I'm trying to implement a quaternion-based camera, but when moving around the X and Y axis, the camera produces an unwanted roll on the Z axis. I want to be able to look around freely on all axis.
I've read other topics about this problem (for example: http://www.flipcode.com/forums/thread/6525 ), but I'm not getting what is meant by "Fix this by continuously rebuilding the rotation matrix by rotating around the WORLD axis, i.e around <1,0,0>, <0,1,0>, <0,0,1>, not your local coordinates, whatever they might be."
//Camera.hpp
glm::quat rotation;
//Camera.cpp
void Camera::rotate(glm::vec3 vec)
{
glm::quat paramQuat = glm::quat(vec);
rotation = paramQuat * rotation;
}
I call the rotate function like this:
cam->rotate(glm::vec3(0, 0.5, 0));
This must have to do with local/world coordinates, right? I'm just not getting it, since I thought quaternions are always based on each other (thus a quaternion can't be in "world" or "local" space?).
Also, should i use more than one quaternion for a camera?
As far as I understand it, and from looking at the code you provided, what they mean is that you shouldn't store and apply the rotation incrementally by applying rotate on the rotation quat all the time, but instead keeping track of two quaternions for each axis (X and Y in world space) and calculating the rotation vector as the product of those.
[edit: some added (pseudo)code]
// Camera.cpp
void Camera::SetRotation(glm::quat value)
{
rotation = value;
}
// controller.cpp --> probably a place where you'd want to translate user input and store your rotational state
xAngle += deltaX;
yAngle += deltaY;
glm::quat rotationX = QuatAxisAngle(X_AXIS, xAngle);
glm::quat rotationY = QuatAxisAngle(Y_AXIS, yAngle);
camera.SetRotation(rotationX * rotationY);
Related
I'm having problems rotating GameObjects in my engine. I'm trying to rotate in 2 ways.
I'm using MathGeoLib to calculate maths in the engine.
First way: Rotates correctly around axis but if I want to rotate back, if I don't do it following the inverse order then rotation doesn't work properly.
e.g:
Rotate X axis 50 degrees, Rotate Y axis 30 degrees -> Rotate Y axis -50 degrees, Rotate X axis -30 degrees. Works.
Rotate X axis 50 degrees, Rotate Y axis 30 degrees -> Rotate X axis -50 degrees, Rotate Y axis -30 degrees. Doesn't.
Code:
void ComponentTransform::SetRotation(float3 euler_rotation)
{
float3 diff = euler_rotation - editor_rotation;
editor_rotation = euler_rotation;
math::Quat mod = math::Quat::FromEulerXYZ(diff.x * DEGTORAD, diff.y * DEGTORAD, diff.z * DEGTORAD);
quat_rotation = quat_rotation * mod;
UpdateMatrix();
}
Second way: Starts rotating good around axis but after rotating some times, then it stops to rotate correctly around axis, but if I rotate it back regardless of the rotation order it works, not like the first way.
