCos + axis.y * sinAngle;
tempMatrix[2][1] = (axis.y * axis.z) * oneMinusCos - axis.x * sinAngle;
tempMatrix[2][2] = (axis.z * axis.z) * oneMinusCos + cosAngle;
Vector3<Type> temp(*this);
Vector3<Type> result;
result.x = (temp.x * tempMatrix[0][0]) + (temp.y * tempMatrix[1][0]) + (temp.z * tempMatrix[2][0]);
result.y = (temp.x * tempMatrix[0][1]) + (temp.y * tempMatrix[1][1]) + (temp.z * tempMatrix[2][1]);
result.z = (temp.x * tempMatrix[0][2]) + (temp.y * tempMatrix[1][2]) + (temp.z * tempMatrix[2][2]);
*this = result;
}
void OpenGLRenderer::startDraw(unsigned long mask)
{
//sortBuffer(); // sort draw queue
clearBuffers(mask); // clear buffers
loadIdentity();
glTranslatef(-1*m_frustumCamera->getViewMatrix().getTranslationAxis().x,-1*m_frustumCamera->getViewMatrix().getTranslationAxis().y,-1*m_frustumCamera->getViewMatrix().getTranslationAxis().z);
glMultMatrixf(m_frustumCamera->getViewMatrix().getMatrix());
glTranslatef(m_frustumCamera->getViewMatrix().getTranslationAxis().x,m_frustumCamera->getViewMatrix().getTranslationAxis().y,m_frustumCamera->getViewMatrix().getTranslationAxis().z);// load identity
//
// push matrix stack
matrixStackPush();
}
You might be experiencing Gimbal Lock; this can happen if you pitch all the way up or down so your look vector becomes parallel with your up vector, In which case a yaw will be the same as a roll.
This can be a downside of constructing rotations piecemeal via Euler angles. You may want to look into quaternions. (Note that you cant rotate with Euler angles; they are just a representation for rotation (you need to convert it to matrix or quats), but the way you are tackling it is very much an 'Euler angle' way of thinking about it)
The strength of matrix multiplication is that any sequence of multiple rotations can be represented (and concatenated) as a single rotation matrix. What you need to be doing is something like this:
void Transformable::yaw(float angle)
{
float4x4 rot; // temp rotation matrix
float3 translate(&_transform._41); // save our translation
float3 up(&_transform._21); // y axis
// build the rotation matrix for rotation around y
MatrixRotationAxis(&rot, &up, angle);
// multiply our transform by the rotation matrix
// note that order of multiplication matters and depends on
// if your matrices are column-major or row-major
MatrixMultiply(&_transform, &_transform, &rot);
// write back our original translation
memcpy(&_transform._41, &translate, sizeof(float3));
// might want to reorthogonalise every now and then
// to make sure basis vectors are orthonormal
// or you will probably get matrix creep after a few operations
}
instead of trying to rotate one basis vector at a time. In this case _transform would be a 4x4 homogenous matrix representing the transformation matrix. (rotation and translation). The topleft 3x3 submatrix is simply the basis vectors of the orientation space.