Nvidia Showcases AI 3D Model Maker That Brews Objects In Under 1 Second

Video screenshot via Nvidia

Nvidia is waking up the 3D graphics world with the aptly named LATTE3D, an innovative artificial intelligence model that transforms text descriptions into 3D models in a split second. The tool gives digital creation an almost caffeinated jolt—with the AI giant affectionately calling it “instant latte”—and provides an accelerated way for professionals in diverse fields to visualize their ideas in three dimensions.

Operating on a single GPU, LATTE3D can produce detailed 3D images of objects and animals in less than a second. This efficiency is a notable improvement over previous technologies, which could take up to 12 seconds for a similar output. The technology could be an essential asset for anyone looking to quickly develop virtual environments, from video game designers to advertisers and educators in robotics.

“A year ago, it took an hour for AI models to generate 3D visuals of this quality—and the current state of the art is now around 10 to 12 seconds,” explains Sanja Fidler, vice president of AI research at Nvidia. “We can now produce results an order of magnitude faster, putting near-real-time text-to-3D generation within reach for creators across industries.”

LATTE3D not only offers speed but also versatility. It allows users to explore various 3D shape options from a single text input, providing flexibility in design choices. These models can be refined and integrated into different software applications or platforms, including NVIDIA Omniverse, which supports workflows based on Universal Scene Description (OpenUSD).

Video courtesy of Nvidia

Image courtesy of Nvidia

The potential applications of LATTE3D extend to various specialized areas. Its adaptable architecture means it could be used for tasks ranging from landscape design, where it could quickly generate plant life for garden renderings, to home simulations for robotic training. The model’s effectiveness is enhanced by its training on NVIDIA A100 Tensor Core GPUs and a wide range of text prompts generated using ChatGPT, ensuring it can accurately process and execute diverse descriptive commands.

[via Nvidia, images courtesy]