Question

I’ve created a simple 3D world using GLKit. I’m trying to find a specific point in my 3D world based on a screen tap. It seems that I would want to use GLKMathUnproject. The signature of the method is:

GLKVector3 GLKMathUnproject (
   GLKVector3 window,
   GLKMatrix4 model,
   GLKMatrix4 projection,
   int *viewport,
   bool *success
);

A few things confuse me about the params it requires. First shouldn’t the window coordinates be GLKVector2? And most importantly wouldn’t projecting a point on the screen into 3D space create a ray instead of a single point.
When you project points in 3D space onto a 2D plain (aka the screen) you loose one dimension of information about that point. So how would it be possible to transform a 2D screen point into a 3D world point without specifying at what depth you’d like to ‘unproject’ your point to? There is absolutely no comment about this method anywhere on the internet (besides apple docs).

Answer 1

The first parameter is GLKVector3 because the z component stores the desired z-depth.

So window x,y are the screen coords and window z is the desired depth.

eg with z=0 GLKMathUnproject will give the coordinates of the object at the near plane
with z=1, GLKMathUnproject will give you the coordinates of the object at the far plane.

This is necessary because GLKMathUnproject maps from a plane to a line in the 3d space formed by the projection frustum. The z – bounds of the frustum are constrained by the near and far planes.

So say you wanted to map the unprojected mouse point at the z=0 (object space) plane in the frustum.

Then unproject 2 points with window z=0 and window z =1 1.
Use these points to form a line in object space. Interesect this line with the plane z=0 to determine the final target point.

Hope this helps someone –

You can read more here CasualDistraction Blog