347 years after his death, one of the greatest Masters of all time, Dutchman Rembrandt, is brought back to life to create one more masterpiece. But this time, data is the painter and technology the brush: The Next Rembrandt.
The Next Rembrandt is a 3D printed painting made completely out of data derived from Rembrandt’s total body of work. All of his 346 paintings were analysed using hi-res 3D scans and digital files upscaled by a deep learning algorithm for maximum resolution.
Facial recognition and machine learning software was designed that could understand Rembrandt’s style and could use those learnings to generate new facial features, which were assembled based on his use of geometrical proportions.
Once the 2D image was complete, a height map was created to mimic the brushstrokes used by Rembrandt. The file was brought to life through an advanced 3D printer that printed 13 layers of paint based UV-ink.
On the 5th April 2016, The Next Rembrandt was unveiled at an exhibited in Amsterdam, the place where Rembrandt lived and worked.
People from all over the world came to experience the unveiling and exhibition of The Next Rembrandt in Amsterdam. There the conversation about where data and technology can take us, started. A conversation that went global.
The Next Rembrandt was global trending topic on Twitter, reporting almost 10 million Twitter impressions on the day of the unveiling.
Over 1.400 articles were written about the 3D printed painting made from Rembrandt-data.
All the major news networks covered the unveiling and commented on the meaning The Next Rembrandt can have for future innovations.
Globally over 1.8 billion media impressions were measured.
The total earned media value of the project amounted to 12.5 milllion euro.
On launch day Fortune Magazine reported an increase in stock value for ING (ING ^1.22%) and partner Microsoft (MSFT ^ 0.49%).
And in the weeks after the launch, Google reported an increase of ING’s Search Interest by 61.29%, and Microsoft’s by 20%.
Next to the ongoing global conversation about what this painting means for our industry and the future of creativity, there was another fundamental result: the technology developed for The Next Rembrandt is now used for the restauration of damaged and partially lost Masterpieces.
Data and creativity have been getting closer together. Data is inspiring ideas, and is becoming part of ideas. The ultimate destination of this trajectory is data and creativity becoming one - data so inextricably linked with an idea that its presence is both fundamental and invisible.
The Next Rembrandt represents this leading edge role of data.
It takes perhaps the highest form of creativity– fine art –and demonstrates how it can be beautifully re-born with simple zeros and ones.
The work raises provocative questions for our industry to consider, and even for the wider world of art to ponder too.
The Next Rembrandt is a painting fully made out of Rembrandt-data. It is bringing back to life the Master of Light and Shadow by a multidisciplinary effort by academic experts, data scientists and software engineers that discovered innovative ways to leverage historical and statistical data. All of his 346 paintings were analysed on a pixel by pixel basis and an extensive database was built. Supporting partner Microsoft lend their cloud platform Azure to host, process and analyse all the data gathered.
First, statistical and demographical data were used to determine the subject of the painting. After that, facial recognition and machine learning software was designed that could understand Rembrandt’s style and used those learnings to generate new facial features, which were assembled based on his use of geometric proportions.
Finally, existing hi-res 3D scans of Rembrandt paintings were used to teach a computer how to apply brush strokes like Rembrandt. That information, assembled in a height map was used to create the texture of The Next Rembrandt painting.
Resulting, after more than18 months, in what could have Rembrandt’s next portrait, fully made of data of over 11 billion pixels.