Nvidia has showcased a new AI that can transform 2D photographs into a realistic 3D world in just seconds. 3D rendering, used heavily in design, architecture, industry, and science, is known for taking up large processing resources. Rendering a 3D object or a 3D digital world is also time-consuming, even for the most powerful computers.
Today, AI is accelerating tasks and processing massive amounts of data at incredible speeds. From medical to autonomous or self-driving technology, AI applications are endless. AI is also playing a leading role in the creation of the metaverse, as the internet evolves and looks to bring real objects and human users into a digital world.
Related: M1 Ultra Vs. RTX 3090: Did Apple Really Beat Nvidia?
Nvidia released a new video to showcase its latest AI innovation. In seconds, Instant NeRF can process 2D images and render them into a 3D complex world. NeRF, or neural radiance fields, is beginning to position itself as a popular new technology for various uses and Nvidia says that its AI is now the fastest NeRF to ever exist. In the video, Nvidia pays tribute to the iconic imagery of the world of Andy Warhol taking a polaroid picture, with the polaroid representing how an image can be instantly generated.
The Inverse 3D Digital World
There are several reasons why Nvidia created this instant 3D inverse world-generation technology. The company explains that it can be used to create avatars, virtual worlds, level-up conferences and events with 3D capacities, create digital 3D maps, and much more. The vice president for graphics research at Nvidia went as far as to say that NeRF could one day be as “important to 3D as digital cameras and JPEG compression have been to 2D photography.”
NeRF collects input data from a collection of 2D images before then starting a process known as inverse rendering. During this inverse rendering stage, AI is used to fill in the gaps and figure out how light behaves from all angles in a 3D world. Nvidia says Instant NeRF is super fast, making it 1000 times faster than most NeRF techniques. As a result, the 3D world takes just seconds to be generated and a single render can be done in just tens of milliseconds.
Robots, autonomous and self-driving vehicles, smart streets, digital twins, and smart city AI systems all require a certain level of environmental awareness to operate. 2D cameras are often used to create this awareness but camera input cannot estimate velocity or distance very well. LiDAR technology has been used to scan environments and provide this missing information in the past. Nvidia believes that Instant NeRF could now be used to create realistic world environments and provide AI systems with environmental awareness.