Artificial intelligence could be about to be made 3D modeling Much faster and easier, if Nvidia has its way.
Fresh out of the company’s research lab, Neuralangelo is an AI model that transforms 2D videos into “super high resolution” 3D structures.
“Creative professionals can then import these 3D objects into design applications, and modify them further for use in art, video game development, robotics, and industrial digital twins,” the company explains in its latest paper.
High accuracy in the real world
Using videos as reference points is nothing new — even when you throw relatively new, experimental AI into the mix. But, as Nvidia notes in its recent paper, “existing methods struggle to restore the detailed structures of real-world scenes.” In other words, they can repeat things and scenes, but it won’t Feel vibrant.
This is the high precision that Neuralangelo promises. According to Nvidia Research, the brains behind the model, they are able to vividly recreate different materials, such as glass, marble, and even ceiling panels. The model has also demonstrated its ability to reconstruct the interiors and exteriors of buildings.
“The 3D reconstruction capabilities offered by Neuralangelo will be very useful to creators, as it helps them recreate the real world in the digital world,” said Ming-Yu Liu, senior director of research and one of the paper’s authors.
The applications of Neuralangelo in developing arts and games are already clear: think of vast cityscapes grand theft autothe realistic historical setting of Doctrine killer. Entire virtual cities based on videos recorded on a smartphone could soon be realized. But it’s only a matter of time before AI models like these also form an essential part of everything Best architecture software And Best landscape design softwareAnd If it can really simplify refactorings in real life.
In a demo released by the company, the results are admittedly impressive: highly detailed AI renderings of Michelangelo’s David based on video footage amount to a boast of “large-scale, detailed scene reconstruction from RGB video.” But we’re still a long way from dropping AI 3D models directly into projects without some finishing touches.