FROM 2D PICTURE REAL 3D AVATARS IN FEW SECONDS

Experts have been improving tools that turn 2D images into 3D models for several years, so the final renders have more details.

Since the talk about the metaverse began, more and more companies are working on their own versions of artificial worlds.

In order for virtual worlds to function, users need 3D avatars and a reconstruction of the environment that is in line with that.

In this context, Nvidia is working on neural radiance fields technology (NeRF), which uses artificial intelligence to make 3D models from ordinary photos.

This is possible because the empty space is filled with additional pixels, which are filled with color based on predictions of how the light in the environment will behave in relation to the object.

The company introduced the tool recently, and the prefix Instant was added to the name because the creators claim that compared to the previous variants, which took a few minutes (after several hours of learning) for this job, the new version is up to a thousand times faster. 

It is emphasized that in practice the speed depends on the complexity of the object and the desired resolution of the final product.

Experts have been improving such tools that turn 2D images into 3D models for several years, so the final renderings have more details thanks to that. 

Nvidia points out that the Instant NeRF model is one of the fastest at the moment since the process was shortened from a few minutes to just a few seconds, so the 3D result was obtained almost immediately.

The company has also published the source code of the project so that others can contribute to it, and technology is advancing faster and faster, so it is certain that it will soon change and improve metaverses.