Green light lifts the NV center up; it falls back and gives off a red flash. If the spin is in the "1" state, it often takes a dark side path instead and waits there without glowing. So "1" looks a bit darker than "0", and that's how a camera can read a qubit.

Optical transitions conserve spin, but the excited-state intersystem crossing to the singlet is about eight times faster from mₛ = ±1 (91.8 /µs) than from mₛ = 0 (11 /µs). Time spent in the long-lived singlet is dark, so mₛ = ±1 emits fewer photons during the first few hundred nanoseconds. The singlet decays preferentially to mₛ = 0, which is why the same light also polarizes the spin.

Five-level rate model (mₛ = ±1 lumped at zero field) with rates from [tetienne2012]; deterministic curves by integrating the rate equations, the animation by a Gillespie quantum-jump trajectory, and the histogram by 500 trajectories with 10% photon collection. Agrees with adamas.photophysics (contrast 0.445 in a 300 ns window at saturation). Real single-shot readout needs many repetitions because of the few collected photons; that is why chapter E4 proposes photocurrent readout.