Mouse to 3D space?

[Varkas]

I'd assume that it should be possible to use the projection matrix "backwards", and calculate a line in 3D space from the 2D mouse coordinate on the screen. Intersecting this line with the objects it should be possible to find the object(s) which was pointed at.

Is this the way to do it? And if yes, how does one do the math? Matrix operations are nontrivial for me, so I will ned a bit of help with that. Thanks

[ra4king]

Have you tried asking the all-knowing Google?

[sproingie]

The term that you might find handy to google for is "ray picking".  A "simple" way to get that ray is with a couple calls to GLU.gluUnproject, using your near and far planes as Z values.  There's probably better ways to do it now though.

Games published by our own members! Check 'em out!

[Agro]

It is actually really easy. this function will return a vector in the direction you are pointing in, assuming you have a yaw and pitch value(you can't just get it from a single coordinate, you need some sense of direction too):

1   2   3

Vector3D getDirectionVector() {    return new Vector3D(-Math.cos(pitch * Math.PI / 180.0) * Math.sin(yaw * Math.PI / 180.0), Math.sin(pitch * Math.PI / 180.0), -Math.cos(pitch * Math.PI / 180.0) * Math.cos(yaw * Math.PI / 180.0)); }

Note: This is assuming yaw and pitch are in degrees.

Tell me if it didn't work, might've messed up something there.

Then after that, you can use ray tracing or some other good  method to find the object(s) you're pointing at.

[ra4king]

Try to keep pitch and yaw as radians to save performance.

/nitpicky

[Agro]

Oh yeah, always use radians. Degrees is just better for debugging/presentation honestly, but personally I like it better. >

[Varkas]

Have you tried asking the all-knowing Google?

I guess I was rather feeling for a forum posting yesterday. But it seems easy enough, I think I can find a way. If I remember right I already had solved that problem in an very old 3D game attempt of mine, and maybe I can revive that code (the vector from eye position to mouse-on-projection-surface position should be just the ray I was looking for. It guess it was more obvious when I did the projection in my own code ...)

Thanks for the responses, and sorry, next time I'll try to google first.