tech2 News StaffApr 08, 2016 10:52:14 IST
A team of scientists and engineers have recreated the style of Dutch painter Rembrandt Harmenszoon van Rijn, 350 years after his death. The team used a combination of data analysis, machine learning and 3D printing to make The Next Rembrandt. The project was a collaboration between Microsoft, ING, Delft University of Technology, The Mauritshuis art museum and the Rembrandt House Museum.
The entire collection of Rembrandt's painting were analysed by the team, pixel by pixel. They generated 150GB of high resolution scans of his works. Machine learning algorithms were used to upscale these high resolution scans, to maximize the resolution and quality. An analysis of 346 paintings was done to determine the subject. Since most of his paintings were portraits, the team focused on analysing the portraits after comparing the age, gender, race, clothing and facial features.
This allowed the researchers to decide their subject – a Caucasian male between the age of 30 and 40 with a beard, wearing a black shirt and hat, with a white collar. The next step used AI to generate new geometries based on the features painted by Rembrandt. A facial recognition algorithm was used to identify and classify the most typical patterns used by Rembrandt. Over 500 hours of rendering was used to generate the new features and bring them together into a 2D painting.
The final step was recreating the brush strokes on canvas. High resolution 3D scans were used to map the thickness or height of the paint strokes used by Rembrandt. There were two maps, one for the canvas, and one for the layer of paint. A 3D printer was used to output 13 layers of ink one on top of the other, to accurately simulate the brush strokes.
Find our entire collection of stories, in-depth analysis, live updates, videos & more on Chandrayaan 2 Moon Mission on our dedicated #Chandrayaan2TheMoon domain.