The effect of camera configuration on multi-camera markerless motion capture for biomechanical analysis

Markerless motion capture technology, such as Theia3D, enables human movement data to be collected outside traditional laboratory environments. However, the extent to which changes in camera configuration influence markerless kinematic output remains underreported. This study evaluated the effect of camera configuration on Theia3D-calculated kinematic parameters during walking, running, and jumping. Ten healthy adults completed five trials of walking, running and counter movement jumps in two volume sizes (10 x 8 m and 6 x 4 m). Movements were captured with eighteen Sony-RX0II synchronised video cameras, which were systematically selected to generate three camera configurations (18-cameras, 12-cameras and 6-cameras), then processed in Theia3D (v2024.1.17). Trunk, pelvis, hip, knee, and ankle kinematics were compared across sagittal, frontal, and transverse planes using one-dimensional statistical parametric mapping. Changes in camera configuration produced mean differences of < 2° during walking, < 4° during running, and < 3° during countermovement jumps. Largest differences were shown for all movements between 18 and 12-camera and 12 and 6-camera configurations in the larger (10 x 8 m) volume. Smallest differences were shown between the 18 and 12 camera configurations in the smaller (6 x 4 m) volume during counter movement jumps. This work demonstrated minimal effects of varying camera configurations on Theia3D markerless kinematics during walking, running and jumping. Camera placement was likely an important factor where small differences were identified. Overall, these findings indicate that Theia3D-derived kinematics are minimally affected by changes in camera configuration during walking, running, and jumping, supporting its use in environments where camera number and space may be constrained.