Jump to content

  • Log In with Google      Sign In   
  • Create Account

We're offering banner ads on our site from just $5!

1. Details HERE. 2. GDNet+ Subscriptions HERE. 3. Ad upload HERE.


Trouble with Quaternion camera motion.


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
3 replies to this topic

#1 gp521   Members   -  Reputation: 105

Like
0Likes
Like

Posted 26 October 2012 - 06:42 PM

Hi all, I hope someone can help. I'm having trouble getting my Quaternion based camera to move correctly. I've been working at this code for a while and searching for answers to no avail. Rotating the camera works perfectly, but where I'm having an issue is with moving the camera forward.

Basically how it should work is that I want it to move forward in the direction the camera is pointing and move freely in any direction, like a camera should in a space sim. Current controls are pitch and yaw, for now there is no roll control.

Currently, pitching and yawing works correctly for a while, the camera flys and objects appear to move correctly, but after a few pitches or yaws around with the camera, eventually the objects end up moving in some direction that makes no sense, for example sideways to the camera or away from it. I can't figure out how to fix this. Any advice or help on this would be greatly appreciated!

How I'm trying to do this is like this:

// For each frame

// Convert the Quaternion to Euler vector, this seems to be correct, the vector always points in the direction the camera is facing.
Vector3 vec = QuatenionToEuler(quaternion); // vector returned is 0 to -PI PI

// Get worldview rotation matrix from the quaterion.
Matrix4 matrix = QuaternionToMatrix4(quaternion);

// Update the camera position.
/***** Something is wrong here, how to fix? ******/
cam_pos.x -= sin(vec.x) * forwardspeed * dt;
cam_pos.y += sin(vec.y) * forwardspeed * dt;
cam_pos.z += cos(vec.x) * cos(vec.y) * forwardspeed * dt;

// translate the worldview matrix.
matrix = Matrix4Translate(matrix, cam_pos.x, cam_pos.y, cam_pos.z);

Sponsor:

#2 zacaj   Members   -  Reputation: 643

Like
2Likes
Like

Posted 26 October 2012 - 07:00 PM

Instead of extracting the eulwr angles, I just create a rotation matrix from the quaternion, and then multiply it by vec3(0,0,-1) for forward, vec3(1,0,0) for right, etc

#3 gp521   Members   -  Reputation: 105

Like
0Likes
Like

Posted 27 October 2012 - 07:37 AM

Thanks for the advice, I got it working, my ship is flying! I didn't quite go this approach, but it got me to abandon the convert to euler dead end and think about this differently. The answer turned out to be quite simple actually, the rotation matrix contains the direction! Here's my modified code:

// For each frame

// Get worldview rotation matrix from the quaterion.
Matrix4 matrix = QuaternionToMatrix4(quaternion);

// Update the camera position.
Vector3 forward = Vector3(matrix.m[2], matrix.m[6], matrix.m[10]);
forward = Vector3Normalize(forward);

cam_pos.x += forward.x * speed * dt;
cam_pos.y += forward.y * speed * dt;
cam_pos.z += forward.z * speed * dt;

// translate the worldview matrix.
matrix = Matrix4Translate(matrix, cam_pos.x, cam_pos.y, cam_pos.z);

Edited by gp521, 27 October 2012 - 07:37 AM.


#4 zacaj   Members   -  Reputation: 643

Like
0Likes
Like

Posted 27 October 2012 - 11:17 AM

That isn't quite safe. Some matrix transformations will rearrange the 3x3 section of the matrix so that 2,6,10 aren't the three components.




Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS