As tech companies are finding new and unique use cases for artificial intelligence during the AI revolution the industry finds its self in, developers are coming up with new techniques to showcase their AI models. NVIDIA has revealed a new AI they are calling “Neuroangelo”, which is able to construct 3D models of objects gathered from 2D video footage.
NVIDIA claims that Neuroangelo is able to accurately capture textures from video footage, including complex patterns or variations in shading, with results that “significantly surpasses prior methods.” Co-author of the paper on Neuroangelo, Ming-Yu Liu, says that the technology will “eventually enable developers to import detailed objects — whether small statues or massive buildings — into virtual environments for video games or industrial digital twins.”
Apparently the technology won’t keep users waiting long, either, with NVIDIA telling Engadget that Neuroangelo can create 3D scenes in under two hours.
