Much ado about guMtxRotAxisDeg(....)

First off, the example code I am compiling is from nehe lesson6 with a few modifications for my own sake. I wanted to take the example of the rotating cube and experiment with setting up a camera that had yaw, pitch, and roll and keep the cube as a center reference point (for testing anyway).

Here is a snippet of code that I have regarding the rotation matrices. I just need a few things demystified

some declarations:

f32 camXRot = -45;
f32 camYRot = 0;

guVector aboutXAxis = {1,0,0};
guVector aboutYAxis = {0,1,0};

and the implementation :

guMtxIdentity(model);
guMtxRotAxisDeg(model, &aboutXAxis, camXRot);
guMtxRotAxisDeg(model, &aboutYAxis, camYRot);
		
guMtxTransApply(model, model, 0.0f, 0.0f, -8.0f);
guMtxConcat(view,model,modelview);
// load the modelview matrix into matrix memory
GX_LoadPosMtxImm(modelview, GX_PNMTX0);

Basically the camXRot and camYRot values are incremented through keypad input, and while x rotations seem to work, y rotations don't. If I comment out the x rotations then my y rotations start working. This makes me think there is some order/limit to specifying multi-axis rotations, but I don't know exactly what the guMtxRot function is doing. I have relatively no experience with matrices. So, although this may seem fundamentally simple, I don't know what the process is for multi-axis rotations. A nod in the right direction would be most appreciated.

Thanks

The guMtxRotAxisDeg function sets a rotation matrix, it does not multiply with the matrix (like glRotate does)

So you need to setup a 'temp' matrix for that.

Example:

    Mtx model, modelview;
    Mtx model2, model3;
    Mtx mrx, mry,mrz;

    guMtxIdentity(model);
    guMtxIdentity(model2);
    guMtxRotAxisDeg(model, &pitchAxis, pitch);
    guMtxTransApply(model, model, x, y, z);
    guMtxRotAxisDeg(mrx, &yawAxis, yaw2);
    guMtxRotAxisDeg(mry, &pitchAxis, pitch2);
    guMtxRotAxisDeg(mrz, &rollAxis, roll2);
    guMtxConcat(mrx,mry,mry);
    guMtxConcat(mry,mrz,model2);
    guMtxTransApply(model2, model2, x2, y2, z2);
    guMtxConcat(model,model2,model3);
    guMtxConcat(view,model3,modelview);
    // load the modelview matrix into matrix memory
    GX_LoadPosMtxImm(modelview, GX_PNMTX0);

And you probably want to rotate your matrix incrementally instead of storing rotation values and applying them as a whole. This is because it is not the same to rotate a matrix 10 degrees around Y, 10 around X and then again 10 about Y that rotating it 20 deg around Y and then 10 around X. The first aproach makes more sense "physically"

Daid:

Thanks a lot! That is exactly what I was looking for. I tried it out with a few modifications and it seems to do the job perfectly.

I was assuming that gu would have a wrapper for a glRotate type function, and the libogc doxygen documentation site only has a couple of these matrix functions listed so it is hard to know what they do. In any case you've just made my day.

Tecnik:

I'm not sure I follow what you are saying. Is choosing incrementing over storing my rotations a choice that comes down to desired behavior or is it more a matter of technical correctness? The code I have in place at the moment acts exactly how I want it to, but I want to make sure it is the correct way.



Edited 1 time(s). Last edit at 02/09/2011 09:32AM by glennxserge.

Quote
glennxserge
I was assuming that gu would have a wrapper for a glRotate type function, and the libogc doxygen documentation site only has a couple of these matrix functions listed so it is hard to know what they do. In any case you've just made my day.

I hardly use the doxygen from libogc. I found out how to work with gu/gx by looking at sources from different projects on wiibrew.org

wait @Daid I have a questio, so how do you know where the new top edge of your object is? Say for instance I draw a cude in GRRLIB and I rotate it by 45 degrees/radians on the Y. How do I find out the new coordinates of the corners? especially when it comes to collision detection.

You could use trigonometric functions to calculate the new points in 3D space.

I'm no expert in 3D collision detection, I always keep my games 2D, sometimes with 3D graphics.

If you want advanced 3D collision detection then bullet: [code.google.com] might help you.