Split thresholds depend on sensor precision and contact area

What it is

How far two fingers must move apart before they become two ids is not a universal millimetre figure. The split threshold tracks electrode pitch, filter strength, and blob area: a fine grid and a dry fingertip can separate at a few millimetres; coarse electrodes and a wet thumb pad may need more than a centimetre. The same person on the same device produces two thresholds with a fingertip versus a pad.

Why it happens

Segmentation looks at peak spacing relative to peak width. A large contact widens the peak, so the same geometric finger gap looks closer in the signal, and the threshold climbs. Edge electrodes are incomplete, equivalent pitch worsens, and the threshold climbs again. Firmware that resists noise demands a valley deeper than a signal-to-noise floor before it will open a new id; glove mode raises gain and noise together, sometimes lifting the threshold further to avoid ghosts. Temperature changes skin conductivity and surface moisture, so winter outdoors and a bathroom do not share a number. The threshold is a function of conditions, not a constant to copy from a datasheet.

Studying it

On one person, measure merge–split hysteresis with a fingertip, a pad, and a thin glove, at center and at the four edges: close slowly until ids collapse, open slowly until they separate—the two critical distances often differ. Independent variables are contact area (the reported major axis is a usable proxy), region, and gain mode. Dependent measures are the merge threshold, the split threshold, and their difference (hysteresis). Repeat on two device tiers to pull “sensor precision” apart from “this finger.” Report skin state; mixing dry and wet smears the distribution into a useless wide peak.

Where it stops holding

An active stylus tip is far smaller than a finger; the stylus–finger split threshold does not predict two fingers. A screen film changes effective area and noise, so factory thresholds fail after a protector is applied. Treating a lab probe’s metal disk as a finger yields an optimistic small threshold. Products whose two-finger gestures always span a tablet width almost never meet this variation; it hurts close pinches on phones and children’s hands.

Applying it

• Measure the split threshold on the target model, region, and finger (thumb pad versus index tip). Set the gesture’s minimum onset span from the worst case plus margin; do not copy another model’s millimetres.
• When glove or wet-hand mode is on, temporarily enlarge the two-finger minimum span, or move zoom onto buttons.
• Accept close pinches with both children and adults; if only adults split reliably, give children a one-finger zoom.

Related

• Same group: C2.19.1 Nearby fingers can be merged into a single reported contact · C2.19.2 Merged contacts disable pinch gestures at onset · C2.19.3 Contact splitting can produce a sudden coordinate jump
• Nearby: C2.01 Finger contact area and touch centroid · C2.24 Gloves, wet hands, and extreme conditions
• Search terms: split threshold · contact area · hysteresis