How to Scan a QR Code From a Screenshot

Yes, and the Route Depends on What You Are Holding

Nothing about a screenshot is hostile to a QR decoder. The pattern is the pattern, and a screen capture records it at whatever resolution your display runs at, which on a modern phone is over a thousand pixels across. Codes fail after a screenshot for one reason far more often than any other: something resized the file between the sender and you.

So the useful question is not which app to install. It is what route your device gives you for handing a file to a decoder instead of pointing a camera at the world. That route exists and is documented on 2 of the platforms below, and is documented nowhere on 4 of them.

When the Code Is on the Screen You Would Scan With

This is the version of the problem most articles skip, and it is the one people actually have. A code arrives in a message on your phone. You cannot point that phone’s camera at its own screen, and the advice to find a second phone assumes somebody else is in the room.

The move is to stop treating it as a camera problem. Turn it into a file, then hand the file to something that reads files.

1. 1Long-press the code in the app it arrived inChat apps and browsers increasingly offer this straight from the image. It costs two seconds to try and it skips every other step on this page.
2. 2Screenshot it, then decode the screenshot as a fileThe route that always exists, because it turns a screen problem into a file problem. Everything below is about that file.
3. 3Open the image in a second browser tab and decode it thereUseful when the code is on a web page rather than in a message. You keep the original resolution instead of capturing it.
4. 4Borrow a second phoneThe advice most articles give, and the one that assumes somebody else is in the room. It also fails on a bright screen at close range.

One thing worth doing before any of them: open the image at full size rather than screenshotting the thumbnail in the conversation. A thumbnail is already a shrunken copy, and capturing it hands the decoder a smaller code than the one your phone already has.

A Failed Decode Does Not Mean the Image Is Too Small

Everybody writing about this, including the sibling guide on this site, samples a handful of widths and reports a floor: above this many pixels a code reads, below it the code does not. Sampled at nine widths that is exactly what the data looks like. Sampled at every width it is not true.

Look at the campaign URL and the contact card. Both have a band where green and red interleave, and both read at widths well below something that failed. The campaign URL failed at 128 pixels wide and decoded at 72, and the contact card failed at 84 and read at 64. That is not noise in the measurement, it is the mechanism: shrinking an image decides where each module edge lands relative to the pixel grid, and at some ratios the edges land cleanly while at others they average into grey. Change the ratio slightly and the alignment changes with it.

The practical consequence inverts the usual advice. If a screenshot will not decode, do not conclude the code is unreadable and go shopping for a better scanner app. Try it at a slightly different size. Resize by ten per cent, or crop tighter around the code so the decoder gets a different ratio, and try again.

Two honest caveats. This is one decoder on clean renders with no blur, glare or JPEG noise, so the exact widths that fail belong to this decoder and this resampler rather than to QR codes in general. And enlarging a small image is still pointless, because interpolation invents pixels rather than recovering detail. What generalises is the shape of the result: success is not a straight line in size.

What a Chat App Does to the Image Before You Get It

Messaging apps and mail clients shrink images to a limit on the long edge. Your screenshot starts with plenty of pixels and arrives with fewer, and the code inside it shrinks by the same proportion. So the number that decides whether a forwarded screenshot works is how much of the screen the code filled when it was captured.

The headline is reassuring. At a 1600 pixel limit a short link still reads from a tenth of the screen width, so an ordinary screenshot of an ordinary code is fine. The trouble starts as the limit tightens and the payload gets denser. A campaign URL with tracking parameters needs a quarter of the screen width to survive an 800 pixel limit, and a code that small on screen is common in a slide deck or an email footer.

Which gives two rules worth following when you send one. Crop to the code before you send it, because cropping spends the whole pixel budget on the part that matters instead of on your notification bar. And send the original file rather than a screenshot of it whenever you can.

That second rule sounds like folklore, so here is the number behind it. Each round below shrinks the previous copy by a fifth, the way a screenshot of a forwarded screenshot does. Beside it, the same code is resized from the original straight to that width in one step.

Density decides how much re-saving you get away with. A campaign URL and a contact card both die on the sixth round, at 105 pixels wide, while the same 105 pixels reached in a single resize still decodes. The pixel count is identical. Only the history differs, and the history is what killed it. So when somebody sends you a screenshot that will not read, asking for the file they screenshotted is not pedantry, it is the fix.

Where the image came from decides how much of that damage has already happened before you touch it.

