PDA

View Full Version : point vector rotation



mphanke
03-25-2002, 12:50 PM
glHi,

I have a point in space and a vector pointing towards another point. I want to rotate an object about the resulting axis.
Could somebody please send me a snippet on how to do that?

Thanks very much,

Martin

[This message has been edited by mphanke (edited 03-25-2002).]

Thug
03-25-2002, 03:30 PM
Uhhhh.

use glRotatef(angle, Vx, Vy, Vz) where V is the vector to the other point and angle is the angle to rotate around that axis/vector.

mphanke
03-26-2002, 12:13 AM
Hi, thanks for that reply but what I have to compute is the physical movement of the object. What I need is the mathematics behind that glRotate().

I'll try to make it more clear:

--------
I*<---*I <- this is where I want to rotate
I V I the object
I I
I I
--------

glGreets,
Martin

[This message has been edited by mphanke (edited 03-26-2002).]

satan
03-26-2002, 02:07 AM
it is all very basic vector/matrix math
(you may also use quarternions, but personally i do not think they are good for anything, 'cause you can achieve the same effects using matrices)

take a look here and you will find all you need:
http://mathworld.wolfram.com/topics/Rotations.html

if this is not good enough for you search google for 'rotation arbitrary axis'

Michael Steinberg
03-26-2002, 10:13 AM
Thanks satan for pointing out this resource of information.

Thug
03-27-2002, 06:35 AM
Are you looking for the location where you will be in the next frame from the axis of rotation and the axis of rotation.

I was working on the same problem last weekend. I am in the process of testing it but my system is doing some funky stuff... I try to open any program and it says that it is "Not responding" in the running process under win98. I am also getting an error that my GL program is still running when I had terminated the program. So I have to reboot my system before I can recompile the program again. I konw the code is fine. I was compiling it on my sisters compter last weekend and it was working fine.

Anyway I am using a local coordinate system where I have the vector Z_UP <0,0,1> and I have another vector that contains the direction on the XY plane (Proj <cos(angle), sin(angle), 0> ). I do not care about the length of the vector. I then find the cross product between the Z_UP and Proj. This produces another vector perpendicular to the direction on the XY plane. I then take the crossproduct of the vector from the previous cross product and my rotation axis (my true up vector of the object). Thus giving me the direction that I am heading in my local coordinate system. Now all I have to do is change the vector's magnitude to the proper distance that I went in that time frame and add the vector to my last position in world X, Y, Z coordinates.

Hope this is clear. I was pulling it all from memory and remember I have not fully tested it but in theory it should work. NOTE: there are special cases since we took Z_UP to be the vector <0,0,1>. Doing a crossproduct with another vector that is paralell to Z_UP or lies in the XY plane. will have some problems with the above.

I also have know idea about how fast this algorithm is compared to multiplying matrixies together.

billy
03-27-2002, 06:43 AM
Cool site satan!! http://www.opengl.org/discussion_boards/ubb/smile.gif

mphanke
03-27-2002, 12:20 PM
glHi,

thanks for your help, I solved my problem with your help and some of my old math books. I have alot of funcs done. So if someone would like to get the source of rotation funcs, throw me some lines I will mail it asap!

glGreets,

Martin