DeviceBenchFree online monitor ghosting test
Ghosting test that shows the trail instead of scoring it
Squares slide across three bands at a speed you set — dark on white, white on black, and a low-contrast gray step — and the thing to look at is the edge each one leaves behind. A soft gray tail means the pixels are still finishing their last transition when the next one arrives; a tail brighter than the background is overshoot, which is a monitor’s overdrive pushed one setting too far and is cured by turning it down rather than up. A still copy of each square sits mid-band as an unsmeared edge to hold the moving one against, and there is deliberately no millisecond figure anywhere on this page, because response time is a measurement of light and a browser cannot see the light coming off your screen.
- 100% free
- No signup
- 3 transitions
- 240 – 1,920 px/s
- Full-screen field
- Band 1
Dark square, white background
A gray shadow lagging behind the square is slow pixel response on the rise.
- Band 2
White square, black background
A milky trail here is the transition most VA panels are slowest at.
- Band 3
Gray square, gray background
A trail brighter than the background is overshoot — overdrive pushed too far.
Follow the square with your eyes rather than staring at one point on the glass — smearing is a property of a tracked edge, and a fixed stare hides it. The pale square parked in the middle of each band is the same color with no motion in it, so if the moving one looks softer or lighter than the still one, that difference is the panel. Nothing here produces a response time in milliseconds, because a browser cannot see the light coming off your screen; a review-grade figure needs a photodiode pointed at the panel and an oscilloscope behind it.
How to test a monitor for ghosting
Look, compare, then change one setting on the monitor and look again.
Set 960 px/s and let your gaze travel with the square
Pursuit is what makes the defect visible: your eye moves smoothly, the panel updates in steps, and the difference between those two things is what appears as a tail. Staring at a fixed point on the glass instead will hide almost any amount of smearing, which is why a monitor can look flawless on a desktop and terrible in a game.
Hold the moving square against the still one
Each band has a stationary copy of its square at the center, drawn in the same color with no motion in it. If the moving square looks softer, lighter or wider than the parked one, the extra is the panel finishing its last transition late. This comparison catches small amounts of smearing that are invisible when there is nothing to hold them against.
Change the overdrive setting on the monitor and repeat
Overdrive lives in the monitor's own menu under a name like Response Time, Trace Free, OD or AMA. Step it up one level and look at the third band: if a trail brighter than the background appears behind the gray square, that level is too high for this panel and the setting below it is the right one. The middle setting wins on most monitors, and the highest setting is almost never correct.
Technical specifications
| Speeds | 240, 480, 960 and 1,920 CSS pixels per second, changed while the test runs |
|---|---|
| Bands | 3 at once — #111111 on white, #ffffff on #0b0b0b, and #9a9a9a on #6b6b6b for the low-contrast step where overshoot shows |
| Square sizes | 40, 72 and 120 px, reduced to fit the band when the field is short |
| Still reference | A copy of each square at 45% opacity parked at the center of its band, so there is an unsmeared edge in frame at all times |
| Striped background | Optional 24 px vertical stripes, one shade apart from the base color — hard edges under the square make a faint trail easier to catch |
| Rendering | Canvas 2D, rectangles blitted at integer positions at up to 2× device pixel ratio; no CSS transform, transition or filter touches the moving object |
| Full screen | Entered from a click, with an exit control drawn on the field itself rather than relying on the browser's own hint, which fades after a few seconds |
| What it cannot produce | A response time in milliseconds — that needs a photodiode against the glass and an oscilloscope behind it, and no browser API exposes emitted light |
Frequently asked questions
How is ghosting different from burn-in?
Ghosting follows the object and burn-in stays where it is. A ghost is a trail that exists only while something is moving and vanishes the instant it stops, because it is pixels taking too long to reach their new color. Burn-in is a shape that has stopped moving on with the picture: whatever sat in one place for hundreds of hours stays faintly visible on a plain background long after that program closed, and it is permanent damage to the emitting layer rather than a timing problem. If what you see does not move with the square, the burn-in test is the page you want.
There is a bright line behind the square instead of a dark tail.
That is inverse ghosting, and it means the monitor's overdrive is set too aggressively. Overdrive briefly overvolts a pixel to force it through its transition faster; when the push is stronger than the transition needed, the pixel sails past its target color before settling back, and the overshoot appears as a bright halo trailing the object. The fix is counter-intuitive to most people the first time: turn the response-time setting down, not up.
Which overdrive level should I use?