• WhatsApp, Telegram and other chat appsThe version sitting in the conversation is usually a preview rather than the file. Screenshotting that preview resizes an already resized copy.Tap the image to open it full screen first, or save it, and work from that.
• EmailAn inline preview is often a scaled-down render of an attachment that is much larger.Download the attachment rather than screenshotting the message body.
• Instagram, Facebook and other social appsThe platform recompresses on upload, so the poster's crisp code is not the one you are looking at.Open the post at the largest size the app offers before you capture, and ask the poster for the file if it still fails.
• A PDFScreenshotting a whole page gives the code a small share of the capture, which is exactly the case measured above.Zoom until the code fills the window, then capture just that.
• A videoMotion blur, and whatever the player is currently rendering rather than the resolution the video was uploaded at.Set playback quality as high as it goes first, then pause, step to the sharpest frame and capture that.

The File Route, Device by Device

Camera scanning and reading a saved file are separate capabilities, and platforms are much better at the first than the second. The scanner guide compares what each device can do with its camera. What follows is only the file route, and only where the vendor documents one.

• Android

  Documented route

  Open Chrome, tap the Lens icon in the search bar of a new tab, then pick the image instead of taking one. Google's own wording for that picker is to select from your screenshots or images, which is exactly this job.

  Source: Google, Search with Google Lens in Chrome (Android)

• iPhone and iPad

  Documented route

  Same route, different button. In Chrome on iOS, open Lens from a new tab and tap Photo picker at the bottom left to use an image you already have. Apple's own Camera app handles codes in front of you, not codes in your camera roll.

  Source: Google, Search with Google Lens in Chrome (iPhone and iPad)

• iPhone, straight from Photos

  Nothing documented

  Long-press the code in the screenshot and see whether a link appears. This works on some iOS versions and not others, Apple does not document it, and behaviour has moved between releases. Five seconds tells you which version you have.

• Windows 11

  Nothing documented

  Nothing built in reads a QR code out of a file. Snipping Tool has Text actions, which is optical character recognition for text and has no barcode step. The Camera app can read a code you hold up to the webcam, which is a different job from the screenshot already sitting on your disk.

  Source: Microsoft, Use Snipping Tool to capture screenshots

• Mac

  Nothing documented

  No documented route either. Preview and Quick Look detect text in images, and Apple does not document them detecting QR codes. Try a long press or a right click on the code first, then fall back to something that reads files.

• Linux

  Documented by the project

  One command, offline, no account, and it prints the decoded string to your terminal. Documented by the zbar project rather than by a vendor, and we have not run it on this machine, so treat it as their claim rather than our measurement.

  zbarimg screenshot.png

• Chrome and Firefox on the desktop

  Nothing documented

  Google documents Lens on desktop Chrome for searching images and page regions, and does not mention QR codes in that documentation. Firefox has nothing for this at all. A page that decodes in the tab is the gap both of them leave.

  Source: Google, Search with Google Lens in Chrome (Computer)

The pattern in that list is the finding. Both mobile platforms have a documented way to decode an image you already have, and both desktop platforms have none, which is why the search results for this question are full of app downloads. A page that decodes in the browser tab works the same way on all five, and you can paste a screenshot into ours without installing anything.

On a Computer, the Clipboard Is the Short Way

Everything above assumes a file. On a computer you can skip the file entirely, and given that neither Windows nor macOS documents a way to read a code out of one, this is the fastest route either of them has.

Capture straight to the clipboard, then paste into something that decodes. Nothing touches your disk, so there is no screenshot left in a folder afterwards holding whatever the code contained.

• Windows

  Capture to clipboard

  Windows key + Shift + S

  Then paste

  Ctrl + V

  Snipping Tool opens a crosshair and puts the crop straight on the clipboard. Drag a box around the code, leaving a margin of blank space, then paste. Nothing is ever saved to disk.

  Source: Microsoft, Use Snipping Tool to capture screenshots

• Mac

  Capture to clipboard

  Control + Shift + Command + 4

  Then paste

  Command + V

  Apple's wording is to hold Control while you press the other keys, and that is the part that sends the capture to the clipboard instead of the desktop. Without Control you get a file on the desktop, which also works, just with an extra step.

  Source: Apple, Take a screenshot on your Mac

The reader on this site listens for a paste anywhere on the page and takes a dropped file too, so the whole job is two keystrokes: capture, paste, read the result.

The One Format That Will Not Open

Almost every image format you will meet is fine, and the exception catches people out because it only bites in one specific case.

