The lines that weren't there on the page

You hold a magazine page up to the light and it looks perfect: a glossy photograph, smooth skin tones, a sky that fades cleanly from pale to deep blue. You scan it. And on the screen, draped across that same photograph, is something that was never on the paper — a faint plaid of wavy lines, a shimmer like heat coming off a road, a rippling grid that seems to move when you scroll.

You didn't do anything wrong. You didn't shake the phone or smudge the lens. The strange pattern has a name — moiré — and it isn't a flaw in your scan so much as a collision between two grids that were never meant to meet. Understanding why it happens is the whole trick to making it stop, and it turns out to be one of the more elegant bits of optics hiding inside an everyday task.

A printed photo is not a photo

Start with a fact that surprises most people: the smooth photograph in your magazine is an illusion. Printing presses cannot lay down a continuous fade from light to dark. They have ink or they have paper — on or off, with nothing in between. To fake the shades your eye reads as "smooth," printers use a centuries-old trick called halftoning: they break the image into thousands of tiny dots of a single ink, making the dots larger in the dark areas and smaller in the light ones. Step back far enough and your eye blends them into tone. Lean in with a magnifying glass over any magazine photo and you'll see it — a fine, regular field of dots, usually tilted at a precise angle.

That dot field is a grid. A very regular, very fine grid, printed at something like 133 to 175 rows of dots per inch, each color of ink screened at its own careful angle so the four grids of a color image don't clash with each other. The printer spent real effort making that grid invisible to your eye at reading distance. The problem is that your scanner has a grid too.

Two grids, one interference

A camera sensor or a scanner bar is also a grid — a fixed array of light-sensing pixels, evenly spaced, sampling the page at regular intervals. So now you have two regular patterns laid over each other: the printed dot screen and the sampling grid of the sensor. When two fine, regular grids overlap at slightly different spacings or angles, they produce a third, coarser pattern that belongs to neither of them. That third pattern is moiré.

You've seen this in the physical world. Two window screens held together and shifted slightly throw up rolling dark bands. A striped shirt on television crawls with color. Overlapping chain-link fences make ghostly diamonds. In every case, nothing is actually there — the pattern is an interference effect, the visual beat frequency of two rhythms that don't quite line up. Scanning a halftone image is exactly this, in miniature, thousands of times across the page.

The deeper reason has a name from signal theory: aliasing. A sensor can only faithfully capture detail up to a certain fineness — roughly, it needs to sample at least twice as often as the finest pattern it's trying to record. This is the Nyquist limit, and it is unforgiving. When the printed dot screen is finer than your sensor can resolve, the sensor can't record the dots honestly. Instead it records a lie — a lower, false frequency that masquerades as real structure. The wavy lines are that lie made visible. The information was too fine to keep, so the grid invented something coarse in its place.

Why this is different from blur or glare

It's worth separating moiré from its cousins, because the fixes are opposites. Blur comes from too little sharpness — the lens or the motion smeared the detail away. Glare comes from stray light bouncing off a glossy surface into the lens. You fight both by getting more faithful, sharper, cleaner capture.

Moiré is the strange exception where more fidelity makes things worse. A cheap, soft scan of a magazine photo often shows no moiré at all, because it never resolved the dot screen in the first place — it blurred the grid into the tone your eye wanted anyway. A crisp, high-resolution scan is the one that catches the screen at just the wrong scale and lights up with ripples. That's why people are baffled: they upgrade their setup, scan more carefully, and the artifact appears. It's not a failure of care. It's a side effect of resolving a grid you were never supposed to see.

What actually makes it stop

Because moiré is a beat between two grids, you break it by changing the relationship between them — or by softening one grid until it can't beat against the other.

Change the angle. The simplest, most reliable move. Rotate the page a few degrees relative to the camera or scanner bar — ten or fifteen degrees is plenty — then straighten the image afterward with software. Tilting the page shifts the interference so the two grids no longer line up into visible bands. This is the single trick that saves the most scans.

Change the distance or resolution. Because moiré depends on the ratio between the dot screen and the sampling grid, nudging that ratio often makes the pattern collapse. Move the phone slightly closer or farther; try a different resolution setting. There's frequently a sweet spot where the ripples simply vanish.

Soften the finest grid before you sample it. This is what professional scanning software calls descreening: a gentle blur applied specifically to smear the halftone dots back into tone before the pattern can alias. It sounds like sabotage — deliberately throwing away sharpness — but remember that the dots were never meant to be seen. Blurring them away doesn't lose the picture; it restores the smooth tone the printer was faking all along. A light blur, then a modest re-sharpen of the actual image edges, and the photograph reads clean.

Fix it after the fact. If moiré is already in a scan, it lives at a specific spatial frequency, which means a targeted blur or a frequency-domain filter can often lift it out while leaving text and edges mostly intact. It's rarely as clean as avoiding it at capture, but it can rescue a page you can't reshoot.

One quiet note: this whole problem belongs to printed images — magazine and newspaper photographs, glossy brochures, anything made of a halftone screen. Original photographic prints, ordinary typed text, and handwriting have no dot grid to beat against your sensor, so they don't moiré. If you're only scanning the article's words, you may never meet the effect at all. It's the pictures that ripple.

The pattern was always a compromise

There's something almost poetic in it. The printer built a grid of dots to fool your eye into seeing smoothness. Your scanner built a grid of pixels to capture the world faithfully. Two honest illusions, each perfectly good on its own, produce a third thing that is purely an artifact of their meeting — visible proof that every image you've ever looked at is a negotiation between what's really there and what a machine can record. Once you can see moiré for what it is, you stop treating it as a defect and start treating it as a signal: these two grids are fighting; move one.

This is the kind of small, physical understanding that makes scanning feel less like guesswork. LumenScan handles the page itself — sharp, corrected, with on-device OCR that keeps your documents on your phone rather than shipping them to someone else's server — but it can't override physics, and neither can any app. What it can do is give you a clean capture and let you re-scan in seconds when a magazine photo ripples, so the tilt-and-retry trick takes a moment instead of a chore. If you'd like a scanner that respects both the page and your privacy, LumenScan is at https://lumenscan.lumenlabs.works — and now, at least, the wavy lines won't be a mystery.