Schematic of our equine back-surface-measurement apparatus and evaluation of its accuracy. (a,b) The study used two arrays of cameras: four high-speed (green) and eight motion-capture cameras (blue) for measuring back surface, as well as gross kinematics and stride timing, respectively. We positioned motion-capture cameras both above the horse on the tower to identify the orientation and position of the saddle region and level with the horse to identify timings of forelimb contact with the ground; both sets of parameters were derived using motion-capture markers (orange circles in b). (c–e) We estimated reconstruction accuracy by reconstructing a life-size, rigid, horse model and compared the surface generated by our high-speed-camera array to that of a professional laser scan of the rigid model. We present the differences in the surface between the laser-scanned model and our measurement apparatus in right lateral (c) and dorsal (d) views and summarized (e) as a histogram and plot of the cumulative percentage of surface points contained across error thresholds (grey line, right y-axis). (c,d) Insets present schematic views of the rigid horse model and the approximate location of the reconstructed surface in blue.