I’ve been training a model to reconstruct fresco styles from the Renaissance, mapping pigment spectra from the surviving fragments. Think we could reverse‑engineer the original pigment mixes from a few faded spots and generate a full‑scale restoration?