• PNGAcceptedWhat every screenshot on every platform already is. Lossless, so nothing is thrown away between the screen and the decoder.
• JPEGAcceptedWhat a photo becomes once it has been through a chat app or an email. Lossy, and each re-save costs a little more edge definition.
• WebPAcceptedCommon when an image was saved off a web page, and on the accepted list here. Lossy like JPEG, so the same warning about re-saving applies.
• GIF and BMPAcceptedRare for this job, both on the accepted list. We have not tested an animated GIF, so if you have one, save a single frame as a PNG rather than finding out.
• HEICNot acceptedThe default for photos taken on an iPhone, and the one gap. Screenshots are not affected because iOS writes those as PNG. For a photo, share it as JPEG, or open it full screen and screenshot that.
• PDFNot acceptedNot an image at all. Zoom the page until the code fills the screen, then screenshot it. Zooming first is the whole trick, because it hands the decoder more pixels per module rather than fewer.

The HEIC line is worth reading twice, because it is narrower than it looks. A screenshot on an iPhone is a PNG, so screenshots never hit this at all. Photographs are the HEIC ones, which means the case that actually breaks is photographing a code on somebody else’s screen and then feeding that photo to a reader. Screenshot the photo and you have a PNG again.

Two Codes in One Image Read as None

A screenshot often catches more than you meant, and the intuitive assumption is that a decoder picks one code and ignores the rest. It does not.

This one is worth knowing because of how the failure presents itself. You get told no code was found, on an image that visibly contains two, so the natural conclusion is that the image is broken. It is not. Crop to one code and it reads immediately.

The Screenshots You Should Not Upload Anywhere

Most pages offering to scan a QR code from an image ask you to upload it. Before you do that with a screenshot, consider what a screenshot contains, because for several common codes the pattern is not a pointer to a secret, it is the secret.

• A two-factor setup codeThe QR code in an authenticator setup screen contains the shared secret itself. Anyone holding that string can generate your codes indefinitely, and rotating it means re-enrolling the account.
• A boarding pass or event ticketThe barcode is the credential. A readable copy is enough to check in, and the name and record locator next to it are usually in the same screenshot.
• A Wi-Fi code from an office or a home routerThe passphrase is inside the pattern in plain text. Decoding is not a hint, it is the whole password.
• A payment request or a wallet addressPayment codes carry an account identifier and often an amount. Circulating one invites a plausible looking request that is not yours.
• Anything with the rest of the screen still in itA screenshot is the whole screen. Notifications, a browser tab title, an email subject line and a phone number all come along with the code you meant to share.

A Wi-Fi code is the clearest case, because the passphrase is inside the pattern as plain text rather than as a reference to something stored elsewhere, which the Wi-Fi guide shows by decoding the payloads back. The answer is not to be careful about which server you trust. It is to not involve one. Decoding is arithmetic on a bitmap, and a browser can do all of it in the tab you already have open, which is how the reader on this site works. The tool page explains what enforces that rather than only promising it.

Check the Local Claim Instead of Believing It

This page and the reader it points at both claim your image stays in your browser. That claim is worth nothing without a way to check it, so here are three, in descending order of how much they actually prove.

1. 1Read the security headerRun the command below against any page on this site and look at the first directive.default-src 'self' with no connect-src override means connect-src inherits it, so the browser refuses any connection this page tries to make to another origin. That is enforcement rather than a promise, and you did not have to trust us to see it.
2. 2Pull the plugLoad the reader, turn off Wi-Fi, then decode an image.That decoding needs no network once the page has loaded. It does not prove much about a page you have already visited, because the browser may have cached it, which is the honest limit of this test.
3. 3Watch the network panelOpen developer tools, switch to the network tab, then decode an image and watch what appears.That no request carrying your image goes out. Worth knowing what it cannot tell you: a page can decode locally and still send analytics, so an empty panel is a stronger result than a busy one is a problem.

curl -sI https://qrmakery.com/tools/scan-qr-code-from-image | grep -i content-security-policy

Run that and the first directive is default-src 'self', with no connect-src of its own. Connections inherit the default, so the browser blocks this page from opening a connection to anywhere but this site. That is the part worth having: it holds whether or not we are honest, because it is the browser enforcing it rather than us describing it.

Being straight about the limit, though. A same-origin request is not blocked by that header, so the policy rules out sending your screenshot to somebody else rather than ruling out all traffic. The reason your image goes nowhere is that the decoding runs in the tab, and the header is what stops any future edit from quietly changing that.

When It Still Will Not Read

One case is not a failure at all. If the image decodes and the destination then leads nowhere, the screenshot did its job and something on the other end expired, which do QR codes expire takes apart properly.

If the image is a fair size, holds one code, and still will not read, the problem has moved from the copy to the code. Two other pages here cover that ground properly rather than in a paragraph.

A cropped quiet zone, weak contrast between light and dark modules, or a logo covering too much of the grid are all faults in the code itself, measured with thresholds in why a QR code will not scan. If the code is in front of you on paper or a screen rather than in a file, camera scanning is a different job with different answers, and the scanner guide covers what each platform can already do without an app.