Photos of Screens Add Rainbows That Were Never There

The bands are not on your television. Look straight at the screen: clean. Raise your phone: shimmering waves of color roll across the photo, and they move when you nudge the camera a centimeter closer. What you are seeing is a disagreement between two grids. The screen is a lattice of pixels; the sensor is a lattice of photosites; when one regular pattern samples another, their small mismatch amplifies into large visible beats—moiré. The rainbow tint arrives as a bonus from the sensor's color architecture. The same pipeline surprises people when a violet flower photographs blue, but this artifact is stranger: the camera isn't misreading a real color, it is inventing one.

Sampling always has an opinion

Any device that measures a continuous scene through a fixed grid must decide, at every photosite, what single value represents the light falling across it. When the scene contains detail finer than or comparable to the grid spacing—like another screen's pixels—the sampling no longer averages innocently. It starts generating structure at low spatial frequencies that exist in neither original signal. Signal processing texts treat this as aliasing, and analyses of digital camera color filter arrays work out the frequency-domain consequences in detail: the sensor's red, green and blue sub-mosaics each sample the scene at slightly different effective positions, so the aliased beats land differently in each color channel. Gray beats become colored beats.

That is the deeper reason screen moiré is so stubbornly colorful. A monochrome sensor would give you wavy gray bands. A Bayer-pattern sensor gives each band a hue that depends on how the beat aligns with the red-green-blue sub-grid, which is why the colors shimmer and swap as you tilt the phone. Classic studies of engineered moiré fringes show the same principle in reverse: deliberately overlaid gratings let you dial the fringe pattern, including its apparent color structure, purely by geometry.

Why your phone sometimes wins

Not every screen photo ripples. The beat pattern depends on the ratio between the two grids, so it appears at particular combinations of distance, zoom and orientation and fades at others. Phone makers also fight moiré on your behalf: optical elements and processing tuned to suppress aliasing, plus aggressive multi-frame software that treats static banding as noise to be smoothed. Dedicated cameras with strong anti-aliasing filters show less moiré at a slight cost in sharpness—a deliberate trade, not a defect.

A field guide to clean captures

Because the artifact is geometric, the fixes are geometric. Change your working distance in small steps and watch the bands widen, fade and re-form—each shift moves the grid ratio to a new beat. Zoom with your feet rather than digital zoom, which resamples the image and re-mixes the pattern afterward. Rotate the phone a few degrees; screens and sensors are both at their most grid-like when perfectly aligned, and a small tilt breaks the symmetry. And if the material allows it, capture several frames while content changes slightly on screen, then keep the one where the pattern is least visible. And if you are documenting a screen for someone else—a repair ticket, a design review—remember that the bands in your photo are evidence about two devices, not a defect of the display alone.

A test that shows the machinery

Here is that grid-on-grid beating, demonstrated directly above. Two fine stripe patterns, each unremarkable alone, overlay to produce broad waves that neither stripe contains—exactly the relationship between your sensor and your screen. When you notice a small difference between two nearly identical things, you are doing the visual version of the same detection task as Spot the Different; here the "different" thing is generated on the fly by interference of patterns.

The same meeting, elsewhere in the house

Once you know the recipe—two fine patterns plus a small mismatch—you start meeting moiré far from photography. Two window screens layered a few centimeters apart weave phantom waves across the glass. A scanned halftone print, itself a grid of ink dots, beats against the scanner's sampling lines—a problem serious enough that the prepress industry built countermeasures around it. Woven fabric photographed at a distance develops shimmering bands that catalog photographers fight with small angle changes and careful focus. In each case, no single object contains the pattern you see; it exists only in the relationship between two regular structures, which is why changing either one—rotating a screen, refocusing a lens, sliding a window screen—rewrites it instantly.

Engineers also use the effect deliberately. Overlapping gratings turn tiny relative movements into large fringe shifts, which is precisely the principle behind the engineered-fringe studies cited above: a beat pattern too coarse to miss, amplifying displacements too fine to see. The artifact in your screenshot and the instrument in a calibration lab are the same physics with opposite intentions—a tidy reminder that in imaging, a pattern is never noise or signal until someone states what is being measured.