How We Test

Last updated

MouseWisp reads input events that the browser already exposes to a webpage. The mouse, receiver, driver, operating system, and browser can all influence those events. The tools run in your tab without an account or uploads. Reloading the page clears the session from the tool.

Buttons and wheel

The mouse test counts a press as soon as the browser supplies a standard button-down event over the test area. It measures the latest hold from down to up using browser timestamps. If the tab loses focus, it clears held highlights rather than inventing a release timestamp. Mouse events report individual button changes even when you are pressing two buttons at once. The diagram uses the browser’s Left, Right, Middle, Back and Forward assignments.

The wheel check counts signed vertical wheel events. A single event may cover more or less than one physical notch. Momentum, touch gestures, or scroll direction settings can change the stream. A button assigned to a keyboard command or intercepted by the OS may never reach the test.

Chatter and accidental double clicks

The double-click detector compares consecutive press timestamps separately for each button. It flags gaps at or below the selected threshold. The default of 80 ms is a practical starting point for deliberate single clicks. There is no universal gap that separates a hardware fault from intentional rapid input. The homepage uses the same default cutoff for its preliminary verdict.

The guided sample captures the first 500 eligible press events for the selected button. The first sets a baseline, leaving 499 gaps in a complete sample. The detection rate is flagged gaps divided by eligible gaps. A physical press that generates two events can contribute twice to the press total. The test can show which button emitted short gaps, while the comparison procedure helps isolate a switch, OS click grouping, or a software macro.

Polling rate

The polling rate test reads mouse pointermove timestamps, expanding getCoalescedEvents() when the browser supplies those samples. It excludes nonpositive gaps and pauses longer than 100 ms. Average Hz is 1,000 divided by the mean gap in milliseconds. Peak is the fastest rate in a completed block of at least 100 ms of accepted intervals. Stability is the share of intervals within 25 percent of the mean interval. The readings describe the most recent two seconds of samples and stay visible when motion stops.

The graph uses animation frames to schedule painting. The timing calculation uses event timestamps. Browser delivery can still track display refresh, and the browser can merge samples before a script sees them. Firefox privacy settings may round timestamps, and Safari and older browser versions can expose fewer underlying events. Results above 1,000 Hz are especially unreliable. Use native device timing for a hardware measurement. W3C Pointer Events specification

DPI and distance

The DPI test requests unadjusted relative movement through Pointer Lock. It sums signed horizontal deltas, takes their absolute value, and divides by the measured distance in inches. It converts centimeters using 2.54 cm per inch. The comparison shows the percentage difference from the claimed DPI.

An older lock implementation can yield browser-relative units. The cursor fallback uses the horizontal difference in CSS pixels between start and release and shows effective screen pixels per inch. Display scaling, pointer speed, browser zoom and acceleration can affect those units. Turn off acceleration, keep the mouse level, measure travel with a ruler, and repeat the same distance at multiple speeds. W3C Pointer Lock specification

Saving results

The button and chatter tests can draw a PNG locally with the result, canonical tool URL, and the UTC timestamp of the last input (or export time for an empty test). You can download the image or copy a text result. Both actions are optional and use the device’s own download or clipboard facilities.