This week, Nvidia presented its 3D MoMa method in more detail. It is a technique that allows 3D objects to be created quickly to work with in graphics engines.
Using instruments as an example, the vendor demonstrated 3D MoMa at the Computer Vision and Pattern Recognition Conference (CVPR) 2022.
It would be suitable to speed up the workflow for architects, designers, concept artists and game developers. Imported objects can be widely manipulated, for example to change the scale or material or to conduct experiments with different lighting effects. This includes so-called inverse rendering, which is aimed at creating 3D objects from 2D photos. This is quite a challenge.
The video shows quite well how this can work. For example, in the video you could create a 3D model of a trumpet from 120 photos. In fact, Nvidia says it’s working to make creating 3D models easier and simpler with the help of machine learning and AI. The idea is also to keep the files as compatible as possible so they can be seamlessly added to popular game engines, 3D modeling tools and movie rendering software.
The material of the created models can then also be exchanged via Nvidia Omniverse, for example: A plastic trumpet can become a more elegant model made of gold, for example. The technologies behind 3D MoMa are also described in more detail by Nvidia in this blog post.