What ‘The Next Rembrandt’ Tells Us About Art and AI

Alice Cattley, Writer
@uxconnections

Revisiting the project amid the pandemic, has our view of virtual art changed?

Science is the antithesis to art – or so we’re often led to believe. On school curriculums and in our universities and museums, we consistently see a distinction between the humanities (languages, painting, poetry, dance) and STEM subjects (Science, Technology, Engineering, Mathematics). Studies have even shown that artists and non-artists have different brains. But in 2016, a portrait that set out to “[blur] the lines between art, technology and emotion” did just that – and it’s as beguiling and bewildering as it was four years ago. 

‘The Next Rembrandt’ isn’t an up-and-coming artist. It’s not even a long-lost painting. Instead, it was the name given to a 2016 project run by ING Bank, J. Walter Thompson Amsterdam, and Microsoft, which involved analysing data points from Rembrandt’s portraits before 3D printing an ‘original Rembrandt’ so statistically accurate it might even have baffled the Dutch master himself.    

The aim wasn’t simply to copy Rembrandt’s art. Rather, it was to create “a visualization of data in a beautifully creative form.” Researchers analysed all 346 of Rembrandt’s paintings to identify key features that were common – or at least, statistically average – across his body of work. This ranged from the subject matter (white moustached male in his forties, wearing a hat and facing to the right) to the specific dimensions (distance between the eyes, width of the face, position on the page). 

The algorithm detected more than 60 data points on each painting, running Linux virtual machines on Microsoft Azure. In collaboration with the Mauritshuis and Museum Het Rembrandthuis in Amsterdam, the team also analysed the depth of the paint which Rembrandt used across his canvas. By 3D printing extra layers of paint in the same areas, they were able to mimic his brushstrokes – recreating not just a static piece of art, but also the physical process behind its creation. 

The finished portrait was unveiled in Amsterdam, the result of 18 months’ work and 150 gigabytes of digital graphics. But the neat calculations and statistics that propelled the project were soon met with some very messy questions about art, AI, and the ethics of imitation. Jonathan Jones in the The Guardian described ‘The Next Rembrandt’ as “a travesty of surface trickery”, an insult to artists whose human understanding would always outdo the soulless paint-by-numbers approach of data analysts and machines. “I cannot wait for Love’s Labours Have Been Successfully Functionalised by William Shakesbot,” he wrote. 

Was Microsoft making a mockery of the heavy emotional labour that elevates a piece of art from aesthetically pleasing to gut-punchingly powerful? Like most questions about art, there’s no straightforward answer. ‘The Next Rembrandt’ was intended as both an homage to the Dutch painter and a starting point for a discussion about the possibilities of data innovation. Only one thing is for certain: it definitely started a discussion.

Today, roughly four years after the painting was created, data innovation and technology is more important than ever. As the world struggles to adapt to the global Coronavirus pandemic, we’re becoming reliant on virtualisation as a way to keep working, socialising, and sane. Many of us are also turning to the consolation of art, and it’s increasingly through technology that we’re able to experience it. 

From MOMA and the Musée d’Orsay to the British Library, National Theatre, and National Gallery, cultural institutions all over the world have made their exhibitions and performances available online. Being able to view their archives through a screen has even been heralded as the silver lining of lockdown – an opportunity to appreciate art without the barriers to entry (expense, location) that often prevent people from seeing masterpieces in person. 

Not many of us would argue that this computerised experience is comparable to standing in front of a painting, but the differences aren’t always inferior. You can’t zoom in on artwork in a gallery or museum, and you do have to wait for people to move out of the way to see the most popular paintings. (If the crowds in the Louvre have ever put you off from viewing the Mona Lisa, you can download the gallery’s ‘VR – Mona Lisa’ app, which gives you a 360-degree guided tour of da Vinci’s work). 

Returning to ‘The Next Rembrandt’ four years after its creation, the relationship between paintings and pixels has changed. Director of SMB Markets at Microsoft, Ron Augustus, described the project as a combination of his two passions: the way technology can benefit companies, and art history. “It is not often that these two worlds come together”, he said. But they’ve arguably come together again during the pandemic – when, without algorithms, most of us wouldn’t be able to enjoy art at all.