Hyperspectral scans of historical pigments and painting reconstructions

Summary

Painting Tools and Dataset Processed paint samples, using the Kubelka-Munk model to compute reflection and the painting_tools package to display the RGB values. See the corresponding notebook to understand how these parameters were estimated. This repository contains code and links to data for painting analysis using hyperspectral data. This data can be used in technical art history and computer graphics applications, for example for pigment mapping and spectral upsampling. Code and data for this type of research is often hard to find and we hope that sharing this repository contributes to open code and data in technical art history. The accompaying data contains: Hyperspectral scans of nine historical reconstructions by professional painters. Hyperspectral scans of ten historical pigments in oilpaint. Several painting stages of one of the captured paintings (Vermeer's Milkmaid). For each scan, we provide: Raw scan files from the Specim IQ hyperspectral scanner. Code to process the raw files (i.e., stitching scans that were made in segments, reading out spectra from samples). Processed data. An example python Notebook for unmixing paints with a reimplementation of Pigmento [1] using a painting and paint database from our dataset. The data was captured as part of research conducted at the CGV group at the TU Delft by Ruben Wiersma for a project in collaboration with Elmar Eisemann and Adrien Bousseau. The reconstructions were painted by Lisa Wiersma, Charlotte Caspers, and Mané van Veldhuizen. Matthias Alfeld generously assisted in capturing the hyperspectral scans. The data is shared under a permissive copyright license (CC-BY-NC-SA) and the code under the MIT license. If this data or code is helpful to your research, please cite this repository and attribute the artists (see below) and send me an email. I would love to hear your thoughts and feedback! Left: The calibrated and stitched hyperspectral scan of Rachel Ruysch's bloemstilleven as reconstructed by Lisa Wie...