A surprising amount of the world communicates by blinking. Routers report their state in LED patterns. Appliances flash error codes. Cheap LED bulbs pulse invisibly fast — until they give you a headache. Fans, drills and wheels spin at rates that matter and that nobody can count by eye. All of these are periodic light signals, and a camera is, among other things, a device that samples light at a steady rate. Point one at the other and you have a timer.

That's the Blinking tool: aim the camera at a blinking light and it times the blink rate, or at a spinning object and it estimates the rotation frequency. The capture watches how brightness in the frame changes over time, finds the period, and reports a frequency — a number you can file, compare and trend, instead of a rhythm you tried to count under your breath.

The flicker you can't see but your head can feel

Start with the use case that surprises people most. Many LED bulbs and displays dim by PWM — pulse-width modulation. The light isn't actually dimmer; it's switching fully on and off hundreds of times a second, and your visual system averages the pulses into "dimmer." Averages them consciously, that is. Plenty of people get eyestrain, headaches or a vague wrongness from low-frequency PWM, especially from cheap bulbs that flicker at twice the mains frequency, and they usually never find out why — the flicker is invisible by design.

A camera doesn't average the way a brain does. Sampled by a sensor, PWM flicker shows up as a periodic brightness signal the tool can time. If the suspicious bulb over your desk flickers at 100 Hz and the one you replaced it with is genuinely steady, that's now a measured difference between two objects in your catalog, not a hunch. Measure the bulbs in a store display before buying, and the headache-machine identifies itself.

Blink codes: the LED is trying to tell you something

The second family is slower and more literal. Routers, chargers, dishwashers, printers and industrial equipment all report status through indicator blink patterns, and the manuals speak in timing: "slow blink (once per second) — connecting; fast blink (twice per second) — error." Human beings are bad at distinguishing 1 Hz from 1.5 Hz, particularly while irritated at a router. The camera isn't. Point, capture, read the rate, look it up. And because the measurement files under the object — this router — the blink rate from last month's outage is still there when the same light starts flashing again.

Rotation: how fast is that fan actually spinning?

A spinning object modulates light too. Fan blades sweep past and the reflected brightness at any point rises and falls once per blade; the periodicity in the frames encodes the rotation. The Blinking tool uses the same time-domain analysis to estimate a spinning part's rotation frequency: the ceiling fan on "medium," the PC case fan you suspect is dying, the workshop motor whose nameplate speed you'd like to sanity-check. A failing fan often slows before it stops — two measurements a month apart, filed under the same object, catch the trend.

The honest limit: aliasing

Now the caveat, because it's a real one and it shapes the whole design. A camera samples light at its frame rate — commonly 30 or 60 frames per second. Sampling theory is blunt about what that buys you: a signal can only be timed directly if it changes slower than half the sampling rate. Blink faster than that and the camera doesn't see "too fast to measure" — it sees a wrong, slower rhythm. This is aliasing, the same effect that makes wagon wheels roll backwards in old films and makes a fan under fluorescent light seem to hover.

So the tool is honest about its window. Indicator blink codes at a few hertz: comfortably inside. Slow PWM and mains-related flicker: reachable, with care about the interplay between the flicker rate and the frame rate. A drill bit at 30,000 RPM — 500 revolutions per second: far outside anything a 60 fps sensor can time directly. The frame rate is the ruler, and you can't measure past the end of the ruler.

The design rule at work: a measurement tool must know — and say — where its physics stops. The camera's frame rate is a hard sampling limit, so the Blinking tool claims only the frequencies it can time unambiguously rather than reporting an aliased number as if it were true. A confident wrong reading is worse than an honest "out of range"; the aliased region isn't hidden, it's handed to a different technique.

Beyond the frame rate: the Strobe experiment

What's on the far side of that limit? This is where aliasing stops being only a nuisance, because there's a classic trick that turns it into the instrument: the stroboscope. Illuminate a spinning object with light flashing at exactly its rotation rate and the object appears frozen — every flash catches it in the same position. Machinists have used dedicated strobes to read RPM this way for a century.

A phone torch can flash. So the research builds of Sentio include a Strobe tool that sweeps the torch's flash rate and looks for the stroboscopic freeze, estimating a spinning or vibrating object's frequency from where the freeze occurs. To be plain about its status: this is a research-build experiment, not a shipped consumer feature. Torch hardware varies wildly in how fast and how consistently it can pulse, and the approach needs validation across devices before it can make consumer-grade claims. But it's the intellectually satisfying counterpart to the Blinking tool — the exact phenomenon that limits passive timing, driven deliberately, becomes the way past the limit. Like every capture in the platform, the analysis runs on device, and research captures can be labelled against known-RPM references afterward, which is precisely the data that validation needs.

Filed, not memorized

The common thread with every tool here: the point isn't the single reading, it's the reading in context. A blink rate means something next to the manual's table; a flicker frequency means something compared across bulbs; a fan RPM means something against last month's. Each capture files under an object in the catalog and can take a ground-truth label once you know the outcome — the bulb that fixed the headaches, the fan that did die. Single glances become records, and records become the comparisons that were the point all along.

The camera has quieter modes too — measuring color and ambient light rather than timing it, covered in Color and light. And for the full picture of what phone sensors can measure, from microphones to accelerometers, start with the tools overview.