Photogrammetry, the use of photography to calculate measurements, includes a suite of algorithms known as Structure from Motion (SfM) that model 3D structure from standard photographic images. While use of SfM for objects or scenes at scales from centimeters to kilometers is common, studies that use macrophotography for small objects or scenes are few. This study tests methodology to optimize 3D modeling for change detection of lichen structure and community dynamics over time. Specifically, tests are performed to clarify uncertainties on software, photographic equipment, spatial control, making measurements, and the potential for accuracy and precision. Test subjects included common lichen communities on wood, bark and rock, plus a calicioid lichen community, each with two or three models over time. The commercial software PhotoScan outperformed the no-cost software alternatives VisualSfM and COLMAP. With PhotoScan, results of 3D modeling showed that while cellphone cameras are capable of producing models with sub-millimeter accuracy, DSLR cameras produce more accurate and detailed models. Photography using a narrow-angle macrolens demonstrated advantages over a wide-angle lens when working at this scale. Inclusion of a scaling object increased accuracy of spatial control relative to hand measurements. Detail and accuracy of models was sufficient to detect radial growth of a substrate tree and changes in calicioid ascomata that likely corresponded to humidity and provided evidence for hygroscopy in Chaenotheca. Models achieved accuracy as low as 0.06 mm root-mean-square-error (RMSE) and data density as high as 1.7 million points per cm² of substrate; accuracy was non-random and may be better described as local positional bias.