Unlocking the Holy Grail: Nvidia Unveils New AI Model to Convert Still Images to 3D Graphics

Nvidia’s new AI model can convert still images to 3D graphics

View insights

Nvidia is trying to give depth to graphics that are shallow. The company is now focusing on editing after converting 2D photos into 3D models, scenes, and videos. GPU giant unveiled today a new AI technique that converts photos into 3D objects, which creators can easily modify. Researchers at Nvidia have created a new pipeline for inverse rendering, Nvidia 3D MoMa, that allows users reconstruct still photos to create a 3D computer-model of an object or scene. This workflow is superior to traditional photogrammetry because it produces clean 3D models that can be imported and edited by 3D visual and gaming engines.

Nvidia 3D MoMa is a photogrammetry program that converts 2D images to 3D models. However, according to reports, it goes a step beyond by generating mesh, lighting, and material information and delivering it in a compatible format with 3D graphics engines. Nvidia claims that 3D MoMa is able to generate triangle mesh models in just an hour, using a single Nvidia tensor core GPU.

David Luebke describes this technique to India Today, Nvidia’s Vice President of Graphics Research, as \”a holy-grail uniting computer vision and computer graphics.\”


Nvidia’s New AI Model can Convert still Images to 3D Graphics