HomeTechnologyTech newsArtificial Intelligence will create holograms in real time on your phone

Artificial Intelligence will create holograms in real time on your phone

A new method created by MIT engineers in the USA, manages to generate holographic images in three dimensions and in real time. The idea does not seem new, but now the whole process of projecting these holograms could be done from simple and inexpensive devices, such as a cell phone or an ordinary laptop.

Holographic images such as those seen in cinema, capable of confusing a less attentive observer, are still far from becoming reality. Today’s 3D holograms even offer a good representation of landscapes and objects, with changes in perspective that allow the human eye to adapt better to differences in depth.

But the biggest problem with this technology is the price. In order to generate minimally satisfactory holograms, supercomputers capable of carrying out physical simulations that take a long time and do not deliver a very convincing result, or that come close to the Star Trek Holodeck, are needed.

What’s new

MIT researchers used a system based on deep learning algorithms to create holograms instantly. “People previously thought that with existing hardware for the consumer, it would be impossible to make 3D holography calculations in real time,” says Professor Liang Shi.

In the so-called “tensor holography”, the team designed a neural network known as convolutional, a processing technique that uses a chain of trainable tensors to mimic how humans work with visual information. Since this training needs a large amount of data to work, the researchers built a database with four thousand pairs of computer-generated images.

Each of these pairs started to combine a different image, with color and depth information for each pixel analyzed. The holograms of this new database were created with scenes containing complex complex shapes and colors, distributing the depth of the pixels between the foreground and the background.

The algorithm

After learning from analyzing each pair of images, the tensor network algorithm adjusts the parameters of your calculations, increasing the ability to create holograms much faster and more efficiently than using calculations based on conventional physics.

“In mere milliseconds, tensor holography can create holograms from figures with depth information, which are provided by computer-generated images and can be calculated from a LiDAR sensor, which already equips some newer smartphones,” says professor Wojciech Matusik.

In addition to the advantages of speed and quality to design holograms in three dimensions, the tensioner network does not need advanced configurations to run. A compact system requires less than a megabyte of memory to run smoothly. Future 3D

For researchers, holograms produced in real time can enhance Virtual Reality systems, eliminating some side effects of VR glasses such as nausea and eye strain after long periods of use.

In another application, 3D holograms could be used in the development of 3D volumetric printing equipment. With this technology, printers that today print objects layer by layer, could use a simultaneous projection scheme in three dimensions to print more quickly and accurately.

In the field of medicine, holographic screens can be used for ophthalmic diagnostics, helping to correct eye disorders by displaying clearer images, which can be viewed from several different angles. It would be as if the doctor could “roll” the eye to find the best way to correct the problem.

Related news