reset password
Author Message
cliffordtzlin
Posts: 11
Posted 14:33 Jun 10, 2014 |

Hi everyone,

On the last page #5, it asks the end points of the lines in CVV. Does that just mean we determine the projection matrix and multiply it to the endpoints of the lines? Thanks for your time.

Clifford Lin

eykang
Posts: 95
Posted 14:58 Jun 10, 2014 |

Yes. x, y, z values of the point in CVV should be in the ordinary coordinate system. So make sure to perform perspective division to get the values.

 

 

cliffordtzlin
Posts: 11
Posted 15:10 Jun 10, 2014 |

Thank you.

For #5, we don't have to use perspective division yet right? Isn't perspective division done after clipping within CVV has been completed?

eykang
Posts: 95
Posted 15:27 Jun 10, 2014 |

Again, CVV itself is defined in the ordinary coordinate system, not Homogeneous coordinate system. So you have to perform the perspective division to see x, y, and z values in CVV.

Perspective transformation (multiplication by the projection matrix P followed by perspective division) transforms object in the camera coordinate system into CVV.  For efficiency, clipping is inserted between "multiplication by the projection matrix P" and the "perspective division" such that the clipping algorithm takes advantage of taking points defined in the homogeneous coordinate system as inputs to identify trivial reject/accept. The clipping algorithm does internally perform perspective division to calculate intersection points. Those trivial accept points will be converted to CVV points after the clipping.