Walk your screen through eighteen fullscreen patterns, spot dead and stuck pixels with your own eyes, and try to revive a stuck one with the flashing fixer. Works on monitors, laptops, phones and televisions; nothing to install, nothing sent to a server.
Before you start, turn the brightness all the way up and darken the room. Wipe dust and fingerprints off the surface with a dry microfibre cloth; dirt looks exactly like a faulty pixel.
If you would rather not start from the beginning, click the pattern you want and fullscreen opens right there.
The fixer draws coloured dots that change dozens of times a second inside the box you place. In people with photosensitive epilepsy, flashing images can trigger a seizure. Do not use this section if you have an epilepsy diagnosis, if you are sensitive to light, or if flashing images give you dizziness, headaches or nausea.
Because each dot is drawn independently, the average brightness of the box stays constant; even so, do not stare at it from close range for long. Start it, turn to something else, and glance back now and then. If you feel unwell, press Esc and leave immediately. Do not let children use it unsupervised.
The fixer switches a stuck pixel's sub-pixels on and off rapidly to try to shake them loose. Drag the box over the faulty pixel and leave it running for at least ten minutes, or an hour if needed. A dead pixel will not come back this way.
What the browser will tell you about the panel. The true resolution comes from multiplying CSS pixels by the device pixel ratio; that is the number printed on the box.
The measurement counts how many frames the browser draws in a second. If the tab is in the background, battery saver is on, or the browser caps the frame rate, the result can come out below the panel's real figure.
The browser could not enter fullscreen. The test still works; press F11 to hide the browser's address bar.
Turn the brightness all the way up, darken the room and wipe the screen with a dry microfibre cloth. Dust and marks look exactly like faulty pixels, so start on a clean surface.
Start the test and the browser interface disappears, leaving the screen a single colour. Move through the eighteen patterns with the arrow keys, the space bar or a click, scanning the whole surface on each one.
A black dot on the white screen is a dead pixel. A bright coloured dot on the black screen is a stuck pixel, and the red, green and blue screens tell you which sub-pixel is at fault.
Read the warning, confirm it, then start the fixer, drag the flashing box over the faulty pixel and leave it running for at least ten minutes. A dead pixel will not come back this way.
Every pixel on the screen is made of three sub-pixels: red, green and blue. Every colour you see comes from mixing those three at different brightnesses. What the fault is called depends on which state the three sub-pixels are stuck in, and that distinction decides both your chance of a repair and your warranty claim.
A dead pixel is a point where all three sub-pixels are permanently off. It shows as a black dot on a white field and does not change whatever you display. The cause is usually the transistor driving that pixel failing outright, which makes it a hardware fault that software cannot solve.
A stuck pixel is a point where one or two sub-pixels stay permanently lit. On a black field it shows as a red, green, blue, yellow, cyan or magenta dot. Here the transistor is still working and has merely locked at the wrong voltage, which is why there is a chance of fixing it.
| Symptom | Dead pixel | Stuck pixel |
|---|---|---|
| On a white screen | A black dot | Usually invisible |
| On a black screen | Invisible | A bright coloured dot |
| When the colour changes | Stays exactly the same | Stays exactly the same |
| Cause | The transistor has failed completely | A sub-pixel is locked at the wrong voltage |
| Can software fix it? | No | Sometimes, and it is worth trying |
| Does it clear up on its own? | No | It can, over days or weeks |
Rule out a third possibility as well: a speck of dust, a dried water spot or a dried splash on the screen looks exactly like a faulty pixel. Telling them apart is easy — if the dot glints as you tilt the screen, or moves when you nudge it gently with a fingernail, it is dirt and not a pixel. Wiping the surface with a dry microfibre cloth before you start heads this mistake off.
The test has eighteen screens and their order is not arbitrary. Flat colours come first, because a faulty dot is easiest to see against a single colour. Ramps and patterns follow; those reveal faults in the panel as a whole rather than in one pixel.
| Screen | The fault you are looking for |
|---|---|
| Black | Stuck pixels, backlight bleed and clouding in the corners. |
| White | Dead pixels, dust shadows and dirty screen effect. |
| Red, green, blue | Faults in a single sub-pixel. If one sub-pixel is dark, a dark dot appears in that colour. |
| Cyan, magenta, yellow | Two sub-pixels lit together, which is where a leak from the third one becomes obvious. |
| Grey tones | Panel uniformity, staining and IPS glow. |
| Greyscale ramp | Colour banding. If a ramp that should be smooth shows steps, the panel or the cable is running below eight bits. |
| Colour spectrum | Coarseness and breaks in the colour transitions. |
| Contrast steps | Lost detail at the black and white ends. If the two darkest steps are indistinguishable, the brightness is set too low. |
| Colour bars | Colour calibration and edge sharpness. |
| Grid | Places where line thickness breaks down, which points to scaling errors and edge softening. |
| Sharpness | A one-pixel checkerboard. At native resolution you see even grey; if it looks blurred or wavy, the image is being scaled. |
| Geometry | Whether the rings stay circular. Squashing towards the corners points to scaling or lens distortion. |
Inside the box you place, the fixer gives every dot its own random colour and repeats that dozens of times a second. The aim is to force the locked sub-pixel to switch on and off continuously so the liquid crystal starts moving again. Because the dots are independent, the screen as a whole does not pulse in sync, which is both easier on the eye and gentler on the panel's power circuitry.
