Repair a painting in 3 hours? MIT did it with AI

Repair a painting in 3 hours? MIT did it with AI

Nicolas Sarazin | Jun 13, 2025 4 minutes read 0 comments
 

The Massachusetts Institute of Technology (MIT) has unveiled a revolutionary technology that can restore damaged paintings in just a few hours using artificial intelligence. This innovative method uses a custom-printed polymer mask, delivering precise, reversible, and non-invasive restoration, promising to transform the conservation of artistic heritage worldwide.

The process steps involve scanning the painting, then the final result © Alex Kachkine, MIT

Key points

  • The system scans damaged areas of a paint job in high resolution to detect them precisely.

  • An AI generates a palette of tens of thousands of colors perfectly matched to the original hues.

  • A mask printed on a transparent polymer film is applied without direct contact with the paint.

  • The restoration is quick (approximately 3.5 hours) and fully reversible, without altering the original work.


Fast, precise and fully reversible restoration

The restoration of works of art, particularly old paintings, is a field where meticulousness, patience, and human expertise have long been the only guarantees of reliable results. Cracks, flaking, and deterioration caused by time, light, or humidity require months, even years, of manual intervention. But a team of researchers from the Massachusetts Institute of Technology (MIT) has just shaken up this paradigm by unveiling a unique technology based on artificial intelligence, capable of restoring damaged paintings in just a few hours.

The process, which is fast, precise, and respectful of the original work, relies on a sophisticated system that begins by scanning the painting. Every detail is scanned in high resolution, allowing the algorithm to automatically detect damaged areas, even at the microscopic level. In a recent demonstration, a late 15th-century painting attributed to the Master of the Adoration in the Prado, a Dutch painter whose name has been lost, served as a case study. The algorithm identified no fewer than 5,612 micro-damages spread across the entire painting surface. Based on this analysis, the artificial intelligence generated a personalized palette of more than 57,000 colors to faithfully reproduce the original hues.

But the real innovation lies in the creation of a restoration "mask" printed on a transparent polymer film. This film, composed of two layers of ink—one white, the other colored—is designed to perfectly overlap the damaged surface. It is applied without direct contact with the painting: it adheres only to the protective varnish, making the procedure completely reversible. In other words, the restored painting can be returned to its original state at any time, without permanent alteration. This method also avoids any addition of material directly onto the work, a practice often a source of debate in the conservation world.

A major step forward for museums and forgotten collections

Alex Kachkine, a master's student in mechanical engineering at MIT, is behind this innovation, which could permanently transform the world of art restoration. Passionate about the intersections between technology and heritage, he designed this precision tool combining artificial intelligence, polymer printing, and absolute respect for the original works. His goal was clear: to create a fast, reversible, and non-invasive method for restoring paintings weakened by time. By combining scientific rigor and artistic sensitivity, Alex Kachkine demonstrates how engineering can today be committed to serving cultural memory.

The entire process, from digital analysis to mask printing, takes only 3.5 hours, approximately sixty times less time than an equivalent traditional restoration. This spectacular time saving opens up unprecedented opportunities for museums, cultural institutions, and private collectors, particularly for works that, due to lack of budget or notoriety, remain stored in reserves with no hope of restoration.

The technology developed by MIT also offers several significant advantages. It is non-invasive, inexpensive compared to traditional methods, and allows for complete digital documentation of each intervention. This transparency facilitates archiving, traceability, and scientific research. Moreover, because the mask is completely removable, it can be replaced or modified without risking damage to the original, thus meeting the strict ethical requirements of heritage conservation.

However, this innovation is not without its limitations. It only works effectively on paintings with flat, smooth surfaces. Relief works, impasto, or thick textures currently escape this type of treatment. Moreover, the question of authenticity remains central. Even if the intervention is reversible, the addition of an external mask raises debates about the boundary between restoration, recreation, and falsification. This is why researchers insist that this technology must remain a tool at the service of conservators, and not a substitute for their judgment.

FAQ

Is this method suitable for all paints?
It works primarily on flat, smooth surfaces. Relief or impasto artwork cannot yet be processed with this technology.

Is the restoration permanent?
No, the polymer mask is reversible and can be removed without damaging the original paint.

How much time does this method save?
It allows a work to be restored in around 3.5 hours, compared to several weeks for traditional manual restoration.

Does this technology replace the work of restorers?
No, it should be used as a complementary tool under the supervision of qualified professionals.

Is this process expensive?
Compared to traditional methods, this technology is faster and potentially less expensive, making restoration accessible to a greater number of works.

View More Articles
 

ArtMajeur

Receive our newsletter for art lovers and collectors