Freezing motion with a torch: the strobe experiment

Wagon wheels in old westerns sometimes roll backwards. Everyone has seen it; the reason is one of the prettiest facts in measurement. Film samples the world twenty-four times a second, and a spoked wheel spinning at almost-but-not-quite a multiple of that rate appears to crawl — forwards, backwards, or to stop dead. The effect is called aliasing: sample a periodic motion too slowly and it masquerades as a different, slower motion. It's the same phenomenon Claude Shannon formalised in his 1949 sampling-theorem paper, wearing a cowboy hat.

A stroboscope is aliasing on purpose. Flash a light at a spinning machine, and when the flash rate exactly matches the rotation rate, every flash catches the machine at the same position — it appears frozen. Read the flash rate off the dial and you've read the machine's RPM without touching it. Mechanics have tuned engines this way, and factories have inspected running machinery this way, for a century.

A phone has a flashable torch, a camera, and a very good clock. So: can it be a stroboscope? Sentio's Strobe tool is the experiment that finds out — and it ships flagged as research, because the honest answer so far is "partly, slowly, and with a caveat that deserves its own chapter."

How the phone runs the experiment

The tool doesn't ask you to hunt for the freeze by hand. It runs a fixed protocol:

• First, three seconds of baseline with the torch steady, while the camera measures how much the image changes frame to frame. A spinning fan under constant light produces constant motion blur and flicker — that's the baseline level of "visual activity."
• Then the torch steps through a ladder of flash rates — 0.5 to 8 Hz — holding each for three seconds while the camera keeps watching.
• For each step, it computes a freeze score: how much the frame-to-frame change dropped relative to baseline. A strobe rate that locks onto the motion (or a multiple of it) makes the image measurably stiller. The best-freezing rate, if any step clears the bar, is the candidate answer.

One detail matters more than it looks: the tool never trusts the flash rate it asked for. Torch hardware has sluggish, vendor-dependent switching latency, so the tool measures the flicker it actually delivered — reading its own strobe back off the camera frames — and a step only counts if the delivered rate confirms within tolerance. The commanded number is a request; only the measured number is a measurement. This self-identifying-stimulus discipline is shared with the two-phone tools, where the receiver likewise verifies the stimulus rather than assuming it.

The factor-of-two problem, or: a frozen image is not an answer

Here is the caveat that keeps this tool experimental, and it isn't a phone limitation — real stroboscopes have it too. A wheel spinning at 4 Hz freezes under a 4 Hz flash. It also freezes under a 2 Hz flash (you see every second revolution) and a 1 Hz flash (every fourth). And if the object is symmetric — a two-bladed fan looks identical after half a turn — it freezes at 8 Hz as well. A frozen image tells you the flash rate divides evenly into the motion's period, not what the period is.

Professional practice resolves the ambiguity by protocol — find a freeze, halve the flash rate, check whether it still freezes single-image, repeat — and the phone tool inherits the ambiguity without yet automating the resolution. So its result is worded with unusual care: possible freeze, with a note that the number may be a harmonic or submultiple of the true rate. That phrasing isn't legal cover; it's the measurement's actual epistemic state.

Why only 8 Hz, when fans spin at 20?

A real stroboscope flashes a xenon tube or LED array at up to hundreds of flashes per second with microsecond precision. The phone's torch was built to be a flashlight: it turns on and off through a driver stack with latency on the order of a hundred milliseconds, inconsistently across vendors. Push it faster than ~8 Hz and the "square wave" you asked for degenerates into mush. That caps the current tool at slow rotations — ceiling fans on low, cement mixers, signage motors — and is the honest reason the tool is research-flagged rather than pitched as a tachometer. The faster path, flashing the screen instead of the torch, is the same trade the glazing tools make, and it's where this experiment goes next if the data says it's worth it.

Meanwhile the same camera, used passively, already estimates faster rotation a different way — watching the blur and brightness ripple of the spinning object itself, no strobe involved: that's the Blinking tool, and for rates within the frame rate's reach it's the more practical instrument today.

What the professionals carry

For where phone measurement competes outright and where it defers to instruments like the humble laser tachometer, see the honest accounting; for the rest of the sensor map, the tools overview.