Artificial Intelligence in the Service of Renaissance Art? MIT Student Presents Revolutionary Technology

Preserving old works of art is a complex and expensive process that could be revolutionized today. Alex Kachkine, a mechanical engineering student at MIT, has developed an AI-based technique that can digitally restore damaged paintings in just a few hours.

Traditional conservation relies on X-ray analysis , pigment studies , and painstaking manual labor. This process can take months and… well, it requires huge financial outlays. This is why museums store thousands of works that will probably never leave the warehouse.

Alex Kachkine , a mechanical engineering student at MIT and a private enthusiast and collector of antique paintings , came up with his idea almost by accident. Due to his limited budget, he usually bought only damaged paintings for his collection, and then amateurishly restored them himself. Over time, however, he decided to combine his engineering skills with his love of art and developed an innovative method of digital reconstruction based on artificial intelligence .

Illustrative photo / photo: Unsplash, Igor Omilaev

It all starts with a high-resolution scan of the image. Then, an AI- based algorithm identifies cracks and gaps in the composition , which are digitally reconstructed. During this process, colors are matched to the surroundings , and the most complex patterns are copied from other parts of the image. But this is where the revolution begins.

Kachkine developed the so-called digital mask . What is it? The digitally recreated image is printed on a thin polymer film , using high-quality pigments . This film is applied to the image and then fixed with a varnish . The most important thing: It can be completely removed at any time with the help of conservation solvents without damaging the image.

Painting conservation using Alex Kachkine's method / photo: MIT, Alex Kachkine

In the above image, the damaged element is on the left. The middle panel shows a map of the different types of damage . The right side shows the restored image with a laminate mask applied.

In a paper published in Nature, Kachkine describes applying the technique to a 15th-century painting , The Master of the Adoration of the Magi, in the Prado Museum . The AI ​​identified 5,612 parts of the painting that needed repair. The mask the student and his technology created contained 57,314 colors and took just 3.5 hours to apply. Kachkine estimates that this is 66 times faster than traditional restoration methods.

- This approach gives conservators much greater flexibility, enabling the restoration of countless damaged paintings deemed unworthy of high conservation budgets - concludes Kachkine .

Currently, Kachkine is raising funds to further develop his own conservation method . Although the art community is cautious about the subject, we suspect that this type of digital tool will soon become extremely popular, especially for works by less prestigious artists.