Methods and Workflows for Colour Change Modelling and Reconstruction

This chapter presents end-to-end methodologies and workflows for modelling, reconstructing, and communicating colour change in cultural heritage across four scenarios: classical sculpture, paintings, historical textiles, and early colour photographs/films. We couple experimental mock-ups, created with chemically appropriate pigments/dyes, with computational imaging and appearance modelling to realistically capture the evolution of the aging process. For sculpture, we propose an atlas-based segmentation and semantic shading pipeline that transfers measured appearance from mock-ups to 3D models, enabling virtual reconstructions of lost polychromy. For paintings, we introduce a dual pathway to predict and visualize light-induced discolouration: a photochemical model linked to radiant exposure and a data-driven spectral unmixing approach; both are demonstrated on Edvard Munch’s The Scream (ca. 1910), and further supported by corrections of archival colour film records. For textiles, we implement appearance-aware 3D restoration and fading simulation, integrating spectral unmixing with multi-temporal photographic documentation to visualize past states and forecast future fading. Finally, for early colour photographs, we present high-magnification multispectral imaging for the recovery of colour in faded autochromes, neural methods for repairing the bleeding of the green dye in some autochromes, and lastly, a refined digital decoding strategy for Kodacolor lenticular film. Together, these methods bridge analytical evidence, appearance capture, and perceptual visualization, providing reproducible, decision-support tools for conservation, display, and public engagement.