QR Code Generator

How it works

The generator takes the exact characters in the input and passes them to a QR encoding library running in the page. The library selects a QR version large enough for the payload, applies the requested error-correction level, calculates the data and format bits, and renders the modules as a PNG. The light margin around the matrix is intentional: scanners use that quiet zone to separate the code from nearby text and graphics.

Error correction does not recover arbitrary design mistakes. It adds redundant data so a scanner can reconstruct a code when some modules are dirty or hidden, but it cannot compensate for tiny modules, insufficient contrast, a missing quiet zone, glare, or a destination that no longer works. A scan test in the final physical context is more useful than looking at the image on a bright monitor.

Choosing a payload

A QR code is a transport container, not a URL shortener, analytics system, or safety filter. It does not know whether a destination is trustworthy, and it cannot make a private URL public or a public URL private. Short stable links make less dense codes and make printed material easier to maintain. For campaigns, use a redirect you control so the destination can change without changing the artwork.

The output is a PNG, which is suitable for ordinary documents, presentations, and print layouts. Keep a copy of the original text beside the image, scan the exported file, and check the destination on both iPhone and Android before distributing it widely.

Size it from the scanning distance

Printed QR codes fail for one reason more than any other: the modules are too small for the distance people read them from.

The working rule is a ratio of about ten to one. A code scanned from one meter needs to be roughly ten centimeters across; from three meters, thirty. A poster read across a room and a business card held at arm's length are different problems, and the same artwork cannot serve both.

The constraint underneath is module size rather than overall size. A URL of thirty characters produces a small matrix whose individual squares stay chunky. The same physical square carrying a two-hundred-character payload packs in far more modules, each one correspondingly smaller, and a camera that resolved the first easily will struggle with the second. This is the practical reason to keep payloads short: not the file size, but the printed legibility.

Two things that also decide it in the real world. Matte stock scans better than gloss, because glare from a light source washes out an entire region and error correction is not designed to recover a hole that large. And curved surfaces — bottles, cups, folded leaflets — distort the matrix; a code that works flat can fail wrapped, so test on the actual object.

Higher error correction is not free

The four levels recover roughly 7, 15, 25 and 30 percent of the code. It is tempting to choose the highest and stop worrying, and that is usually the wrong call.

The redundancy has to go somewhere, and where it goes is more modules. Moving from the lowest level to the highest can push a payload into a larger QR version, shrinking every module at the same printed size — so a choice made to improve reliability makes the code harder to scan.

Match the level to the risk. Low or medium is right for anything printed cleanly on paper and scanned in reasonable light, which covers most documents, slides and signage. The higher levels earn their cost where damage is genuinely expected: labels on equipment that will be handled, codes on packaging that gets scuffed, anything outdoors, and codes with a logo placed over the center — that logo is damage, deliberately introduced, and the redundancy is what makes it survivable.

If a code needs high error correction and a long payload, shorten the payload instead. A redirect you control turns two hundred characters into twenty and solves both problems at once.