High-Performance Computing as an Accelerator for 3D Photogrammetry in Heritage Conservation

Kolarek, Branimir; Gamulin, Ljubo; Vuković, Lucija; Knežević Kuzman, Barbara; Lučić, Nađa (2025) High-Performance Computing as an Accelerator for 3D Photogrammetry in Heritage Conservation. In: Mihaljević, Slaven; Veldić, Ivana, (eds.) Srce DEI 2025: Knjiga sažetaka. Hrvatska, Sveučilište u Zagrebu Sveučilišni računski centar, p. 58 .

Preview	Image (JPEG) - Submitted Version - poster Available under License Creative Commons Attribution No Derivatives. Download (1MB) | Preview
	Microsoft PowerPoint - Accepted Version - presentation Available under License Creative Commons Attribution No Derivatives. Download (4MB)

Abstract

By leveraging HPC clusters to run structure-from-motion algorithms on overlapping photographs, our method generates highly detailed and accurate 3D meshes of artwork surfaces in a fraction of the time required by consumer-grade laptops. This approach enhances computational efficiency. Using the same dataset from our study on Carmelo Reggio's Portrait of Marija Ghetaldi-Gondola (ca. 1812), we transitioned from traditional laptop-based processing to an HPC-driven workflow. This shift allowed us to remove colour information from the photogrammetry data and focus on constructing precise 3D models, which revealed subtle surface anomalies including a faint outline of an eye on the sitter's neck and indications of a secondary portrait beneath the visible layer. These findings underscore the potential of HPC to enhance minimally invasive conservation methods by providing faster and more robust analysis capabilities.

Item Type:	Conference or workshop item published in conference proceedings (UNSPECIFIED)
Uncontrolled Keywords:	HPC; Photogrammetry; Heritage Conservation; 3D Mesh; High-Performance Computing; Accelerated Processing; Minimally Invasive Restoration
Subjects:	TECHNICAL SCIENCES > Computing > Information Systems
Divisions:	Center for Informatics and Computing
Depositing User:	Branimir Kolarek
Date Deposited:	08 May 2025 11:54
URI:	http://fulir.irb.hr/id/eprint/9802

Actions (login required)

View Item