The mysterious images hidden in the masterpieces of art

The history of painting always contains masterpieces that have been lost, or destroyed. There were even paintings painted over existing ones, and this was only discovered many centuries later, with the help of technology.

After Mount Vesuvius had a terrible eruption in AD 79, engulfing the city of Pompei and the surrounding area, many libraries with ancient books were turned to coal, and paintings were wall too fragile to be excavated.

One of the masterpieces of the Dutch painter Rembrandt - The Night Watch (1642), was cut to fit in the door of Amsterdam's city hall in 1715. For a long time, art researchers accepted the fact that these losses would be forever.

Few people know that Rembrandt's The Night Watch (now on display at the Rijksmuseum in Amsterdam) was cut off on both sides.

Looking for hidden things

In recent years, with the help of ever-evolving technology, there is a wave going against this view, aiming to recover parts or all of the lost works.

Just like AI, machine learning, also known as machine learning, has crept into many parts of everyday life. Now, this technology is helping to revive precious artefacts that have disappeared over time.

Throughout history, it is not uncommon for artists to paint over an old painting. This will be detected when one places a picture under an X-ray.

A project called Oxia Palus in recent years has attracted attention. Founded by two doctoral students, George Cann and Anthony Bourached, this project aims to revive the original paintings hidden under the paintings.

Oxia Palus began publishing their first images in 2019, relying on AI to determine the original color and texture of overdrawn paintings. The two researchers then used 3D technology to reconstruct the original paintings. These paintings will be sold as a single print.

"I think a lot of people are interested in this because they like new things. It would be great to see these masterpieces, because they exist but you couldn't see them before, because you couldn't see them. Who wants to shave off the top layer of a million-pound painting," Bourached said.

George Cann and Anthony Bourached also want to remind people that using technology to restore paintings is not an easy thing.

"The articles brought attention to the topic of 'AI technology to restore a picture', as if we could press a button and it would all be over," Bourached added.

According to Mr. Bourached, this can give the impression that scientists are giving full power to machines, when in reality there is a lot of human contribution in the process.

Specifically, team members will have to collect data from the artist's other works to understand the author's painting style, then filter the results from the X-ray scanner to remove elements that are not. necessary, needs.

According to Cann, the final images from the X-ray scanner will be "our understanding of what's underneath". The two scientists admit the process is not very meticulous, as they see it as an experiment in their spare time, not a well-invested project.

The image of a woman hidden under the painting Portrait of a Girl (1917) by Italian artist Amedeo Modigliani.

"One thing we want to emphasize, is that although this is a relatively naive approach, it is completely doable. I hope that others will adopt this nascent field and do better things," Bourached said.

In the process, Mr. Cann and Mr. Bourached's team faced a major roadblock. That is, the use of X-rays to scan pictures - first used in the 19th century - proved ineffective in gathering information.

The power of technology

Conservators sometimes also have to sample directly on the painting to learn about the type of fuel, the ingredients that make up the pigments used in the painting. However, modern scanning technologies allow them to collect all this data without having to touch the picture.

Five years ago, the National Gallery in London purchased a state-of-the-art scanner capable of collecting a great deal of data about a painting. The problem is that the data is so overwhelming that most cultural institutions are not equipped with the resources to analyze them.

To undo this knot, now cultural institutions like the National Gallery have partnered with universities to gain access to high-end computers and the expertise of scientists.

In one such project called Art Through the ICT Lens (ARTIC), the National Gallery joined forces with University College London (UCL) and Imperial College London to analyze the Dona Isabel de Porcel painting. by the Spanish painter Francisco de Goya, painted in 1805.

This work by Goya shows a woman wearing a black shawl. However, in 1980, when it was put into the scanner, a second mysterious portrait of a man in a coat was discovered underneath this painting.

To get a clearer picture of this hidden picture, the researchers had to place Goya's painting under the scanner several times, combining different regions of the electromagnetic spectrum. Initially, this process had to be done by human power, but now the whole process is left to computers.

UCL researchers led by Dr. Miguel Rodrigues had previously participated in a similar project to restore the Dutch artist Jan van Eyck's work Ghent Altarpiece (1432).

While some research efforts help us understand what lies beneath the paintings, others help scientists restore what has been lost. In 2011, the Rijks museum (Rijksmuseum) in Amsterdam invited the public to see a full reproduction of Rembrandt's The Night Watch.

The original version of The Night Watch was fully reconstructed using AI technology.

Luckily for us, before this huge painting was cut to enter Amsterdam's town hall, a little-known artist named Gerrit Lundes fully reproduced the original. However, there are still very small differences between the painting styles of the two artists, and the scientists had to guide the algorithm to reconstruct the missing part of the painting in the correct Rembrandt style.

"The algorithm even mimics tiny cracks in pictures, so I would say it's as science-based as possible. That doesn't mean it's perfect, technology will always improve." , said Rob Erdmann, senior scientist at the Rijksmuseum.