Code:
void ComponentTransform::SetRotation(float3 euler_rotation)
{
editor_rotation = euler_rotation;
quat_rotation = math::Quat::FromEulerXYZ(euler_rotation.x * DEGTORAD, euler_rotation.y * DEGTORAD, euler_rotation.z * DEGTORAD);
UpdateMatrix();
}
Rest of code:
#define DEGTORAD 0.0174532925199432957f
void ComponentTransform::UpdateMatrix()
{
if (!this->GetGameObject()->IsParent())
{
//Get parent transform component
ComponentTransform* parent_transform = (ComponentTransform*)this->GetGameObject()->GetParent()->GetComponent(Component::CompTransform);
//Create matrix from position, rotation(quaternion) and scale
transform_matrix = math::float4x4::FromTRS(position, quat_rotation, scale);
//Multiply the object transform by parent transform
transform_matrix = parent_transform->transform_matrix * transform_matrix;
//If object have childs, call this function in childs objects
for (std::list<GameObject*>::iterator it = this->GetGameObject()->childs.begin(); it != this->GetGameObject()->childs.end(); it++)
{
ComponentTransform* child_transform = (ComponentTransform*)(*it)->GetComponent(Component::CompTransform);
child_transform->UpdateMatrix();
}
}
else
{
//Create matrix from position, rotation(quaternion) and scale
transform_matrix = math::float4x4::FromTRS(position, quat_rotation, scale);
//If object have childs, call this function in childs objects
for (std::list<GameObject*>::iterator it = this->GetGameObject()->childs.begin(); it != this->GetGameObject()->childs.end(); it++)
{
ComponentTransform* child_transform = (ComponentTransform*)(*it)->GetComponent(Component::CompTransform);
child_transform->UpdateMatrix();
}
}
}
MathGeoLib:
Quat MUST_USE_RESULT Quat::FromEulerXYZ(float x, float y, float z) { return (Quat::RotateX(x) * Quat::RotateY(y) * Quat::RotateZ(z)).Normalized(); }
Quat MUST_USE_RESULT Quat::RotateX(float angle)
{
return Quat(float3(1,0,0), angle);
}
Quat MUST_USE_RESULT Quat::RotateY(float angle)
{
return Quat(float3(0,1,0), angle);
}
Quat MUST_USE_RESULT Quat::RotateZ(float angle)
{
return Quat(float3(0,0,1), angle);
}
Quat(const float3 &rotationAxis, float rotationAngleRadians) { SetFromAxisAngle(rotationAxis, rotationAngleRadians); }
void Quat::SetFromAxisAngle(const float3 &axis, float angle)
{
assume1(axis.IsNormalized(), axis);
assume1(MATH_NS::IsFinite(angle), angle);
float sinz, cosz;
SinCos(angle*0.5f, sinz, cosz);
x = axis.x * sinz;
y = axis.y * sinz;
z = axis.z * sinz;
w = cosz;
}
Any help?
Thanks.
Using Euler angles and or Quaternions adds some limitations as it creates singularities which if not handled correctly will make silly things. Sadly almost all new 3D games using it wrongly. You can detect those by the well known things like:
sometimes your view get to very different angle that should not be there
object can not rotate anymore in some direction
object start rotating around different axises than it should
view jumps around singularity pole
view is spinning or flipping until you move/turn again (not the one caused by optic mouse error)
I am using cumulative transform matrices instead:
Understanding 4x4 homogenous transform matrices
Read the whole stuff (especially difference between local and global rotations) then in last 3 links you got C++ examples of how to do this (also read all 3 especially the preserving accuracy ...).
The idea is to have matrix representing your object coordinate system. And when ever you rotate (by mouse, keyboard, NAV,AI,...) you rotate the matrix (incrementally). The same goes for movement. This way they are no limitations or singularities. But also this approach has its problems:
lose of accuracy with time (read the preserving accuracy example to deal with this)
no knowledge about the Euler angles (the angles can be computed from the matrix however)
Both are solvable relatively easily.
Now when you are rotating around local axises you need to take into account that with every rotation around some axis you change the other two. So if you want to get to the original state you need to reverse order of rotations because:
rotate around x by 30deg
rotate around y by 40deg
is not the same as:
rotate around y by 40deg
rotate around x by 30deg
With cumulative matrix if you want to get back you can either iteratively drive your ship until it faces desired directions or remember original matrix and compute the rotations needed to be done one axis at a time. Or convert the matrix difference into quaternion and iterate that single rotation...
I'm currently in the process of finishing the implementation for a camera that functions in the same way as the camera in Maya. The part I'm stuck in the tumble functionality.
The problem is the following: the tumble feature works fine so long as the position of the camera is not parallel with the up vector (currently defined to be (0, 1, 0)). As soon as the camera becomes parallel with this vector (so it is looking straight up or down), the camera locks in place and will only rotate around the up vector instead of continuing to roll.
This question has already been asked here, unfortunately there is no actual solution to the problem. For reference, I also tried updating the up vector as I rotated the camera, but the resulting behaviour is not what I require (the view rolls as a result of the new orientation).
