A phone camera is a grid of millions of tiny light meters. That's all a camera sensor is: photons in, numbers out. Everything else — the tone curves, the white balance, the "computational photography" — is software working hard to make those numbers not look like measurements, because measurements make ugly photos. Two tools in Sentio go the other way. They turn the automation off and let the sensor be what it physically is.
Color: what shade is this thing, actually?
The Color tool captures the object's color and brightness as a spot measurement. Point at a patch of surface, capture, and the reading files into your catalog as numbers rather than a photo you'll squint at later.
The use cases are the small, recurring annoyances of physical life:
- Paint matching. You're at the hardware store and the wall is at home. A measured reading of the wall beats "sort of a warm gray, I think."
- Produce color as a ripeness cue. Bananas, tomatoes and mangoes announce their state chromatically. Measure the same fruit daily and the trend is the signal — a companion to the acoustic tap test, which listens for what color can't show.
- Comparing batches. Two dye lots, two print runs, two mixes of the same stain — are they the same, or does one drift warmer?
- Fading over time. Measure the sun-facing side of a fabric or a painted fence monthly, under the same conditions, and you get a fade curve instead of a vague sense that it "used to be brighter."
Why does this need a tool at all, when the camera app is right there? Because the camera app is your adversary here. Auto white balance exists to make every scene look neutrally lit, which means it actively erases the color cast you're trying to measure. Auto exposure rewrites brightness frame to frame. So the Color tool locks auto-exposure and auto-white-balance during capture — the same discipline every level-sensitive tool in the platform follows. If the light levels must stay honest, the camera must stop "helping."
There's one more trick available: the torch. Ambient light is the biggest confound in color measurement — the same wall reads differently at noon and at dusk. The torch can serve as a known illuminant: it's always the same LED, at a fixed position relative to the lens, on every capture. Light the object with the torch and you've replaced "whatever light happened to be around" with a repeatable one, which is most of what makes two readings from different days comparable.
Light: is this room actually bright enough?
The Light tool flips the question. Instead of measuring an object under light, it samples the ambient brightness itself over time, using the camera as a light meter.
People reach for this when a vague impression needs to become a number:
- Is this desk bright enough? Workspace lighting is something everyone has opinions about and nobody measures. Sample the desk, sample the desk by the window, and the argument resolves itself.
- Comparing rooms and window orientations for plants. "Bright indirect light" is the most-cited and least-defined phrase in houseplant care. Measure the north sill, the east sill and the shelf across the room — same tool, same phone, same hour — and you can rank them instead of guessing.
- Tracking light over a day. Because the tool samples over time, a spot doesn't have to be judged by one moment. The morning-blasted, afternoon-dark windowsill and the steadily dim corner can average to the same impression while being very different places to live, if you're a fern.
The same capture discipline applies: with exposure locked, more light means brighter frames in direct proportion, and the sensor behaves like the light meter it physically is. With auto-exposure left on, the camera would fight to make every room look identically well-lit — which is exactly the difference you came to measure.
The design rule underneath both tools: build on relative features, never absolute levels. An absolute reading depends on the sensor, the lens, the device model — a comparison between two readings taken the same way on the same phone cancels all of that out. This is the platform-wide principle that lets features travel between devices without per-device calibration, and it's why every claim these tools make is phrased as a comparison.
The honest caveat
Here is what a phone camera is not: a certified lux meter or a laboratory colorimeter. Those instruments are calibrated against traceable standards, and their absolute numbers mean something in court. A phone camera's absolute numbers mean something on that phone, with that lens, under that firmware.
So the reliable claim is comparative. "The east sill is roughly twice as bright as the north sill, measured the same afternoon" — solid. "This wall drifted warmer over six months of torch-lit readings" — solid. "This desk is exactly 480 lux" — that's the claim to be suspicious of. Comparisons between measurements of the same setup, taken the same way, are what the instrument honestly supports, and the tools are built to make exactly those comparisons easy.
Where the readings go
A comparative instrument is only as good as its memory, and human memory is terrible at brightness and hue. That's why every capture files under an object in the catalog — this wall, this windowsill, this ficus — so the fade curve and the light survey accumulate instead of evaporating. And like every measurement in the platform, each reading can receive a ground-truth label after the fact: the paint match that turned out right, the sill where the plant actually thrived. The reading cost seconds; the label makes it a data point about what the numbers predicted.
Everything runs on device. The analysis happens on your phone, the catalog lives on your phone, and no capture needs a network round-trip to become a number.
Color and light are the camera at its most passive — sensing what's there. The same sensor can also work in the time domain, timing blinks and estimating rotation: that's the Blinking tool. For the full map of what phone sensors can measure, start with the tools overview.