In Bringing the Dutch Master Back to Life, Has Digital Technology Finally Killed Painting?
Since the 1830s, when the first photographs were made, the "Death of Painting" has been proclaimed more times than anyone can remember. Throughout the almost two centuries that it has existed, photography has rocked painting, challenging its technical primacy and giving rise to both reactionary movements, like Abstract Expressionism, and assimilatory ones, like Photorealism. Painting seemed to be in crisis, fighting to stay relevant. Then Susan Sontag pointed out photography's intrinsic differences from painting (more like "footprints of reality" than painting's synthesized "expressions of interpretation"), and painting survived.
Now, digital technology could be a viable rival, and artists who paint are paying attention. Painting's latest challenge comes from The Next Rembrandt project, led by Bas Korsten of J. Walter Thompson advertising agency in Amsterdam, and a collaboration between ING, Microsoft, the Mauritshuis and Het Rembrandthuis. Korsten brought the worlds of banking, tech and art together to "bring the Great Master back to life" and "distill the artistic DNA from his work," to create a "typical" Rembrandt painting with cutting-edge digital 3D imaging technology, algorithms and Big Data.
The goal was to create a new masterpiece, algorithmically designed to pass as a Rembrandt, and in the process unlock the secrets of artistic genius and "make life itself more beautiful."
The Dutch master was an understandable choice, since there was enough consistent and distinctive source material for a Big Data project like this to be feasible at all. Moreover, Rembrandt is the Ur-painter, an icon of genius who took the oil painting medium to a new level of virtuosity, imbuing his work with psychological atmosphere and theatricality, while deftly exploiting all registers of the oil medium, from thick to thin, and from dark to light. Rembrandt was chosen because he represents a summit of human artistic achievement, his masterpieces therefore the hardest objects for a machine to replicate. (As a prolific printmaker and savvy promoter, it's also very possible that he would have embraced this new technology himself.)
Using Microsoft's Azure cloud platform, the team built a database on the physical artworks, including high-resolution 3D scans of over 160,000 painting fragments from Rembrandt's oeuvre, as well as height maps, which would help to render the final object in 3D, and give it the look of an authentic painting. Then, using deep-learning algorithms, based on 60 data points around the face of his portrait subjects, a typical Rembrandt painting was conceived, a digital file comprising over 140 gigabytes of data. The image was printed in 3D with 13 layers of paint-based UV ink, to mimic the illusion of impasto (visibly thick, brushy paint).
Naturally, this bristles with some painters, who have spent years developing an analogue art practice built on discipline and sacrifice - to be told that every part of their craft can now be achieved with buttons and computers. But the 18-month project met with instant acclaim, and this summer, the Cannes Lions International Festival of Creativity awarded TNR the first ever Grand Prix in its year-old category "Creative Data," sponsored by, um...ING. But does the TNR portrait have the hallowed "aura of the original?" Like Gutenberg's press in the 15th century, this technology challenges the idea of originality, as it enables multiple copies of the same, high-resolution image to be recreated ad infinitum.
Not only is the illusion of TNR being a "painting" successful, but the painting within the painting too: the as yet untitled TNR portrait is a convincing depiction of a Rembrandt subject: a middle-aged, 17th-century bearded Dutchman. It has character, personality and presence just as an original does.
As an oil portrait painter myself, I have questions about the final result that cannot be resolved without seeing it in person. For example, does the transparency of oil paint transfer successfully to the "paint-based UV ink," or are the 12 lower layers purely 'bulk'? Are there any traces of digital noise? Do the brushstrokes read as authentic, hand-made daubs, or statistically averaged forms, devoid of character? How are accents (spontaneous, often improvised final touches to a painting) treated?
Just how "beautiful" is this robo-portrait?
The technologies at work here will one day create opportunities for artists. Imagine releasing new paintings like iTunes downloads, that anyone in the world could bring to life on their home printer. Or creating copies of paintings at varying scales for planning purposes, for product diversification or for flexibility when hanging work. Imagine hybridizing your favorite artists of the past. Or creating paintings without having to touch oil paint...
Sadly, the revolutionary part about TNR - the algorithmic creation of an "original" painting based on existing artistic data - is also the least artistically original, at least to an analogue artist. The so-called "distillation of artistic DNA" sounds like a euphemism for banalization.
So does this represent an end to the tradition of painting? It may seem paradoxical to ask this question, given the medium's intrinsic and visible role as source material in this project. But the disappearance (and in some cases, resurrection) of analogue media in the last half century still has painters wondering about the future of their craft. Whereas photography (which automated half the picture-making process, the analytical side) represented only half a challenge, this technology also automates the production and execution of the final image, and could therefore be an existential threat to oil painting, one day. Digitally applied "paint-based UV ink" could also be neater, less toxic, more efficient, more accessible, and more permanent than oil paint. It may open up more options for those visual artists working under the constraints of the notoriously difficult oil painting medium.
Translation, Imitation, Emulation
This is not the first time that mass-production has challenged our ideas of originality. When dealing with the abundance of excavated marble remains labeled as "Marble Roman copy of a Bronze Greek original," classical art historians used three literary terms to classify them: Translation the straightforward marble copy of an individual bronze statue. Imitation the creation of a new marble statue based on multiple examples in bronze. Emulation a new composition making use of some classic forms and elements, but conceived afresh.
In merging high-resolution photo technology with algorithmic design and the latest generation of 3D printers, The Next Rembrandt has created a system of digital mass production on a level with the "imitators" of the Roman world. Emulation may prove more of a challenge.
Project The Next Rembrandt
The Next Rembrandt is a project that will, after nearly 350 years, bring to life painter Rembrandt van Rijn, with the help of computers and 3D printing. In collaboration with Tu Delft University, ING and Microsoft sponsors, the team from the Mauritshuis museum in The Hague, which showcases the paintings of the Golden Age of Dutch and Flemish painters, created a new painting, after conducting a precise computer analysis of all of Rembrandt's works. They began with the digitalization of several of his paintings. Through computer analysis they studied the portrait faces, their features, traits, the coloring, the thickness of the paint and so on. Based on the extracted data, special computer algorithms then discerned the traits specific to Rembrandt and created a new portrait, which was first 3D printed (to recreate a three-dimensional color effect on canvas) and finally computer-painted to look as if it were by Rembrandt's hand.