How to test a screen properly
A few minutes of testing now can save you from keeping a faulty panel. Most screens can only be returned or exchanged inside a short window, so check yours while you still can. For results you can trust:
- Go fullscreen. Each test fills the whole screen so no menu, cursor or scrollbar sits on the panel. Press F to toggle fullscreen and Esc to exit.
- Test in a dark room for the black and uniformity checks. Backlight bleed and IPS glow are only visible with the lights off.
- Use native resolution and a normal picture mode. Avoid “vivid” or “dynamic” presets that crush detail, and set brightness to what you’ll actually use.
- Clean the screen first so dust or smudges aren’t mistaken for dead pixels.
- Look at the whole panel, including the corners, from a normal viewing distance.
- Mind the return window. If you find dead pixels, heavy bleed or obvious banding, exchange the unit while you still can.
What each test checks
Dead / Bright Pixel
Fills the screen with solid red, green, blue, white, black and more. A dead pixel stays black on every color; a stuck pixel shows as a fixed colored dot, easiest to spot on black or white.
Uniformity
Pure black plus gray levels to reveal backlight bleed, IPS glow, clouding and the “dirty-screen effect.” Best judged in a dark room.
Contrast
Near-black and near-white step patches. If you can’t tell the lowest or highest steps apart, the panel or your settings are crushing shadow or clipping highlight detail.
Gradient
Smooth ramps that expose banding (visible steps) and posterization. Some banding can come from an 8-bit panel or the browser, not only the display.
Color
SMPTE-style color bars, saturated fields, and wide-gamut P3 / Rec.2020 swatches where the browser supports them. Swatches are indicative, not a calibrated measurement.
Sharpness
1px line grids and a 1px checkerboard rendered at true device pixels, plus sample text, to judge focus, scaling blur and subpixel rendering.
Geometry
An alignment grid for overscan and projector setup, a viewing-angle field, and a gamma reference pattern.
Refresh Rate
Measures the browser’s animation frame rate live and reports an approximate value in Hz with a frame-time graph.
Motion
A moving object to judge motion blur, trailing and overshoot using the pursuit (eye-tracking) method. Perceptual only, with no millisecond figure.
Stuck Pixel
A movable square that rapidly cycles colors and may coax a stuck subpixel back to life. It flashes quickly, so it sits behind a photosensitive-seizure warning. It may help, but isn’t guaranteed.
HDR
Detects whether HDR is reported by your browser and display, and shows a wide-range highlight pattern. A support check, not a brightness measurement.
Frequently asked questions
How do I test a new TV or monitor for dead pixels?
Run the Dead / Bright Pixel test fullscreen and step through the solid colors. A dead pixel stays black on every color; a stuck pixel shows as a fixed red, green or blue dot, easiest to see on a black or white field. Look across the whole panel, corners included, and clean the screen first so dust isn’t mistaken for a defect. A few stuck subpixels are common, and many warranties only replace a panel once it passes a set number of dead pixels, so check the policy, and if you’re inside your return window, exchange the unit rather than wait.
What is backlight bleed, and how much is normal?
Backlight bleed is light leaking from the edges or corners of an LCD/LED panel on what should be a pure-black screen; on IPS panels you’ll also see “IPS glow,” a cooler haze that shifts as you move your head. Check it with the Uniformity test’s black field in a dark room at normal brightness. A little even bleed is normal on edge-lit LCDs; bright, uneven patches or obvious corner “flashlighting” are not. OLED panels have no backlight and show none of this. There’s no universal pass/fail number, so judge it by how distracting it is in a dark room, and return a unit that bothers you while you can.
How do I check my screen’s refresh rate?
Open the Refresh Rate test and let it run for a few seconds; it averages the browser’s animation frames and reports an approximate value such as ≈ 60, ≈ 120 or ≈ 144 Hz. This is the rate the browser is actually animating at, which normally matches the panel but can be capped by power-saving, the GPU or the browser, so treat it as a measured estimate, not the panel’s rated maximum. If it reads low, make sure the higher refresh rate is selected in your OS display settings and that a laptop is plugged in.
Can I test HDR in a browser?
Only partly, and we won’t pretend otherwise. The HDR section detects whether your browser and display report HDR support and shows a wide-gamut highlight pattern, but a web page cannot reliably drive true HDR peak brightness or measure it in nits. No browser tool can. Use it to confirm HDR is being recognized and to eyeball highlights; for real HDR brightness and tone-mapping you need dedicated patterns and a measurement device.
Does this work on a phone?
Yes, but a desktop or laptop is better for serious testing. Tap to advance, swipe to navigate, and use the on-screen ✕ to exit. Mobile browsers don’t all support true fullscreen (iPhone Safari in particular), some patterns rely on exact pixels that phones scale, and you can’t easily inspect a TV from a phone. Use a phone for a quick look, a computer for a careful check.
Why isn’t there a response-time or input-lag number?
Because a browser can’t measure either honestly. The Motion test is a perceptual check: follow the moving object with your eyes to judge blur, trailing and overshoot, but it can’t print a millisecond response time, because frame timing is governed by the browser and vsync. Input lag, the delay from your input to the screen, needs external hardware to measure at all, so we don’t fake a number for it.
Are my results private?
Completely. Everything runs in your browser, on your device. Nothing is uploaded, there are no accounts, no cookies, and no analytics or trackers. After the first load the tool works offline. Your screen, and whatever you see on it, never leaves your machine.