The restoration of art takes regular hands and a demanding eye. For centuries, the Conservatives have restored paintings by identifying the areas requiring repair, then mixing an exact shade to fill an area at a time. Often, a paint can have thousands of small regions requiring individual attention. The restoration of a single painting can take from a few weeks to more than a decade.
In recent years, digital catering tools have opened a path to create virtual representations of original and restored works. These tools apply computer vision techniques, image recognition and color correspondence, to generate a “digital restored” version of a relatively quickly paint.
However, there has been no way to translate digital restorations directly on an original work, so far. In a paper Appearing today in the newspaper NatureAlex Kachkine, a student graduated in mechanical engineering at MIT, presents a new method he has developed to physically apply a digital catering directly on an original painting.
The restoration is printed on a very thin polymer film, in the form of a mask which can be aligned and adhered to an original painting. It can also be easily deleted. Kachkine says that a digital mask file can be stored and mentioned by future curators, to see exactly what changes have been made to restore the original painting.
“Because there is a digital recording of what Mask was used, in 100 years, the next time someone will work with this, he will have an extremely clear understanding of what has been done on the board,” explains Kachkine. “And this has never been really possible in conservation before.”
As a demonstration, he applied the method to a very damaged oil paint from the 15th century. The method has automatically identified 5,612 separate regions that need repair and filled these regions using 57,314 different colors. The whole process, from start to finish, took 3.5 hours, what it estimates is about 66 times faster than traditional catering methods.
Kachkine recognizes that, as for any restoration project, there are ethical problems to consider, to find out if a restored version is an appropriate representation of the original style and intention of an artist. Any application of his new method, he says, should be carried out in consultation with the Conservatives with knowledge of the history and the origins of a painting.
“There is a lot of damaged art in storage that might never be seen,” says Kachkine. “I hope that with this new method, there is a chance that we will see more art, which I would be delighted.”
Digital connections
The new restoration process began as a parallel project. In 2021, while Kachkine was heading for MIT to start her doctoral program in mechanical engineering, he led the East Coast and wanted to visit as many art galleries as he could along the way.
“I have been in art for a very long time now, since I was a child,” explains Kachkine, who restores paintings as a hobby, using traditional hand painting techniques. While visiting galleries, he realized that art on the walls is only a fraction of the works that the galleries hold. A large part of the art that the galleries acquired are stored because the works are aged or damaged and take the time to restore properly.
“Restoring a painting is fun, and it's great to sit down and fill things up and have a good evening,” explains Kachkine. “But it's a very slow process.”
As he learned, digital tools can significantly accelerate the restoration process. Researchers have developed artificial intelligence algorithms that quickly paint through huge amounts of data. Algorithms learn connections in these visual data, which they apply to generate a digital restored version of a particular paint, in a way that looks closely at the style of an artist or a period of time. However, such digital restorations are generally displayed practically or printed as autonomous works and cannot be directly applied to retouch original art.
“All this made me think: if we could simply restore a painting digitally and affect the results physically, it would solve a lot of points of pain and disadvantages of a conventional manual process,” explains Kachkine.
“Align and restore”
For the new study, Kachkine developed a method to physically apply a digital restoration on an original painting, using an 15th century painting which he acquired when he arrived at the MIT for the first time. His new method first involves using traditional techniques to clean a paint and eliminate all past catering efforts.
“This painting is almost 600 years old and has undergone conservation several times,” he says. “In this case, there were a lot of surpluses, which all had to be cleaned to see what is really there to start.”
He scanned the cleaned paint, including the many regions where painting had faded or cracked. He then used existing artificial intelligence algorithms to analyze the scan and create a virtual version of what painting in its original state likely looked like.
Then, Kachkine developed software that creates a map of the regions of the original paint that require filling, as well as the exact colors necessary to correspond to the digital restored version. This card is then translated into a two -layer physical mask which is printed on thin films based on polymers. The first layer is printed in color, while the second layer is printed in the same pattern, but in white.
“In order to completely reproduce the color, you need both white ink and in color to get the full spectrum,” explains Kachkine. “If these two layers are poorly aligned, it's very easy to see. So I also developed some calculation tools, depending on what we know about the perception of human colors, to determine how much we can align and restore the small region. ”
Kachkine used high -fidelity commercial ink jets to print the two layers of the mask, which he carefully aligned and superimposed by hand on the original paint and adhered to a thin spray of conventional varnish. Printed films are made from materials that can be easily dissolved with quality of conservation solutions, in case the preservatives must reveal the original and damaged work. The digital mask file can also be saved as a detailed recording of what has been restored.
For the painting that Kachkine used, the method was able to fill thousands of losses in just a few hours. “A few years ago, I restored this baroque Italian painting with probably the same amplitude of losses, and that took me nine months of part-time work,” he recalls. “The more losses, the better this method.”
He believes that the new method can be orders of magnitude faster than hand -painted traditional approaches. If the method is widely adopted, he emphasizes that the Conservatives should be involved at each stage of the process, to ensure that the final work is in accordance with the style and the intention of an artist.
“It will take a lot of deliberations on the ethical challenges involved at each stage of this process to see how it can be applied in a most coherent manner with the principles of conservation,” he said. “We are setting up a framework to develop other methods. Like others work on this subject, we will end up with more precise methods. ”
This work was supported, in part, by the commemorative fund John O. and Katherine A. Lutz. The research was carried out, in part, thanks to the use of equipment and installations at MIT.NANO, with additional support from MIT Microsystems technology laboratories, the MIT Department of Mechanical Engineering and MIT libraries.
