Adapting radiance fields for high-fidelity digital reproduction of fine art paintings

Thumbnail Image

Files

master_Le_Hoanh_2025.pdf (139.1 MB)

URL

Journal Title

Journal ISSN

Volume Title

School of Science | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2025-09-16

Department

Major/Subject

Machine Learning, Data Science and Artificial Intelligence

Mcode

Degree programme

Master's Programme in Computer, Communication and Information Sciences
Master's Programme in Computer, Communication and Information Sciences
Master's Programme in Computer, Communication and Information Sciences

Language

en

Pages

69

Series

Abstract

This thesis investigates how modern radiance field methods can be adapted for the high-fidelity digital reproduction of paintings. Unlike typical 3D scenes, paintings are predominantly planar, yet exhibit subtle depth variations, complex surface textures, and view-dependent effects, such as gloss and specularity. These properties are often lost or diminished in traditional photographic capture, limiting authentic digital engagement. We explore the use of Radiance Fields and propose a set of practical adaptations that improve their performance on near-planar, high-detail surfaces like paintings. Our contributions include bounding box optimization, encoding refinements, and anti-aliasing strategies specifically tailored to the visual characteristics of artworks. Through extensive evaluations, comparing against state-of-the-art methods such as Instant-NGP and Gaussian Splatting, we demonstrate that these adaptations enable significantly more accurate and perceptually faithful novel view synthesis of paintings. This work lays the foundation for more immersive and authentic digital art preservation and presentation.

Description

Supervisor

Lehtinen, Jaakko

Thesis advisor

Härkönen, Erik

Keywords

novel view synthesis, radiance fields, NeRF, machine learning, computer graphics, rendering

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202512179367

Collections

[dipl] Perustieteiden korkeakoulu / SCI