The highest one that produces no bright halo on the gray band. Manufacturers ship three or four levels and tune them for the fastest possible transition in a lab rather than for the mix of transitions real content contains, so the top level nearly always overshoots somewhere. Work upward one step at a time and stop at the level before the halo appears — and note that on many panels the correct level changes when the refresh rate changes, which is why some monitors have a variable-overdrive mode that adjusts itself.
My laptop smears badly and my desktop monitor does not.
Panel technology explains most of that gap. Laptop screens are overwhelmingly IPS tuned for color and power rather than speed, and they usually have no overdrive control at all; a gaming TN or a fast IPS desktop panel changes state several times faster and has a menu setting to push it further. The worst case for smearing is a VA panel on dark transitions — the second band here, white on black — where a slow rise from near-black can take long enough to leave a visible smudge even on an expensive monitor.
Can this give me a response time in milliseconds?
No, and any browser page claiming otherwise is guessing. A real response time is measured by pointing a photodiode at the glass and recording how long the luminance takes to travel between two levels, usually reported as gray-to-gray between fixed points with the 10% and 90% ends of the curve excluded. JavaScript can control when a rectangle is submitted for drawing and has no way to observe what the pixels then do, so the honest output of a browser test is a picture your eye judges rather than a number.
The squares stutter and jump rather than smear.
Then the problem is upstream of the panel. A jump means frames are arriving unevenly or being skipped, so the square is genuinely teleporting rather than being blurred by slow pixels, and the pause message on this page appears when the browser has stopped delivering frames at all. Judge that on the frame rate test first — a panel cannot be blamed for a picture it was never sent.
Do OLED screens and phones ghost?
OLED pixels switch in well under a millisecond, so they effectively do not smear, and a trail you see on an OLED is your own eye tracking a sample-and-hold image rather than the display lagging. Phones behave the same way with an extra wrinkle: touching the glass usually raises the refresh rate, so a handset can look noticeably worse in the two seconds after your finger leaves than it does while you are moving it.
About pixel response, overdrive and the trail behind a moving object
An LCD pixel does not switch, it travels. Applying a new voltage twists the liquid crystal into a new orientation and that rotation takes real time — a few milliseconds on a fast panel, sometimes twenty or more on a slow one, and never the same duration for every pair of colors. That last point is where the marketing goes astray: “1 ms” on a box is one transition under one method, usually gray-to-gray with the first and last tenth of the curve trimmed off, and the transitions that actually smear in daily use are the dark ones a specification sheet quietly leaves out. When the panel has not settled by the time the next frame arrives, the pixel is still partway between two colors when it is asked to move again, and the eye reads that history along the path of the object as a trail.
Overdrive is the manufacturer’s answer, and it is the reason the third band on this page is so dull to look at. The controller briefly drives a pixel harder than the target color requires, so that it arrives on time and then relaxes; when the push is correctly sized, the transition finishes inside one frame and no trail appears. When it is too strong — which is what the top setting on most monitors is — the pixel overshoots past its target, and the result is a bright edge behind a dark object or a dark edge behind a light one. Small gray-to-gray steps expose this before anything else does, because they need the least push and therefore suffer the most from too much of it. This is why raising the response-time setting can make a monitor look worse, and why the correct level often changes with the refresh rate.
Three other faults get called ghosting by people describing them for the first time, and separating them saves a return. A mark that stays put on a plain background is retention or burn-in, which the burn-in test checks with a full-field wash; a single point of wrong color that never changes at all is a defective sub-pixel, which the dead pixel checker hunts one color at a time; and motion that lurches rather than smears is uneven frame delivery, which the FPS test graphs. There is also a blur that is nobody’s fault: every sample-and-hold display holds each frame steady for its whole duration while your eye keeps moving, which smears motion even on a panel with instant pixels, and the only real cures are a higher refresh rate — check yours on the refresh rate test — or a strobing backlight mode. When the picture is clean but arrives late rather than smeared, that is a different complaint entirely and belongs on the input lag test. For a sweep of everything else a panel can get wrong, the monitor test runs the full pattern set in one pass.
What this test knows about your screen
Every number on this page is worked out by JavaScript running in the tab you are reading it in. Nothing you type, paste or open is uploaded, logged or kept, which is also why the tools carry on working after you disconnect from the network.
This page keeps no measurements at all, because it takes none: the squares are drawn, you look at them, and nothing is recorded on either side. The only property of your display it ever reads is how wide the drawing area is in CSS pixels, which it needs in order to know where to turn the squares around.