News
An automated scanning system to create photorealistic digital 3D objects
Abstract
This paper is devoted to the creation of photorealistic digital 3D objects – a task in demand in the retail, design, and entertainment industries. An analysis of current reconstruction methods revealed their key limitations: difficulties working with mirrored, transparent, monochromatic, and thin surfaces; the impossibility of simultaneously accurately reconstructing geometry and materials; and the high labor expenses for obtaining reliable models using reverse engineering methods. The goal of the study is to minimize manual labor and obtain reliable copies in a single scanning cycle. A method for separate reconstruction of geometry and materials using washable texture markers is proposed, ensuring material recovery and stable geometry reconstruction. The system architecture supports the integration of implicit neural reconstruction methods and active scanners. Experiments on a representative dataset have demonstrated high accuracy in the reconstruction of geometry and materials, correct relighting of models, and low operator time costs (less than 15 minutes). The resulting digital objects are suitable for use in AR/VR/web environments. Testing identified a common automated-system limitation: fidelity decreases as the virtual camera approaches, and solutions to mitigate this were proposed.
Keywords
Edition
Proceedings of the Institute for System Programming, vol. 38, issue 3, part 3, 2026, pp. 79-100
ISSN 2220-6426 (Online), ISSN 2079-8156 (Print).
DOI: 10.15514/ISPRAS-2026-38(3)-37
For citation
Full text of the paper in pdf (in Russian)
Back to the contents of the volume