QR Code Generator

Yes. Every type, every design option, the logo, the frames, the scan check and all five download formats are free, with no account, no sign-up, no watermark and no limit on how many codes you make. The codes never expire, because there is nothing to expire — the data is inside the symbol.

No. The QR code is generated, styled, checked and exported by JavaScript running in your browser. Nothing is sent to a server, nothing is stored and nothing is logged. You can disconnect from the network after the page loads and it keeps working — which matters more here than almost anywhere, because the things people put in QR codes are WiFi passwords, home addresses, phone numbers and wallet addresses.

Never. These are static QR codes: the data is encoded in the black and white squares themselves. There is no server in the middle, no account attached, and nothing that stops working when a company changes its pricing or shuts down. A code printed today will scan in twenty years.

A static code holds your data directly. A dynamic code holds a short link to somebody's server, which redirects to wherever you point it — so it can be edited after printing and it can count scans, and it also stops working the moment that service does, or the moment you stop paying. This tool makes static codes only, and does not pretend otherwise. If you need to change a destination later, point a static code at a URL you control and change the redirect at your end.

Yes — PNG, JPEG, SVG or WebP, resized and centred automatically, with an optional backing plate so it reads cleanly against the modules. A logo destroys the modules underneath it, and the error correction rebuilds them up to its limit. That is why the check below the preview matters: it actually decodes the finished image and tells you whether the correction managed it, rather than showing a tick and hoping.

Between 10% and 30% of the width. At level H the error correction can rebuild about 30% of the symbol, so a 20% logo is comfortable and a 30% one is on the edge — and the edge is a bad place to be, because a phone at an angle in poor light has less to work with than the perfect image being checked here. If the check says it scans and the coverage warning is showing, it means exactly that: fine on screen, risky in the wild.

How much of the symbol can be lost and still read. L recovers about 7%, M about 15%, Q about 25% and H about 30%. Stronger correction takes up room, so the same data needs a bigger symbol — which is the trade. Use M for a screen, Q for print, and H for anything with a logo on it or anything that will be handled, folded or left outdoors.

It is checked, not guessed at. The finished image — colours, module shapes, logo and all — is drawn to a canvas, read back a pixel at a time into a module matrix and put through a real QR decoder with genuine Reed–Solomon error correction. If the data that comes out is the data that went in, it scans. The warnings underneath cover what a decode cannot see: contrast, quiet zone and printed size, all of which matter to a camera in bad light and not at all to perfect pixels.

Yes. Enter the network name, the password and the security type, and scanning it offers to join the network without anyone typing anything. Semicolons, colons, commas and quotes in the password are escaped for you — an unescaped one is the usual reason a WiFi code appears to work and then fails to connect, because the scanner reads the password as ending early.

WPA covers WPA, WPA2 and WPA3, which is nearly every network made this century. WEP is there for genuinely old equipment and comes with a warning: it has been broken since 2001 and many current devices refuse to join a WEP network at all. Open is for networks with no password, and the password field is ignored.

Yes. Enter a UPI ID, optionally a payee name, amount and note, and it produces a upi://pay link that Google Pay, PhonePe, Paytm and BHIM all recognise. Leaving the amount empty lets the payer type it, which is what a shop counter usually wants. Check the UPI ID against your own app before printing anything — nothing here can verify it, and a payment code with a typo sends money to a stranger.

Yes, for Bitcoin, Ethereum, Solana and USDT. Each chain has its own URI scheme and this uses the right one — BIP-21 for Bitcoin, an amount as `value` for Ethereum. USDT is the exception and gets a bare address, because USDT exists on a dozen networks and its address alone does not say which; a URI naming one would send funds to the wrong chain. Verify every address in your own wallet before you print it. Nothing here talks to any chain.

A contact card. Scanning it offers to save the name, company, phone, email, website and address straight to the phone's address book, which is what makes it the most useful thing to put on a business card. Commas and semicolons in the fields are escaped, because both are structural in the vCard format and an unescaped comma in a company name splits it into two fields.

The usual rule of thumb is a tenth of the scanning distance: a code read from a metre away wants to be about 10 cm across. Below about 2 cm printed, most phones struggle regardless of resolution. In pixels, aim for at least three or four pixels per module — the panel shows you how many you have at the current size, and warns when it drops too low.

SVG for anything going to print or into a design tool: it is vector, so it scales to a billboard with no loss. PNG for the web and for anywhere transparency matters. WebP for the smallest file that still has transparency. JPEG only if something demands it — it has no alpha channel, so a transparent background gets flattened, and its compression puts soft halos around the module edges. PDF goes through your browser's print dialog, which produces a real vector page.

They multiply the pixel dimensions by one, two and four. Standard is the size you chose, which suits a screen. HD suits a retina display or a small print. Print gives you enough resolution that a code placed at 300 dpi still has crisp module edges. It makes no difference to SVG, which has no pixels to multiply.

Yes — foreground, background and the three corner eyes separately, plus a transparent background. What matters is contrast: a scanner works from a camera image in poor light and needs far more separation than a screen does. Aim for a ratio of 7:1 or better, keep the modules darker than the background, and check the panel — it measures the ratio and says when it is marginal.

You can, and it is legal. A good many scanners still refuse it, because the specification describes dark modules on a light background and plenty of implementations took that literally. The tool warns when the foreground is lighter than the background. If the design demands it, test on more than one phone before committing to print.

Appearance, and they cost a little reliability. Square fills each cell completely and is the most robust. Rounded is nearly as good. Dots, circular and diamond shrink the mark inside its cell, which leaves gaps between neighbouring modules — it looks softer and gives a camera slightly less to work with, especially in print. The check will still tell you whether the result reads.

Because they are how a scanner finds the code at all. A reader looks for three concentric squares with a 1:1:3:1:1 ratio of dark to light, and those proportions are what identifies a QR code in a camera frame full of other things. The eye styles here change the corner radius and never the proportions, which is the difference between a styled code and one that quietly stops being found.

The blank border around the symbol. The specification asks for four modules, and it is not decoration: it is what tells a scanner where the code stops and the rest of the world starts. Cropping it — putting a code hard against a box edge, or against a busy photograph — is the single most common reason a printed QR code fails, and it fails after ten thousand have been printed.

At the very largest size and the weakest correction, 7,089 digits, 4,296 letters and digits, or 2,953 bytes of arbitrary text. In practice much less is a good idea: more data means more modules, smaller modules at the same physical size, and a harder scan. A long URL is better shortened, and a paragraph is better linked to.

Yes. Anything outside the digits-and-capitals set is encoded as UTF-8 in byte mode, which is what every modern scanner assumes and what makes accents, Cyrillic, Arabic, Japanese and emoji all come out the other side intact. Plain digits and plain uppercase get denser encodings automatically, which is why a phone number produces a much smaller code than the same number of letters would.

Three ways: the image to your clipboard as a PNG, the SVG markup as text to paste into a design tool or a page, and the raw data the code holds — useful for checking exactly what a scanner will see, or for pasting into something else.

Ctrl/Cmd+Shift+C copies the image, Ctrl/Cmd+Shift+D downloads it in the chosen format, Ctrl/Cmd+Shift+Delete resets everything, and Ctrl/Cmd+F searches the types and templates. There is no generate shortcut that does anything, because the code is rebuilt as you type.

Once the page has loaded, yes. Everything runs locally — the encoder, the renderer, the decoder that checks it and the export. A connection is only needed to load the page the first time.