Accurate specimen digitization through glass achieved and validated using 3D surface scanning

Three-dimensional digitization and morphometric analyses of specimens have become a crucial source of data for natural sciences. However, the quest for thorough sampling has been impaired by exhibition practices of locking specimens behind glass. In this contribution, we tested how well structured white-light scanning and phone-powered photogrammetry (helped with parallel lasers) capture the morphology of a variety of fossil specimens behind modern (i.e., flat) and old (i.e., undulating) glass. Postprocessing of scans was often longer, essentially to remove duplicated surfaces. Nevertheless, we were able to generate precise and accurate meshes through modern glass (validated by scanning objects without the glass): 91–100% of comparable points deviated < 1 mm from the reference model and 89–99% deviated < 0.5 mm. Older glass multiplies the deviations, especially for large, three dimensional specimens. However, these models still have 75 to 94 % of their points deviating less than 0.5 mm from the reference model, which thus remains manageable. We also conducted a linear morphometric analysis on a diagenetically flattened ichthyosaurian and a high-density geometric morphometric analysis of an hippuritid bivalve. The differences in surface digitization techniques (e.g. laser VS structured white-light scanning) and the human error in landmark placement generate more deviation than the presence or absence of glass. We conclude that, with the right scanning methods, specimens can be confidently digitized through glass and incorporated into morphometric (or other digital modelling) analyses.