Undersampled fast swipes look lagged and polygonal

What it is

When the finger moves fast and coordinate packets are sparse, ink, rubber-bands, or the seed of an inertial fling become a polyline: straight segments between samples, sharp corners, a curve faceted into a polygon. The object also sits on the last real sample until the next packet, so it does not look glued to the fingertip. The failure is trajectory undersampling, not compositor frame jitter, and not the steady lag people describe as sticky.

Why it happens

A circular or arcing motion is continuous in time; sampling takes points on a fixed clock. Larger intervals mean larger spatial jumps, and connecting them with line segments throws away curvature. A fast swipe maps the same interval onto a bigger pixel leap, so faceting shows up first in flicks, signatures, and game swipes, and almost never in slow drags. Tracking feels poor because the displayed position can only rest on the previous real sample; in the tens of milliseconds between packets the finger has already traveled. Velocity estimated from two adjacent points also turns smooth acceleration into a train of impulses, so the launch speed of inertial scrolling jumps around.

Studying it

Have people draw calibrated-speed arcs, sign their name, and fling pages while an optical marker records the true fingertip path; compare that path to the polyline the app received with a Hausdorff distance. Independent variables are report rate (or a driver drop fraction), peak swipe speed, and whether drawing uses raw segments or a spline. Dependent measures include segment count, maximum sagitta error, variance of fling launch speed, and a rating of “trajectory looks like a folding ruler.” Slow target-dragging studies produce a false “report rate does not matter.” When aligning camera and event paths, subtract display delay or faceting error bleeds into lag error.

Where it stops holding

A high-refresh panel cannot rescue undersampling by itself: the display can strobe at 120 Hz, and if touch still reports at 60 Hz there is no new coordinate inside the frame, so the polyline remains. Conversely, dense reports drawn with coarse segments make a polygon that is a stroking choice, not a sensor limit. Denoising that discards adjacent points as jitter manufactures faceting as well. At extreme speed the capacitive blob itself elongates and the centroid wanders; denser reporting then records a denser crooked curve. Small child fingers and gloved contacts add localization noise on top of undersampling; separate the polyline from the jitter.

Applying it

• Treat high-speed strokes as mandatory for signature, doodle, and gesture trails; inspect recorded frames for polygonal corners.
• Connect ink at least at report-rate samples; if you must thin, use curvature-sensitive decimation rather than dropping every nth point.
• Estimate inertial launch speed from several samples in a short window, not from the last two sparse points.

Related

• Same group: C2.17.1 Touch report rate is the sensor's position samples per second · C2.17.2 Super-refresh report rates feed predictive rendering with earlier samples · C2.17.4 Software interpolation cannot fully replace a hardware report rate
• Nearby: C2.10 Touch latency and directness · C3.06 Fling and inertia
• Search terms: undersampling · polyline trajectory · swipe speed