Influence of photography distance and image network geometry on ground-based SfM photogrammetry accuracy

The photography distance and image network geometry are critical influencing factors on ground-based structure-from-motion photogrammetry (G-SfM) accuracy. However, their selection is often subjective, resulting in highly variable photogrammetric precision ranging from centimeters to sub-millimeters. Quantifying the influence of image network geometry and distance provides a scientific basis for designing G-SfM surveys across diverse applications, ensuring more reliable results. To address these uncertainties, this study systematically compares the commonly used Circumvent image network geometry (CING) with the Nadir (NING) and Nadir+Inclined (NIING) geometries—both widely adopted in drone-based SfM. Fifteen experiments were conducted under 3 geometries and 5 photography distances (0.6 m, 0.8 m, 1.0 m, 1.5 m, and 2.0 m). Results show that NIING achieved the highest accuracy, with an absolute bias of 0.04 mm at 0.6 m, representing a 55% improvement over CING and a 102% improvement over NING. Unlike CING and NING, whose accuracies remain insensitive to distance, NIING benefits from reduced photography distance, highlighting its potential for high-precision applications.