A real wheel spins with a single continuous angular velocity, but a camera (or the retina sampling a strobe) only records its orientation at discrete instants. If the wheel's spoke pattern moves more than half a spoke-spacing between two captured frames, the brain matches each new frame to the nearest spoke position instead of the true one — spokes appear to slow, stop, or spin backwards. This is temporal aliasing, the same math behind audio sample-rate aliasing.
f_true = RPM / 60 (true revolutions/s)
f_pattern = f_true * N_spokes (pattern repeats N times/rev)
f_alias = f_pattern - round(f_pattern / fps) * fps
apparent_RPM = (f_alias / N_spokes) * 60 (negative = reversed)
Separately, "persistence of vision" is the ~40-100 ms window over which the human visual system keeps integrating a stimulus after it disappears. If that persistence window is at least as long as the interval between frames (1000/fps), successive still frames overlap in perception and blur into continuous motion; if it's shorter, each frame is seen as a discrete flash.
- True wheel speed — the wheel's actual, continuous rotation rate.
- Capture rate (fps) — how often a new orientation is sampled; in Sampled mode the wheel visibly jumps only at each tick.
- Persistence of vision — how long each captured frame's afterimage (glowing trail marker) lingers before fading, simulating retinal persistence.
- Continuous / Sampled — toggles whether the wheel mesh shows the true smooth motion or the stroboscopically held frames.
Film cameras historically chose ~24 fps because it is cheap on film stock yet, combined with ~40-100 ms persistence, still reads as fluid motion to most viewers — while the classic "wagon wheel" reversing-spoke artifact reveals the underlying sampling process.