MIT researchers have pioneered a new method to create ‘holograms’ for VR, 3D printing and medical imaging, capable of running on a smartphone.
The new method, called tensor holography, using a deep learning-based method to run on
a laptop and generate real-time ‘holograms’.
The work was supported by Sony, with the
team developing a convolutional neural network, using a chain of trainable tensors to mimic how
humans process visual information, building a custom database of 4,000 computer-generated images, including colour and
depth information for each pixel with a corresponding hologram.
The tensor ‘holography’ can craft
holograms from images with depth information, provided by computer-generated images and can be calculated from
a multicamera setup or LiDAR sensor, requiring less than 1 MB of memory.
Wojciech Matusik,
advisor and co-author of the study, MIT, commented: “It’s a considerable leap that could
completely change people’s attitudes toward holography. “We feel like neural networks were born for this task.”