Point a phone camera at the tip of a television remote and press a button. On the screen, a pale violet flare pulses out of the little dark dome. Lower the phone, press the same button again while watching directly, and nothing happens—at least nothing you can see. The remote is fine and so are your eyes. The handset is talking in near-infrared, centered near 940 nanometers, and that wavelength sits well past the deepest red your retinal cones can report. Camera sensors have no such deadline; they keep registering light far beyond it.
This gap between what eyes and sensors receive is not an exotic laboratory condition. It is the ordinary operating principle behind remote controls, security cameras, and automatic door sensors, and you can watch its boundary from your couch. The same mismatch between eye and detector is why night photos grow color speckles that never appeared in your visual field: a camera's record is never simply a copy of what a person saw.
A wavelength past the edge of the rainbow
Human color vision is a committee of three cone types, each with its own tuning. Their combined sensitivity spans roughly 400 to 700 nanometers, and it is not symmetrical: response to deep red is already fading at 700 nanometers and disappears shortly after. A remote's 940-nanometer emission is not a faint red at the horizon of vision. It is beyond the horizon entirely. No cone signals, so no color experience follows—it is not "invisible red," it is no hue at all. Vision-science references such as the Webvision chapters on photoreceptors lay out these cone fundamentals in detail, and they make plain how abruptly the system ends.
That ending looks like a design flaw until you consider the alternatives. The sun floods the ground with near-infrared, and much of it is heat radiation rather than useful image information. Building receptors there would add glare without adding much signal about fruit, faces, or cliffs. Evolution kept vision inside a narrower, information-dense band, and infrared was left to snakes, beetles, and eventually engineers.
The border itself has a strange geography. Just below the visible edge, in the deep red from about 700 to 750 nanometers, sensitivity fades gradually rather than clipping off, so some people can just barely sense intense light in that range as a dim cherry glow. By 940 nanometers, no such luck: the pigment chemistry of the cones simply has nothing left to absorb with. When engineers pick wavelengths for gadgets, they choose deliberately on either side of that fence. Remote controls sit safely deep in the invisible zone so they cannot disturb anyone's color perception; remote-control designers, in other words, are doing color planning in reverse, selecting light specifically to be unseen.
Silicon keeps listening
The chips inside digital cameras are made of silicon, and silicon's photoelectric behavior is more generous than your cones. Reviews of silicon photodetectors describe sensitivity stretching from the near-ultraviolet out toward roughly 1,100 nanometers, which brackets 940 comfortably. To a bare sensor, your remote's LED is not a whisper; it is a searchlight.
Camera makers therefore work hard to un-hear it. Color accuracy depends on the sensor responding only to wavelengths a human would call colors, so serious cameras place an infrared-cut filter over the chip to block the band. Rear phone cameras filter aggressively, which is why the demonstration often fails there. Front-facing cameras, doorbells, and security cameras leave more leakage on purpose or by economy, and that is exactly where the trick works: enough near-infrared survives to register, and the flash appears on screen.
Why the flash reads pale violet
People usually describe the remote's flash as whitish pink or pale lavender, which raises a puzzle: near-infrared has no color, so where does the tint come from? The answer is that the sensor is improvising. Its red-filtered, green-filtered, and blue-filtered pixel sites do not switch off cleanly at 700 nanometers; their sensitivities trail into the infrared with different slopes. The camera's processor then maps that unbalanced signal into the nearest plausible visible color, and the result tends toward a washed-out violet-pink rather than a pure red. The exact tint varies between devices and manufacturers, so treat it as an artifact of the pipeline, not a property of the light.
A ten-second diagnostic that actually works
The remote trick has practical value because the LED is doing all the signaling. A handset that seems dead but still flashes brightly on camera is transmitting normally—your problem lies with the receiver or the pairing. A faint, sluggish flash usually means dying batteries, visible on the sensor before the eye could ever judge. And if you film a security camera at night, the scattered purple pinpoints you see are its infrared illuminators, working a shift your eyes were never scheduled for.
The same near-infrared neighborhood runs a surprising amount of the background world. Automatic faucets and paper-towel dispensers watch for reflected infrared pulses. Night-vision doorbells flood your porch with light you cannot see, then photograph what bounces back. Some phone face-unlock systems project tens of thousands of invisible dots to map a face in three dimensions. None of these gadgets is "glowing" in any human sense; they are all simply working a wavelength to which you hold no pass. The demo with a remote is just the cheapest ticket into that hidden layer—and unlike a security camera, a remote will sit still for you while you check.
Noticing exactly where visible light ends is also good practice for the color exercises on this site. Games such as Spot the Different live entirely inside your eye's narrow band, rewarding the smallest differences a human visual system can register—the same band your remote just stepped outside of. One screen, two worlds: yours, and the machine's wider, colorless lease on the spectrum.