Read a QR code from an image

See where a link goes before you tap it

This is the difference between here and your phone camera. The camera shows a preview banner and one careless tap opens it. Here the raw string sits still on screen for as long as you want to look at it.

Codes stuck over the real one on a parking meter, on a package insert, on a flyer taped to a door — this is a routine scam now, and the trick nearly always hides in the URL. The JSON view separates the host from everything else, which is where a lookalike domain shows itself, and lists each query parameter on its own line, which is where a redirector buries the destination it actually sends you to.

Be clear about the limit: this page tells you what the code says. It does not tell you whether the site behind it is safe. Recognizing the domain is still your job.

When nothing is found

Failures are almost always about resolution or contrast. Before decoding, the image is scaled down to 1600 px on its longest side. So a 4000 px photo where the code occupies one corner loses most of the detail the decoder needs — crop to the code, leave a white margin around it, and try again. That one fix solves the majority of failed reads.

• Shot at an angle, or wrapped around a bottle or wristband — the grid has to be roughly square-on
• Motion blur, or a dim photo where the modules blend together
• Photographed from a monitor: glare and moiré wreck it, so take a screenshot instead
• A logo or sticker covering more of the code than its error correction can rebuild
• Pale ink on white — too little contrast to locate the modules at all

What the JSON view recognizes

Nothing is interpreted for you beyond being split into parts, and the original text is always kept as-is under the Text tab.

• Web links — host, path, and every query parameter listed individually
• WIFI: strings — network name, encryption type, password, hidden flag
• mailto:, tel: and SMSTO: — address, number, and any prefilled message
• vCard and MECARD contact cards — flagged as a contact, kept raw so nothing is lost
• EMV-style payment payloads — identified, with the full payload preserved character for character
• JSON that was already stored inside the code — parsed and re-indented instead of left on one line
• Anything else — handed back as plain text, unchanged

The camera option, and why you may not see it

There is a live camera mode, but the checkbox for it only appears on a phone: it needs a real camera, a touch screen, a small viewport and a secure connection. On a laptop the option is hidden on purpose, because webcam frames of a printed code decode badly and the retries are more frustrating than useful.

So if the code is on paper and you are at a desktop, take a photo with your phone and drop the photo here. Everything decodes locally either way — the camera frames are read in the page and never leave the device, exactly like a dropped file.

One more asymmetry worth knowing: a dropped image is tested in both polarities, so white-on-dark codes decode fine. The live camera only looks for standard dark-on-light codes, to keep the frame rate up.