While virtual reality has been the way to realistically bring objects into the real world, prolonged use of VR headsets can cause nausea and eye strain. The use of 3D holograms, a fantasy of technology lovers for decades, would alleviate these problems. However, elaborate processes and specialized equipment are needed to generate them.
The good news is that researchers at the Massachusetts Institute of Technology (MIT), along with Sony, have found a way to create 3D holograms in real time, using a regular computer and a method called tensor holography. The trick? Teach a neural network about physics, so that it can understand visual and depth information.
Real-time 3D holograms
In an article published in the scientific journal Nature, the team explained that the new deep learning method, called tensor holography, is capable of “running instantly on a simple computer”. The researchers developed this AI using trainable tensors that were taught to process visual and depth information similarly to humans.
As reported Entrepreneur, this technology requires a large and comprehensive dataset. A set that takes into account the color and depth data of each pixel. This is necessary, since all points in an image vary in depth, and applying one operation across the board would not work.
Tensor holography, a "computer egg" of Columbus
Instead, the researchers found that tensor holography was capable of generating 3D holograms from typical computer-generated images with depth data, calculated from a multi-camera setup, in just a few milliseconds.
This technology could be a game changer for virtual reality, medical imaging and even 3D printing, as thanks to tensor holography and 3D holograms it would be faster and more precise than layer-by-layer printing.
See how it works this artificial intelligence in the video below. And good virtual reality at all.