Here's the code for my camera:
using namespace glm;
// point is the position of the cursor in screen coordinates from GLFW
float deltaX = point.x - mImpl->lastPos.x;
float deltaY = point.y - mImpl->lastPos.y;
// Transform from screen coordinates into camera coordinates
Vector4 tumbleVector = Vector4(-deltaX, deltaY, 0, 0);
Matrix4 cameraMatrix = lookAt(mImpl->eye, mImpl->centre, mImpl->up);
Vector4 transformedTumble = inverse(cameraMatrix) * tumbleVector;
// Now compute the two vectors to determine the angle and axis of rotation.
Vector p1 = normalize(mImpl->eye - mImpl->centre);
Vector p2 = normalize((mImpl->eye + Vector(transformedTumble)) - mImpl->centre);
// Get the angle and axis
float theta = 0.1f * acos(dot(p1, p2));
Vector axis = cross(p1, p2);
// Rotate the eye.
mImpl->eye = Vector(rotate(Matrix4(1.0f), theta, axis) * Vector4(mImpl->eye, 0));
The vector library I'm using is GLM. Here's a quick reference on the custom types used here:
typedef glm::vec3 Vector;
typedef glm::vec4 Vector4;
typedef glm::mat4 Matrix4;
typedef glm::vec2 Point2;
mImpl is a PIMPL that contains the following members:
Vector eye, centre, up;
Point2 lastPoint;
Here is what I think. It has something to do with the gimbal lock, that occurs with euler angles (and thus spherical coordinates).
If you exceed your minimal(0, -zoom,0) or maxima(0, zoom,0) you have to toggle a boolean. This boolean will tell you if you must treat deltaY positive or not.
It could also just be caused by a singularity, therefore just limit your polar angle values between 89.99° and -89.99°.
Your problem could be solved like this.
So if your camera is exactly above (0, zoom,0) or beneath (0, -zoom,0) of your object, than the camera only rolls.
(I am also assuming your object is at (0,0,0) and the up-vector is set to (0,1,0).)
There might be some mathematical trick to resolve this, I would do it with linear algebra though.
You need to introduce a new right-vector. If you make a cross product, you will get the camera-vector. Camera-vector = up-vector x camera-vector. Imagine these vectors start at (0,0,0), then easily, to get your camera position just do this subtraction (0,0,0)-(camera-vector).
So if you get some deltaX, you rotate towards the right-vector(around the up-vector) and update it.
Any influence of deltaX should not change your up-vector.
If you get some deltaY you rotate towards the up-vector(around the right-vector) and update it. (This has no influence on the right-vector).
https://en.wikipedia.org/wiki/Rotation_matrix at Rotation matrix from axis and angle you can find a important formula.
You say u is your vector you want to rotate around and theta is the amount you want to pivot. The size of theta is proportional to deltaX/Y.
For example: We got an input from deltaX, so we rotate around the up-vector.
up-vector:= (0,1,0)
right-vector:= (0,0,-1)
cam-vector:= (0,1,0)
theta:=-1*30° // -1 due to the positive mathematical direction of rotation
R={[cos(-30°),0,-sin(-30°)],[0,1,0],[sin(-30°),0,cos(-30°)]}
new-cam-vector=R*cam-vector // normal matrix multiplication
One thing is left to be done: Update the right-vector.
right-vector=camera-vector x up-vector .
Lets say I have an object, and that object has a quaternion representing its orientation.
Currently, I can rotate on all 3 axes without gimbal lock, however, each rotation on any axis SHOULD rotate the other 2 axes of rotation.
What I mean by this is if I pitch an object towards the camera, but then yaw the object 90 degrees away, pitching the object will still rotate it relative to where it was before, not where it is now.
