Direction diagnostic
Four guided phases separate normal down/up from faster down/up. Wrong-sign vertical events are counted only against the direction you declared by following the phase prompt.
Run a browser-local wheel diagnostic with guided direction phases, timing and speed evidence, horizontal-input inspection, raw WheelEvent logs, and repeatable local comparisons without confusing browser events with hardware telemetry.
Observe wheel events, run controlled direction/speed protocols, inspect timing and delta units, compare repeat runs, and export raw evidence without pretending browser events are hardware telemetry.
Vertical points are blue/down or red/up; horizontal-dominant points are purple. The chart scales the visible sample for inspection only. It does not convert line/page units into pixels.
| # | t ms | ΔX | ΔY | Unit | Dominant | Ctrl |
|---|
A comparison is marked compatible only when mode, duration, declared input and connection match. Setup labels are descriptive and are not device detection.
Four guided phases separate normal down/up from faster down/up. Wrong-sign vertical events are counted only against the direction you declared by following the phase prompt.
Median and p95 inter-event intervals, a peak one-second count and robust timing-outlier candidates help compare repeated runs on the same setup. They are browser-event timing, not hardware polling telemetry.
ΔX is kept separate from ΔY. Ctrl+wheel events are reported because trackpad/browser zoom gestures can generate wheel events even when content does not scroll.
A free event stream is useful for seeing whether the browser receives wheel input, but it cannot tell intentional direction changes from an unexpected reversal. Guided Diagnostic therefore separates normal down, normal up, faster down and faster up into declared phases and counts only wrong-sign vertical events inside those phases.
WheelEvent.deltaMode can report pixels, lines or pages. This studio does not combine those units into a fake physical travel distance. The report keeps separate absolute ΔX/ΔY totals per unit and marks mixed-unit runs explicitly.
Events per second, median and p95 inter-event intervals, peak one-second count and timing-outlier candidates describe the browser event stream. Firmware, drivers, operating-system settings, browser scheduling, acceleration and device type can all change those values. A gap does not prove a skipped wheel notch, and browser event rate is not hardware polling rate.
A wheel gesture can fire even when an element does not scroll, and trackpad zoom gestures can arrive as Ctrl+wheel events. Conversely, content can scroll from keyboard input, scrollbar dragging or script. The capture surface intentionally observes WheelEvent data rather than claiming to measure physical content travel.
1. Run Guided Diagnostic twice with the same motion. 2. If a reversal or timing pattern repeats, compare another browser or application. 3. Retest after one change such as a USB port, wireless receiver position, battery, OS setting or browser extension. 4. Save compatible runs locally and compare the metrics rather than relying on one sample. 5. Use vendor or lower-level tools when a hardware diagnosis matters.
Raw wheel events stay in this tab unless you explicitly export or save a run. Local history is capped at twelve runs and saved run traces at the latest 500 events. A settings link includes only mode, duration and declared setup categories; it excludes event traces and the free-form setup label.