Depth & Transparency
Make z-fighting happen on purpose, then fix it from the frustum — and find out why transparency needs sorting.
What you should come away with: That depth precision is set by the near plane rather than the model, and that the depth buffer cannot answer the question transparency asks.
Assumes: Projection & the Frustum. It will still make sense without it, but that one comes first.
Two panels 0.010 apart, 40 units away · near 0.10
What the depth buffer can tell apart
The prediction above is arithmetic, not a guess: a 24-bit depth buffer distributed hyperbolically resolves roughly z²(f−n) / (n·f·2²⁴) at distance z. Set the separation below that number and the panels must fight; set it above and they cannot. Move the near plane and watch the prediction and the picture change together.
Almost all of that precision is decided by the near plane, because the buffer spends most of its range close to the camera. The far plane is the control everybody reaches for first, and dragging it from 10 to 1000 barely moves the number. That is why the fix for z-fighting is so rarely in the model.