🔒 Locked

Mouse Interpretation

Started by Bennettovia May 4, 2003 at 4:02 AM 8 replies 800+ views

Original Post

May 4, 2003 at 4:02 AM

What I need to do is process a mouse click, figure out where that click occured in my GL space, and plot a point there. This is for a modelling tool and that plotted point would be the anchor of a line, or the anchor of a polygon etc. I''m not trying to select or pick anything. Just figure out where I clicked in GL space so that I can plot a point. I obviously have the windows location of where I clicked. Any help would own, since I''m severely frustrated, and if the solution is simple I''m going to kick myself in the head.

666_1337

May 4, 2003 at 8:30 AM

this should be in the graphics theory forum

you''re gonna need some information about your camera.
1. position in space (in GL_PROJECTION-mode)
2. orientation as normalized vector
3. the *up*-vector
3. optionally the right vector, but it can be determined by cross producing up an orientation vector.
4. the near plane of your view frustum

now, scalar-multiply the orientation-vector with the distance to the near plane. then add the mouse x (relative to the midpoint of your monitor & divided by the half of your x resolution) and the y (also relative ...). now you have the vector pointing from the camera''s origin to the mouse pointer - you''ll have to scale it to be in your view frustum. at last, transform the vector by the projection matrix, and you have the coordinate (at the end of the vector*).

*vector --- actually it''s a ray, since a vector doesn''t have a origin. so, you have to translate the camera''s position stored in the last column of the projection matrix by that vector.

Bennettovia

Author

Bennettovia

May 4, 2003 at 12:12 PM

quote:
Original post by 666_1337
this should be in the graphics theory forum

I posted here because I didn''t know if there were GL specific commands I could make use of. Thanks for your help though

666_1337

May 4, 2003 at 1:17 PM

quote:
Original post by Bennettovia
I posted here because I didn''t know if there were GL specific commands I could make use of. Thanks for your help though

hmmm... well... there is a command you could use which is opengl-specific: glGetFloatv with GL_PROJECTION_MATRIX as argument :D

Bennettovia

Author

Bennettovia

May 4, 2003 at 5:01 PM

quote:
Original post by 666_1337
[quote]Original post by Bennettovia
I posted here because I didn't know if there were GL specific commands I could make use of. Thanks for your help though

hmmm... well... there is a command you could use which is opengl-specific: glGetFloatv with GL_PROJECTION_MATRIX as argument :D

Although I haven't programmed this bit (I was sleeping) I was just going to base my coordinate translator on your above psuedo algorithm

The thing is, I'd want all the mouse clicks processed in a fixed XY plane. Essentially my window is split into four, XY, XZ, YZ and view. so any point placed in window XY would have a Z of 0.0 until it was moved in another window.....

[edited by - Bennettovia on May 4, 2003 6:05:50 PM]

666_1337

May 5, 2003 at 1:22 PM

a cad program, i see

so you are surely looking down the z axis for one of your *windows*, the y axis on another and the x axis on the last one.

then you might have something like

  
struct GLCanvas //or maybe it''s class. doesn''t care this time

 {
  int width, height, xpos, ypos;
  GLint _xleft, _xright, ... //parameters to glOrtho

  ...
 };

with xpos beeing the x distance to the left side of your main window. ypos the distance to the top, width and height are that what they are named. now if you catch a click on p(x, y) you have to do the following:
1. divide your mouseposition_x through your viewport coordinates and store the result
2. find the difference from _xleft and _xright
3. multiply the result of 1. with that one of 2.
4. add _xleft to the result of 3.
5. do the same thing with y

now you''ve got the point (x, y, 0) if it was the xy plane. else, simply interpret it as (x, 0, z) or (0, y, z)