QR Code Finder Pattern: The Three Corner Squares

The short answer

The finder patterns are the three nested square rings you see in the top-left, top-right, and bottom-left corners of every QR code — the recognizable “eyes” that make a QR look like a QR. Each one is a 7×7 block of modules built from a 3×3 solid dark square inside a 5×5 dark ring, surrounded by a one-module light gap, and the dark and light runs along any line through it follow a fixed 1:1:3:1:1 ratio. That ratio is what a scanner looks for to find the code in an image and work out its rotation.

There are exactly three, not four, because the missing fourth corner is reserved for the alignment and timing patterns the scanner uses to compute the code’s skew and sampling grid rather than to locate it. You can restyle the eyes for branding, but the 1:1:3:1:1 ratio must hold or the scanner never locks on — the trade-off our qr code styling and qr code design guides treat as the central rule of eye customization. This page is the deep dive; our how does a QR code work guide covers the whole scan pipeline.

What a finder pattern actually is

Each finder pattern is a fixed 7×7 module structure that looks the same in every QR code regardless of what data it carries. Working from the outside in: a one-module light separator, then a 5×5 dark ring (a dark square outline one module thick), then a one-module light gap inside the ring, then a 3×3 solid dark square at the center. Drawn on a grid it is a dark hollow square inside a dark square outline, floating on the light background.

The reason this exact shape was chosen is its ratio. If you draw a horizontal or vertical line through the center of the finder, the dark and light runs you cross follow the pattern 1 module light, 1 module dark, 3 modules dark, 1 module dark, 1 module light — written as the 1:1:3:1:1 ratio. That sequence is long enough and specific enough that it almost never appears by accident in ordinary images, which is what lets a scanner search the whole camera frame for it and reliably find the code. The 3-module-wide center bar is the giveaway: three dark modules in a row, bounded by single dark modules with single light gaps on each side.

Why there are exactly three (not four)

A QR code has three finder patterns, in the top-left, top-right, and bottom-left corners. The fourth corner — bottom-right — is deliberately left without one. Three non-collinear points are enough to define a square in 2D and fix the code’s position and orientation, and leaving the fourth corner free makes room for the alignment pattern (a smaller 5×5 nested square the scanner uses to correct skew and perspective) and the timing pattern (the alternating dark/light line between finders that tells the scanner the module spacing).

If a fourth finder sat in the bottom-right, the scanner would have no room for the alignment pattern that handles distorted, tilted, or photographed-at-an-angle codes, and real-world scans (a poster shot from below, a curved label on a bottle) would fail far more often. Three finders locate and orient; the alignment and timing patterns in the fourth corner let the scanner map the rest of the grid accurately. That division of labor is the whole reason the code reads at any angle and tolerates perspective distortion.

How a scanner uses the finder patterns

When you point a phone camera at a QR code, the scanner does not look for a square of dots. It scans the whole camera frame for the 1:1:3:1:1 ratio — three dark modules between single dark modules with single light gaps — in both the horizontal and vertical directions. Each place that ratio appears is a candidate finder. When the scanner finds three of them arranged as two corners across the top and one down the left, it knows where the code is, how it is rotated, and which way is up.

From the three finders the scanner computes the module size and grid spacing (using the timing pattern that runs between them), corrects any skew with the alignment pattern in the bottom-right, and then reads every data module in a fixed zig-zag path. The finder patterns are read first and are not protected by Reed-Solomon error correction — there is no redundancy to repair them. That is why covering a finder with a logo or a frame is fatal: the scanner never finds the code, so it never gets to the data. Our qr error correction guide covers the redundancy that protects the data area; the finders have to survive by being intact.

The customization trade-off

Because the finder patterns are the most visible part of a QR code, they are the first thing a brand wants to restyle — rounded eyes, circular eyes, colored eye rings, a separate color for the eye pupil. All of that is fine, and QRForge exposes per-eye shape and per-eye color for exactly that reason, but the 1:1:3:1:1 ratio must hold. As long as the eye still reads as a dark square-within-a-square with the right proportions against a light gap, the scanner can still lock on; the moment the shape breaks the ratio (a ring so thick the center bar is gone, a pupil so light it reads as background, an eye stretched into a rectangle), the locate step fails.

The practical rules: keep the eye pupil as dark and high-contrast as the body of the code, keep the dark/light/dark structure intact even when you round or color it, and always scan-test the eyes on a real phone after restyling. A colored eye ring with a dark pupil on a light background reads well; a low-contrast eye that blends into the background drops the locate step before a single data module is read. The qr code colors guide treats the finder contrast as a first-class rule because the eyes need the strongest contrast on the whole code — more on that below. Our qr code with logo guide covers the separate rule for centered logos, which live in the data area, not on the eyes.

Finder patterns at a glance

The table below summarizes the parts of a finder pattern, what each does, and the customization rule that keeps it scannable.

How QRForge lets you style the eyes safely

QRForge exposes per-eye shape (square, rounded, circle) and per-eye color as live controls in the free generator, so you can brand the eyes without leaving the tool — and the preview renders live so you can scan-test on a real phone before you export. The guardrail is structural: as long as the eye still reads as a dark square-within-a-square with the 1:1:3:1:1 ratio against a light gap, the scanner can find it, and the default settings keep that ratio by construction. If a custom eye color drops contrast, the fix is almost always to darken the pupil or pull the ring back toward the body color, then re-test.

When you add a centered logo, keep it in the middle of the data area with level H (our qr code with logo guide covers the 20–22% width rule), and leave the three finder eyes and the four-module quiet zone untouched. Run the final exported SVG, PNG, or EPS through the free /decode helper to confirm the payload round-trips, and pair the eye check with the contrast and quiet-zone habits in our qr best practices guide. The eyes are the code’s handshake with the scanner — keep their ratio intact and the rest of your styling is free to roam.