From manufacturing robots and self-driving cars to smart assistants and disease mapping, Artificial Intelligence (AI) software has many use cases. Recently, things have taken a more artistic turn, and AI-powered software began to be used to recreate lost art. Thanks to newly developed algorithms, we can now hear Beethoven’s last symphony or admire Klimt paintings although they have been destroyed by the Nazis.
Google used AI to recreate Gustav Klimt paintings
Google recently used AI-powered software to recreate Klimt paintings destroyed by the Nazis. In 1945, in the final months of World War II, an Austrian castle holding numerous important works of art burst into flames. The Nazis used the castle as a repository for important art pieces and historians now believe they deliberately set fire to the castle to destroy those they considered “degenerate”.
Three of the paintings destroyed were the so-called “Faculty Paintings”, works by the famous Austrian painter Gustav Klimt called “Medicine”, “Philosophy”, and “Jurisprudence”. All that remained of these paintings are the black and white photographs that were taken of them. But Google, in partnership with the Belvedere Museum in Vienna, decided to use AI to recreate the paintings in color using machine learning techniques.
The researchers used multiple sources to create an algorithm that colors the paintings as they would have looked. Among the sources used were newspaper accounts from the time the paintings were presented that described their colors, such as the vibrant green used for the ‘Philosophy’ sky. Google’s AI then analyzed other Klimt works painted around the same period.
“For me the result was surprising because we were able to color them even in places where we had no knowledge but with machine learning, we have good guesses that Klimt used certain colors,” says Dr. Franz Smola, curator at the Belvedere Museum, in a press release issued by Google and cited by Mashable.
Beethoven’s last symphony was finished with the help of artificial intelligence
With a helping hand from AI, Beethoven’s Tenth Symphony has now been completed, more than two centuries after the German composer’s death.
Beethoven began composing the symphony three years before his death but never finished it, the only surviving records of it being some of his notes and musical sketches. But in 2019, nearly 250 years after his death, a team of musicologists and AI experts began reconstructing the symphony.
“It was a huge challenge. We had to teach the AI to take a short passage or just a musical motif, and develop it into a more complex musical structure, just as Beethoven would have done.”, Dr. Ahmed Elgammal, a professor specializing in artificial intelligence at Rutgers University, explained.
Scientists used works by Beethoven, as well as other composers of the time such as Bach, Haydn, and Mozart to train the AI, with the musical lines it produced then being examined and arranged by musicologists.
The full recording of Beethoven’s Tenth Symphony was released on October 9, 2021, the same day as the world premiere performance took place in Bonn, Germany – the culmination of a two-year-plus effort.
It’s important to mention there have been some past attempts to reconstruct parts of Beethoven’s Tenth Symphony. Most famously, in 1988, musicologist Barry Cooper ventured to complete the first and second movements. He wove together 250 bars of music from the sketches to create what was, in his view, a production of the first movement that was faithful to Beethoven’s vision.
Hidden Picasso painting brought to life by AI software
AI experts Anthony Bourached and George Cann recently revealed and brought to life a hidden Picasso painting. The original was discovered hidden under Picasso’s 1903 masterpiece ‘The Blind Man’s Meal’ after the piece was x-rayed in 2010.
“At the time that Picasso painted The Lonesome Crouching Nude and The Blind Man’s Meal he was poor and artist materials were expensive, so he likely painted over the former work with reluctance,” Cann said in the statement.
Bourached and Cann, who are both doctoral researchers at Britain’s University College London, sought to recreate the hidden nude by training AI to replicate Picasso’s brushstrokes using an algorithm that allowed it to analyze dozens of his past works.
Using the 2010 x-ray as a starting point, the duo used a combination of Artificial Intelligence, advanced imaging technology, and 3D printing, thus being able to reproduce a version of the painting, which was given texture and printed onto canvas.
Art is a way of “documenting information” about moments in time, including in an artist’s life, Bourached told NBC in a phone interview. “And I think (this) is a new frontier. I think the future of AI helps us to understand ourselves better as a society.”
