i want to do backface culling like this

http://processing.org/hacks/hacks:backface_culling

but without using processings camera calls and positions.

how can i do backface culling using rotation and translation matrice calls and positions as opposed to the cameras?

for example if i want to do something like

Code:

float rotationX = 0, rotationY = 0;

void setup() {
  size(300, 300, P3D);
}

void draw() {
  background(0);

  lights();

  translate(width/2,height/2);
  rotateX(rotationX);
  rotateY(rotationY);
  box(150);
}

void mouseDragged() {
  rotationX += (mouseX - pmouseX) * 0.01;
  rotationY -= (mouseY - pmouseY) * 0.01;
}
 

how can i know which face is facing the "camera"... assuming the box being drawn from a custom box class which has a method which takes a position (pvector) and calculates which face is facing a position using the faces objects centroid and normal... as was done in the hack.

The first trick is you have to know where the face actually is. After transforms and rotations the world space co-ordinate could be anywhere. A simple method is to use modelX() modelY() and modelZ() to get the actual location. From there you just check the dot product of the camera normal to the face normal and that'll give you your answer.