Here's a visual example of my problem:
I'm using quaternions and not euler rotations because these objects can rotate on all 3 axes, and I don't want to gimbal lock/reach singularity
I rotate my object's orientation quaternion like this:
orientation = Quaternion(Vector(0,1,0),angle) * orientation;
I then trigger a rebuilding of my object's vectors, and apply it to them (after transforming the object relative to 3D space origin point):
Quaternion Point = Quaternion(( orientation * (Vertex) ) * (orientation.inverse()));
vertices[x] = QVector3D(round(Point.v.x),round(Point.v.y),round(Point.v.z));
And when I multiply my quaternions by other quaternions, this is the multiplication operator's function:
Quaternion Quaternion::operator*(Quaternion& q) const
{
Quaternion r;
//"w" is the angle, "v" is the vector
r.w = w*q.w - glm::dot(v, q.v);
r.v = v*q.w + q.v*w + glm::cross(v,q.v);
//r.normalize();
return r;
}
I swear to god, I'm only posting because this topic is confusing me to no end. Solidifying this system will be make me unfathomably happy.
I'm implementing a first person camera using the GLM library that provides me with some useful functions that calculate perspective and 'lookAt' matrices. I'm also using OpenGL but that shouldn't make a difference in this code.
Basically, what I'm experiencing is that I can look around, much like in a regular FPS, and move around. But the movement is constrained to the three axes in a way that if I rotate the camera, I would still move in the same direction as if I had not rotated it... Let me illustrate (in 2D, to simplify things).
In this image, you can see four camera positions.
Those marked with a one are before movement, those marked with a two are after movement.
The red triangles represent a camera that is oriented straight forward along the z axis. The blue triangles represent a camera that hasbeen rotated to look backward along the x axis (to the left).
When I press the 'forward movement key', the camera moves forward along the z axis in both cases, disregarding the camera orientation.
What I want is a more FPS-like behaviour, where pressing forward moves me in the direction the camera is facing. I thought that with the arguments I pass to glm::lookAt, this would be achieved. Apparently not.
What is wrong with my calculations?
// Calculate the camera's orientation
float angleHori = -mMouseSpeed * Mouse::x; // Note that (0, 0) is the center of the screen
float angleVert = -mMouseSpeed * Mouse::y;
glm::vec3 dir(
cos(angleVert) * sin(angleHori),
sin(angleVert),
cos(angleVert) * cos(angleHori)
);
glm::vec3 right(
sin(angleHori - M_PI / 2.0f),
0.0f,
cos(angleHori - M_PI / 2.0f)
);
glm::vec3 up = glm::cross(right, dir);
// Calculate projection and view matrix
glm::mat4 projMatrix = glm::perspective(mFOV, mViewPortSizeX / (float)mViewPortSizeY, mZNear, mZFar);
glm::mat4 viewMatrix = glm::lookAt(mPosition, mPosition + dir, up);
gluLookAt takes 3 parameters: eye, centre and up. The first two are positions while the last is a vector. If you're planning on using this function it's better that you maintain only these three parameters consistently.
Coming to the issue with the calculation. I see that the position variable is unchanged throughout the code. All that changes is the look at point I.e. centre only. The right thing to do is to first do position += dir, which will move the camera (position) along the direction pointed to by dir. Now to update the centre, the second parameter can be left as-is: position + dir; this will work since the position was already updated to the new position and from there we've a point farther in dir direction to look at.
The issue was actually in another method. When moving the camera, I needed to do this:
void Camera::moveX(char s)
{
mPosition += s * mSpeed * mRight;
}
void Camera::moveY(char s)
{
mPosition += s * mSpeed * mUp;
}
void Camera::moveZ(chars)
{
mPosition += s * mSpeed * mDirection;
}
To make the camera move across the correct axes.
I'm trying to figure out how to make the camera in directx move based on the direction it's facing.
Right now the way I move the camera is by passing the camera's current position and rotation to a class called PositionClass. PositionClass takes keyboard input from another class called InputClass and then updates the position and rotation values for the camera, which is then passed back to the camera class.
I've written some code that seems to work great for me, using the cameras pitch and yaw I'm able to get it to go in the direction I've pointed the camera.
However, when the camera is looking straight up (pitch=90) or straight down (pitch=-90), it still changes the cameras X and Z position (depending on the yaw).
The expected behavior is while looking straight up or down it will only move along the Y axis, not along the X or Z axis.
