Numerical settings

The hand demo uses 20 g sensor mass, 10,000 N/m contact stiffness, and 50 N s/m damping at a 1/240 s timestep. These are simulation parameters.

Values	Meaning
mount_translation_m, mount_rotation_wxyz	Fixed transform from the original taxel coordinates into the mounted fingertip frame; metres and (w,x,y,z) quaternion.
registration_shift_in_sensor_m	Translation fitted to the overmold interior, with no scaling or rotation of the distribution.
0.1 kg and 9.81 m/s²	Known-load fixture; their product is the expected 0.981 N.
20 g, 10,000 N/m, 50 N s/m	Compliant rigid-body simulation parameters, not measured hardware properties.
1/240 s timestep	Advances 4.17 ms of simulation time per step: 240 steps per simulated second.
192 position and 4 velocity iterations	Repeated PhysX solver passes within each step to refine contact and joint constraints; these are not taxel counts. Conservative validated demo settings, not an optimal configuration claim.
0.1 mm contact offset; SDF resolutions 384/128	Numerical contact margin and field resolutions for the sensor/test objects; unrelated to taxel count.
1 ms SDF probe time	Finite-difference half-width used to estimate depth rate, independent of the physics timestep.
5% load and 1 mm thumb-centroid tolerances	Verification acceptance bounds, not physical sensor specifications.

The DEX-01 frame still contains 738 taxel values in a 32 × 32 layout. A sensor updated every step can provide 240 frames per simulated second; wall-clock throughput depends on the host. The stationary-press demo uses a 1/120 s timestep.

Camera positions, colors, and plot cadence affect presentation. Gear scale changes physical collision geometry and contact footprints and requires validation. Solver settings and actuator gains control the simulation; changes should pass the validation workflow. The contact distribution and supported counter-object behavior are described in the model guide.