The model required only a few seconds to take a few dozen photos, and the camera angles, data about the company (Photo: Playback/Nvidia)
NVIDIAA technology multinational based in the United States, a research project has developed that allows to convert 2D photos into 3D scenes on the fly. Uses the process, known as inverse rendering Artificial Intelligence (AI) To reconstruct a 3D scene from a group of 2D images taken at different angles.
The NVIDIA research team has developed an approach that accomplishes this task almost instantly, and is one of the first models to combine lightning-fast neural network training and fast rendering. The company has the technology to be applied to a popular new technique called “neural radiation field” or Aniarf. The result, called Instant NERF, is the fastest technology to date, according to the company.
The models needed seconds to take a few dozen photos, as well as data on the camera angles from which they were taken. With it, it can render a 3D scene in tens of milliseconds. This week appear in a session NVIDIA GTC, immediate access Aniarf avatar or view the virtual world, video conference participants and can be used to re-create the scenes to capture in 3D to their environment or 3D digital maps Is.
The technology could also be used to train robots and autonomous cars to sense the size and shape of real-world objects, capture them or 2D video images. It can also be used in architecture and entertainment to quickly generate digital representations of real environments that developers can modify and build.
“Instant NERF can be just as important to 3D as digital cameras and JPEG compression were to 2D photography, greatly increasing the speed, ease and range of 3D capture and sharing,” said David Lubke, vice president Graphics search engine said.
Interpretation techniques See video:
Would you like to see special content Poca NEGÓCIOS? Get access to the digital edition,
