ARCHAEOVISION: A Multi-Camera Microphotography System for Field Digitization of Small Vertebrate Remains

Carmina Baylon1*, Chara Punzal1, Patricia Cabrera2, Juan Rofes2,3, and Giovanni Tapang1,4

This paper presents a portable, multi-camera, integrated microscope that combines images from an array of smaller cameras to photograph and create 3D models of small archaeological bone specimens from multiple angles, with minimal handling of the material. The prototype was developed as part of the ARCHAEOVISION project at the University of the Philippines Diliman. Synthetic models with known geometry were used to calibrate the camera array, validate image-stitching, and establish spatial correspondence before testing on actual bone specimens. The system successfully captured multi-angle images of small-vertebrate bone fragments for machine-learning classification and 3D mesh generation using NeRFs and Gaussian splatting. Key challenges encountered included ensuring consistent illumination across all cameras in the array, precise spatial calibration for accurate image stitching, and minimizing vibration artifacts at high magnification. Handling fragile and irregular bone fragments without introducing additional damage or displacement between shots also proved difficult. Field deployment introduced additional constraints related to portability, dust and water resistance, and ambient lighting, which required iterative design adjustments between the alpha and beta versions. The developed setup addresses the critical sorting bottleneck in zooarchaeological workflows where manual identification of large assemblages is slow and labour-intensive. The system allows fast, consistent multi-angle digitization, which, combined with machine learning and image processing, will greatly cut the time needed to sort, classify, and catalogue bone fragments on a large scale. It builds the groundwork for an intelligent online reference catalogue for archaeological vertebrate collections.