Coordinate Spaces
Follow one vertex from model space to the pixel it lands on, one stage at a time.
What you should come away with: Where each matrix in the chain hands over to the next, and what the GPU does between them.
1 / 6 · Model space
- frustum
- tracked vertex
One vertex, all the way down
| Space | x | y | z | w | |
|---|---|---|---|---|---|
| Model | 0.50 | 0.50 | 0.50 | 1.00 | |
| World | 1.23 | 0.80 | -0.79 | 1.00 | |
| View | 1.23 | 0.78 | -3.71 | 1.00 | |
| Clip | 1.85 | 1.88 | 3.27 | 3.71 | w now carries depth |
| NDC | 0.50 | 0.50 | 0.88 | — | ÷ w |
| Screen | 719 | 149 | — | — | pixels |
One corner of the cube, followed the whole way. Watch w pick up the depth at the clip stage — every row after it is that same point divided by that number. The frustum you see in the canvas is not drawn separately: it is the camera’s own frustum pushed through the identical pipeline, which is why it lands exactly on the −1…1 cube at the NDC stage.