MIT mechanical engineering student Alex Kachkine has developed a new AI technique that could dramatically speed up the restoration of aged or damaged paintings. The method uses a high-resolution scan of the artwork, an AI algorithm to identify cracks and missing patches, and a digitally printed polymer film—called a "digital mask"—that is overlaid onto the painting and sealed with varnish. The mask can be removed without trace using conservators' solvents. Kachkine tested the process on a 15th-century oil-on-panel painting by the Master of the Prado Adoration of the Magi, where the AI identified 5,612 damaged sections and the restoration took just 3.5 hours—66 times faster than conventional hand inpainting.
This innovation matters because it could make high-quality restoration accessible to far more artworks. According to Kachkine's paper published in Nature, 70 percent of paintings in museum collections are kept in storage partly due to the prohibitive cost of restoration. By drastically reducing time and expense, the technique could allow countless damaged or neglected works to be returned to public view. It also gives conservators greater foresight and flexibility, while remaining reversible—a critical safeguard in the conservation field where botched interventions have historically caused irreversible damage.