This method can work on a stuck pixel and will not work on a dead one. In a dead pixel there is no circuit left to drive, so no amount of waiting changes anything. On a stuck pixel there is no guarantee either: some come back within ten minutes, some are unchanged after an hour, and some never recover.
Put the box over the faulty pixel and leave it running for at least ten minutes, or forty-five minutes to an hour if needed. If nothing changes, take a break of a few hours and try once more. Pressing on the screen, heating it or rubbing it is not advisable: those do not rescue the pixel and can leave a permanent mark on the panel and void the warranty.
Not every manufacturer replaces a panel over a single faulty pixel. The yardstick the industry uses is ISO 9241-307, which sorts panels into four classes and sets how many faults each class allows per million pixels. Faults themselves fall into three types: permanently lit pixels, permanently dark pixels, and individual faulty sub-pixels.
| Class | Allowance per million pixels | Where you find it |
|---|---|---|
| I | No faults allowed at all | Medical imaging and made-to-order panels |
| II | Two lit, two dark, five sub-pixels | Most professional monitors |
| III | Five lit, fifteen dark, fifty sub-pixels | Most consumer monitors and laptops |
| IV | Fifty lit, a hundred and fifty dark, five hundred sub-pixels | Almost never used in practice |
Most panels sold in consumer products are class II or class III. Some brands add their own commitment on top and promise a zero-faulty-pixel guarantee on particular product families; that promise comes from the brand's own campaign, not from the standard. Whatever the class, testing the device the moment you buy it and reporting any faulty pixel straight away puts you in a stronger position. A photograph taken on day one is far more persuasive than a case opened months later.
Flat colours, grey tones, ramps and patterns. The browser interface disappears completely, so you can inspect the screen down to its last pixel.
Advance with the arrow keys, the space bar, a click or a tap. Esc always leaves the test.
Pick an interval between one and fifteen seconds and walk through every screen hands-free.
A draggable, resizable flashing box with a stopwatch counting the elapsed time. It asks for a clear warning to be read and confirmed before it starts.
True resolution, device pixel ratio, colour depth, colour space and a measured refresh rate, all in one panel.
Both the test and the fixer run entirely in your browser. Nothing gathered about your screen reaches us.
Turn your screen brightness all the way up, darken the room and wipe the surface with a dry microfibre cloth. Then start the test in fullscreen and walk through the screens one by one. Look at each pattern from about half a metre away and scan the whole surface. On the black screen you are looking for a bright coloured dot; on the white screen for a dark one. The red, green and blue screens expose faults in a single sub-pixel.
In a dead pixel all three sub-pixels are permanently off; it shows as a black dot on a white field and software cannot fix it, because the transistor driving it has failed completely. In a stuck pixel one or two sub-pixels stay permanently lit; on a black field you see a red, green, blue, yellow, cyan or magenta dot. The transistor is still working there, so a stuck pixel has a real chance of being revived.
Sometimes. The fixer forces the locked sub-pixel to switch on and off dozens of times a second, trying to get the liquid crystal moving again. Some stuck pixels come back within ten minutes, some take over an hour, and some never recover. On a dead pixel it has no effect at all, because there is no working circuit left to drive. It is not guaranteed, but it costs nothing to try.
At least ten minutes. If nothing changes, park the box directly over the faulty pixel and leave it for forty-five minutes to an hour. The on-screen stopwatch counts the elapsed time so you do not have to. If there is still no change after an hour, take a break of a few hours and try once more. Running it far longer than that has no demonstrated extra benefit.
The fixer carries a risk for people with photosensitive epilepsy, which is why it asks you to read the warning and confirm before it starts. Do not use it if you have an epilepsy diagnosis, or if flashing images give you dizziness or headaches. Because each dot is drawn independently, the average brightness of the box stays constant and the screen does not pulse in sync; even so, do not stare at it from close range for long. There is no known harm to the panel, though running it for hours on an OLED screen is not advisable.
Yes. The test advances by tapping, and in browsers that support the Fullscreen API it hides the interface completely. Safari on iPhone does not support that API, so there the test fills the page but the address bar may stay visible; that does not stop you inspecting the rest of the screen. Rotating the phone and checking in both orientations makes it easier to catch faults near the edges.
That depends on the panel's class under ISO 9241-307. Most consumer monitors and laptops are class III, which allows five permanently lit and fifteen permanently dark pixels per million. Professional panels are usually class II, with a far tighter limit. Some brands promise a zero-dead-pixel guarantee on specific products as their own commitment rather than as a standard. Whatever the class, testing the device the moment you buy it and photographing and reporting any fault straight away puts you in a much stronger position.
It is easy to tell. If the dot glints or disappears when you tilt the screen, or moves when you nudge it gently with a fingernail, what you are seeing is dust, a dried water spot or a splash mark. A genuine faulty pixel sits in exactly the same place from every angle and never moves. Wiping the surface with a dry microfibre cloth before you start removes most of this confusion up front.
The measurement counts how many frames the browser draws in a second, and that number can only go as high as the browser allows. If the tab is in the background, battery saver is on, the laptop is unplugged, or the operating system is capping the frame rate, the result comes out low. On displays with variable refresh rate the instantaneous figure also moves around constantly. For an accurate reading, bring the tab to the front, plug the device in and measure again.
No. The colour cycle, the patterns, the display information panel and the fixer all run entirely in your browser. The resolution, device pixel ratio, colour depth and measured refresh rate are shown on your screen only; none of it is sent to a server, stored, or shared with anyone. Close the page and nothing is left behind.