QR Code Reader (from an image)

How it works

If you already have a picture of a QR code — a screenshot, a photo of an invoice, a shot of a poster — choose the file and then tap or click "Convert". The tool scans the image, pulls the text out of the code and shows both the content and its kind in the result line: a link, Wi-Fi details, a contact card or plain text. The full content can also be downloaded as a text file, which is where you see every line of a long contact card. Large photos are scanned at a reduced size first and, if no code is found, once more at full resolution, so codes printed very small are still caught. If nothing is found you get a message explaining what to check: the whole code must be inside the frame, there should be no shadow or glare on it, and the photo should be sharp. When the content is a web address, the result line also reminds you to check that address before opening it. The image is never uploaded; the reading happens inside your browser.

This tool is also known as qr code reader, read qr code from image, qr code from screenshot, decode qr code, scan qr code from photo, qr scanner image.

What is QR code?

A QR code is a two-dimensional mark made of small black and white cells laid out in a square. A reader first finds the three large corner patterns, works out the angle and scale of the code from them, then turns the cells into a string of bits and rebuilds the text. That is why a code that is tilted, partly shaded or slightly creased usually still works: reading it is a piece of geometry rather than a straight scan.

What is Quiet zone?

The quiet zone is the empty margin left on all four sides of a QR code, at least four cells wide according to the standard. It is how a reader tells where the code ends; if the code is dropped into text, a border or a patterned background with no margin, reading fails again and again. When an image will not read, the culprit is very often not the code itself but this margin, destroyed while cropping.

What is Error correction margin?

A mathematical backup of the data is built into every QR code, and thanks to it the content can be rebuilt even when part of the code is dirty, torn or covered by a logo. How much can be recovered ranges from seven to thirty per cent, depending on the level chosen by whoever generated the code. The reader has no say in this: how much resilience your code was given was decided by the person who printed it.

What is the difference between reading and generating a QR code?

Generating is a one-way, lossless job: the text you type is turned into cells by fixed rules, and the result is identical every time. Reading is guesswork; the picture in your hand may be tilted, shadowed, compressed or low resolution, so the reader has to find where the code is, straighten its angle and estimate where each cell boundary falls. That is why generating a code always works while reading one depends on the quality of the image. Reading back a code you generated yourself is a good final check before it goes to print.

When is this useful?

A camera app can only read a code that is in front of it. Yet a QR code is often not on paper but on your own screen: an invoice that arrived by email, a screenshot, a picture lifted from a PDF, a poster you saw on social media. You cannot hold your own screen up to your camera; this tool fills that gap.

The second use is safety. Looking at a QR code tells you nothing about where it leads. Feed a suspicious code into this page, read its content as text first, and only then decide whether to open it.

The third is record-keeping: the same route works when you want to confirm the campaign address on an old poster, or check which text the label you printed yourself actually carries.

What to do when an image will not read

When reading fails, the problem is almost always the picture itself. Work through these in order:

• Are all four corners of the code in the frame? A code cropped at the edge will not read.
• Is the white margin around the code still there? That margin is part of the code.
• Is there glare, a shadow or a crease across the photo?
• Is the image too small? A code under 200 pixels rarely reads.
• Can you take a screenshot instead of photographing a screen? It comes out much cleaner.

What does the kind of content mean?

A QR code really only carries text; the reason phones offer to "join the network" or "add to contacts" is that the first few characters of that text match a known pattern. This tool looks at the same patterns and names the result: starting with http makes it a link, WIFI makes it network details, BEGIN:VCARD a contact card, tel a phone number.

If the pattern is not one it knows, the kind comes out as plain text. That is not a failure; the code may genuinely contain free text. Saying that it does not recognise the pattern is more honest than inventing a label.

Frequently asked questions

Does it use my camera?

No. This tool only reads the image file you choose; it never asks for camera permission and never turns the camera on. If the code is on a piece of paper in front of you, your phone's own camera app is the quicker route.

Is my image uploaded anywhere?

No. The image is decoded and scanned inside your browser; the file never reaches a server. The address, password or personal detail inside the code stays on your device as well.

Will a blurry or tilted photo still read?

Usually, yes. The reader finds the corner patterns of the code and corrects the perspective itself, so a slight tilt and moderate blur are not a problem. It fails when a corner of the code falls outside the frame or the photo is heavily out of focus.

What if there are several QR codes in one image?

The tool reads the first code it finds and returns its content. To read several codes separately, crop the image and give the tool one file per code.

Can I trust the link inside the code?

Not on its own. A QR code gives no clue about where its address leads, and sticking a fake code over a real one is a common scam. That is why the tool shows the address as text and reminds you to check it before opening: make sure the domain really belongs to the organisation you expect.