Studies investigating the concurrent validity of markerless motion capture (MMC) systems for obtaining joint and segment kinematics have largely failed to consider the impact of ambient lighting and camera position, which can both affect video quality, thus likely impacting landmark tracking. This study aimed to systematically quantify the impact of room lighting and camera positioning on three-dimensional kinematics obtained from MMC. Fourteen participants walked (3 km/h) and ran (12 km/h) on a treadmill whilst 14 infrared cameras obtained marker-based kinematics and 12 video cameras, which were processed with Theia3D (Theia Inc., Canada), provided MMC-based kinematics. Four lighting conditions (1069, 692, 455, 289 lux) were tested, whilst different camera setups (surrounding, corridor, and anterior-biased) were recreated post-collection. Room lighting affected joint range of motion obtained from MMC (p < 0.05), particularly in frontal and transverse planes, possibly because of compromised tracking at lower light intensities as the marker-based kinematics did not change between trials. Camera position affected range of motion in all three planes across hip, knee, and ankle joints (p < 0.05), possibly because of less visibility of tracking landmarks used by the MMC tracking algorithm. These results showcase the importance of transparency in reporting camera setups and lighting conditions when using MMC, particularly in concurrent validity or system comparison studies where the difference between systems is clearly affected by data collection setup. Agreement between marker-based and MMC might therefore be lighting- or camera arrangement-specific, so reporting of lighting and camera positions should therefore be considered essential in future work.
Thomas et al. (Wed,) studied this question.