Here's the code that calculates the new camera position
void PositionClass::MoveForward(bool keydown)
{
float radiansY, radiansX;
// Update the forward speed movement based on the frame time
// and whether the user is holding the key down or not.
if(keydown)
{
m_forwardSpeed += m_frameTime * m_acceleration;
if(m_forwardSpeed > (m_frameTime * m_maxSpeed))
{
m_forwardSpeed = m_frameTime * m_maxSpeed;
}
}
else
{
m_forwardSpeed -= m_frameTime * m_friction;
if(m_forwardSpeed < 0.0f)
{
m_forwardSpeed = 0.0f;
}
}
// ToRadians() just multiplies degrees by 0.0174532925f
radiansY = ToRadians(m_rotationY); //yaw
radiansX = ToRadians(m_rotationX); //pitch
// Update the position.
m_positionX += sinf(radiansY) * m_forwardSpeed;
m_positionY += -sinf(radiansX) * m_forwardSpeed;
m_positionZ += cosf(radiansY) * m_forwardSpeed;
return;
}
The significant portion is where the position is updated at the end.
So far I've only been able to deduce that I have horrible math skills.
So, can anyone help me with this dilemma? I've created a fiddle to help test out the math.
Edit: The fiddle uses the same math I used in my MoveForward function, if you set pitch to 90 you can see that the Z axis is still being modified
Thanks to Chaosed0's answer, I was able to figure out the correct formula to calculate movement in a specific direction.
The fixed code below is basically the same as above but now simplified and expanded to make it easier to understand.
First we determine the amount by which the camera will move, in my case this was m_forwardSpeed, but here I will define it as offset.
float offset = 1.0f;
Next you will need to get the camera's X and Y rotation values (in degrees!)
float pitch = camera_rotationX;
float yaw = camera_rotationY;
Then we convert those values into radians
float pitchRadian = pitch * (PI / 180); // X rotation
float yawRadian = yaw * (PI / 180); // Y rotation
Now here is where we determine the new position:
float newPosX = offset * sinf( yawRadian ) * cosf( pitchRadian );
float newPosY = offset * -sinf( pitchRadian );
float newPosZ = offset * cosf( yawRadian ) * cosf( pitchRadian );
Notice that we only multiply the X and Z positions by the cosine of pitchRadian, this is to negate the direction and offset of your camera's yaw when it's looking straight up (90) or straight down (-90).
And finally, you need to tell your camera the new position, which I won't cover because it largely depends on how you've implemented your camera. Apparently doing it this way is out of the norm, and possibly inefficient. However, as Chaosed0 said, it's what makes the most sense to me!
To be honest, I'm not entirely sure I understand your code, so let me try to provide a different perspective.
The way I like to think about this problem is in spherical coordinates, basically just polar in 3D. Spherical coordinates are defined by three numbers: a radius and two angles. One of the angles is yaw, and the other should be pitch, assuming you have no roll (I believe there's a way to get phi if you have roll, but I can't think of how currently). In conventional mathematics notation, theta is your yaw and phi is your pitch, with radius being your move speed, as shown below.
Note that phi and theta are defined differently, depending on where you look.
Basically, the problem is to obtain a point m_forwardSpeed away from your camera, with the right pitch and yaw. To do this, we set the "origin" to your camera position, obtain a spherical coordinate, convert it to cartesian, and then add it to your camera position:
float radius = m_forwardSpeed;
float theta = m_rotationY;
float phi = m_rotationX
//These equations are from the wikipedia page, linked above
float xMove = radius*sinf(phi)*cosf(theta);
float yMove = radius*sinf(phi)*sinf(theta);
float zMove = radius*cosf(phi);
m_positionX += xMove;
m_positionY += yMove;
m_positionZ += zMove;
Of course, you can condense a lot of this code, but I expanded it for clarity.
You can think about this like drawing a sphere around your camera. Each of the points on the sphere is a potential position in the next timestep, depending on the camera's rotation.
This is probably not the most efficient way to do it, but in my opinion it's certainly the easiest way to think about it. It actually looks like this is nearly exactly what you're trying to do in your code, but the operations on the angles are just a little bit off.