Related Experiment Video
Updated: Sep 7, 2026

Photorealistic Learned Landscapes for Augmented Reality
Published on: June 27, 2025
Next-generation forensic 3D surface documentation with projection-assisted digital spray multicamera photogrammetry
Michael Thali1, Jasmin Herr2, Dominic Gascho3
1Virtopsy Research Lab, Institute of Forensic Medicine, University of Zurich, Zurich, Switzerland.
Abstract:
Accurate injury documentation is essential in forensic medicine, however conventional two-dimensional imaging methods are limited in their ability to capture spatial morphology and enable retrospective three-dimensional (3D) reassessment. Although multicamera photogrammetry has significantly advanced forensic surface documentation, challenges remain when imaging with low texture, homogeneous coloration, or reflective properties. This study presents an integrated workflow combining multicamera photogrammetry with projection-assisted acquisition using a digital spray mode to improve surface reconstruction under optically challenging conditions. Technical feasibility was evaluated using a handbag, a mannequin, and a voluntary participant with simulated injuries applied as adhesive tattoos, acquired using the Botspot full-body scanner NEO system. The workflow generated anatomically coherent full-body 360° models with high-fidelity texture representation, enabling clear visualization, identification and retrospective 3D measurement of simulated injuries. Compared with conventional passive acquisition, the digital spray mode improved reconstruction completeness and feature-matching stability. These findings demonstrate the feasibility of integrating enhanced acquisition techniques with advanced visualization methods in a unified forensic workflow and highlight the potential of non-contact projection-assisted technologies for documenting optically challenging surfaces, including visually homogeneous or reflective regions of human skin, without requiring physical surface treatment that could interfere with forensic evidence. In addition, image datasets acquired within this workflow were processed using 3D Gaussian Splatting (3DGS), a radiance field-based rendering method enabling efficient real-time rendering and high-fidelity visualization of reconstructed scenes. While current 3DGS representations are primarily optimized for visualization rather than metrically validated surface reconstruction, the technique demonstrates considerable potential for future forensic applications, particularly in interactive documentation, immersive case review, and advanced 3D visualization workflows.
