Stylus Pressure

From Position to Force

A digitising tablet that reports only where a stylus sits is already useful — the RAND Tablet of 1964 did exactly that, encoding X and Y into a stream of coordinates that a program could act on. But a skilled draughtsperson's mark is not reducible to position. Pressure, speed, angle: all of them modulate the line. Ignore pressure and you lose the most immediate of those three signals — the one a hand communicates most naturally when it presses a nib into paper and the line swells.

The engineering problem is straightforward to name and harder to solve cleanly: transduce the force applied along the stylus axis into a value the computer can read, without adding so much mass or spring resistance that the instrument stops feeling like a pen. Too stiff, and the user is fighting the sensor. Too loose, and small vibrations produce noise that the software reads as fluctuation.

How the Sensor Works

Most production digitisers solved this with a variable-capacitance or variable-resistance stack inside the tip. In the capacitive approach, the tip compresses a small foam or elastomeric layer between two conductive plates; the closer the plates, the higher the capacitance, and the capacitance maps to a voltage that the tablet's controller converts to a numerical value. In the resistive approach, a conductive foam or carbon element changes resistance under compression; the circuitry reads that resistance and linearises it.

Wacom, whose first tablet appeared in 1983 and whose electromagnetic resonance system became the dominant architecture, uses a different method: the pen has no battery. A coil in the tablet broadcasts a signal; the stylus contains an LC circuit (inductor and capacitor) that resonates at a frequency shifted by a pressure-sensitive capacitor in the tip. The tablet reads the returned frequency, extracts the pressure from the shift, and decodes position from the resonance pattern of a grid of loops beneath the surface. Wacom's core patents on this electromagnetic approach date to the late 1980s and describe the coupling between the pen's LC circuit and the tablet's loop array in detail. Because no power lives in the pen, the instrument is light and the tip travel can be minimal — the sensor detects load well before the tip physically deflects much.

Too loose, and small vibrations produce noise that the software reads as fluctuation.

The numerical output is a pressure value over a range that has expanded with each generation of the technology. Early implementations offered 256 levels (8 bits). By the mid-1990s, 1,024 levels (10 bits) had become a practical standard; current devices reach 8,192 levels (13 bits). Whether a painter can actually distinguish 8,192 discrete pressures is a separate question from whether having them reduces quantisation artefacts — finer resolution means the mapping from pressure to, say, brush width can be smooth rather than stepped, and smoothness matters even when the individual steps are imperceptible.

Mapping Pressure to a Mark

Hardware captures a number; software decides what the number means. The most common mapping sends pressure to line width, to opacity, or to both simultaneously. A typical Bézier-based vector brush records pressure at each sampled point along a stroke and varies the rendered ribbon accordingly, so the spline that describes the stroke's centreline carries a width function as well as a shape. In raster painting, pressure usually scales a per-pixel opacity value, so a light touch produces a translucent dab and full pressure produces full coverage — behaviour that resembles the way a watercolour brush releases pigment in proportion to how firmly it meets the paper.

The mapping need not be linear. A curve that expands the response at the low end and compresses it at the high end rewards delicate work and still allows a decisive full-pressure stroke; the opposite curve suits inking, where the early range is mostly flat and the line snaps to full width only under strong pressure. The Commission internationale de l'éclairage built analogous nonlinearity into perceptual colour encoding for similar reasons: the eye's sensitivity is not flat, and neither is the hand's.

The result of reading pressure correctly is a mark that has dynamics — a record of force over time, not just path through space. That is what makes digital drawing feel like drawing.