Scribd
Upload
350 views

Unproject Explained

The document explains how to use the OpenGL function gluUnProject() to convert 2D screen coordinates from a mouse click into a 3D ray in the virtual world. GluUnProject() inverts the projection calculations to determine the corresponding 3D coordinates. It must be called twice, once with the near clipping plane z value and once with the far value, to obtain the two points defining the ray. This ray can then be used to determine which object in the 3D world was selected by intersecting it with object geometries.

Uploaded by

raduman2
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
350 views

Unproject Explained

The document explains how to use the OpenGL function gluUnProject() to convert 2D screen coordinates from a mouse click into a 3D ray in the virtual world. GluUnProject() inverts the projection calculations to determine the corresponding 3D coordinates. It must be called twice, once with the near clipping plane z value and once with the far value, to obtain the two points defining the ray. This ray can then be used to determine which object in the 3D world was selected by intersecting it with object geometries.

Uploaded by

raduman2
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Unproject Explained

Now that weve broken down the math for


projections, we can explain the unproject program that was distributed earlier drag), how do we translate that events mouse coordinates into our 3D world?

The core issue is: given a mouse event (click, move, With our projection analysis, we can phrase this
question more specically now: given a set of coordinates on the 2D viewport, what are the corresponding coordinates in 3D space?

(click)

monitor/window

Observations
First off, note from the diagram that a 2D mouse click
does not translate into a point, but into a ray after all, we are adding an entire dimension 3D point; the best we can do is identify the line along which the mouse points 3D equivalent must lie line but what you do with it after that is up to you

So, we cant really go from a mouse point to a single

OpenGLs gluUnProject() function will help give you that

What gluUnProject() Does


gluUnProject() inverts the projection calculation; instead
of going from a world point to a screen point, which is what projection does
point in world ! model-view ! projection ! viewport ! point on screen

we take the screen point and go the other way:


point on screen ! viewport1 ! projection1 ! model-view1 ! point in world

With this in mind, the signature of the gluUnProject()


function should now be pretty self-explanatory:
GLint gluUnProject (GLdouble winX, GLdouble winY, GLdouble winZ, const GLdouble *model, const GLdouble *proj, const GLint *view, GLdouble* objX, GLdouble* objY, GLdouble* objZ);

gluUnProject() Double-Take
Given what we have said so far, some parts of
Why does the screen (win) point have a z-coordinate? Since the result of the function is a 3D point, the output arguments are passed as pointers; so what is that integer that the function returns directly?

gluUnProject()s signature may have you wondering:

We answer the second question rst: not all matrices Now back to that z-coordinate on the screen

are invertible thus, gluUnProject() might not succeed, in which case it will return GL_FALSE, with successful inversion returning GL_TRUE

Recall that, during the nal drawing to the viewport, we


happen to not need the z coordinate; however, as you have seen from the matrices, we do get a value for the z axisso, even though we dont use z in the nal drawing, it can (and does) get calculated winZ == 0.0 (the screen) corresponds to objZ == N (the near plane), and winZ == 1.0 corresponds to objZ == F (the far plane) call gluUnProject() twice: once with winZ == 0.0, then again with winZ == 1.0 this will give us the world points that correspond to the mouse click on the near and far planes, respectively

It turns out that, the way OpenGL calculates things,

Since two points determine a line, we actually need to

Typical gluUnProject() Sequence


Now that we know what gluUnProject() specically does, we can sketch out its general use, given some screen coordinate (mx, my):

Invert the my coordinate (since the screen y-axis goes


in the opposite direction as the 3D y-axis) view, projection, and viewport again for (mx, my, 1.0)

Grab the current values for the three matrices: model Call gluUnProject() twice, once for (mx, my, 0.0) and

Once you have the two points, what you do next now
depends on how youre representing the objects in your model intersect that line, then choose one of them as the hit object, and act accordingly two endpoints, you can represent their line in terms of a single argument u, where u = 0 corresponds to the near point, and u = 1 corresponds to the far point L(u) = nearPoint + u(farPoint nearPoint)
In the sample program, were testing against a xed plane with a known z, so we solve for u using bilinear interpolation using the z coordinates, then use u to subsequently calculate x and y; the resulting (x, y, z) is the point on the plane that was clicked on by the mouse

Generally, you would test to see which objects

Bilinear interpolation is useful here: since you know

You might also like