Instant NeRF is a novel technology developed by the NVIDIA Research team that can transform numerous 2D photos into a 3D scene. The method employs Neural Radiance Fields, a type of artificial intelligence that can be trained to recreate a 3D scene from a collection of 2D photos collected from various perspectives. Instant NeRF, according to NVIDIA, is one of the quickest NeRFs available, requiring only seconds to train on a few dozen still photographs and rendering the final 3D scene in tens of milliseconds.
Instant NeRF might also be used to produce avatars or scenes for virtual worlds, capture video conference participants and their environments in 3D, or reconstruct scenes for 3D digital maps, according to the researchers.
According to NVIDIA, gathering data for NeRF is similar to a photographer trying to catch a celebrity's attire from every perspective — the neural network requires a few dozen photographs taken from various locations surrounding the area, as well as the camera position of each of those shots.
By anticipating the hue of light radiating in every direction, from any location in 3D space, the AI can also fill in the blanks and reconstruct the scene. The method can even work around occlusions, which occur when objects in one image are obscured by impediments in another.
"If traditional 3D representations like polygonal meshes are akin to vector images, NeRFs are like bitmap images: they densely capture the way light radiates from an object or within a scene," comments David Luebke, Vice President for graphics research at NVIDIA. "In that sense, Instant NeRF could be as important to 3D as digital cameras and JPEG compression have been to 2D photography – vastly increasing the speed, ease and reach of 3D capture and sharing."
留言