At the forefront of artificial intelligence, NVIDIA seeks to simplify the rendering work of 3D creators.
On the occasion of the Conference on Computer Vision and Pattern Recognition (CVPR) which takes place in New Orleans from June 19 to 24, NVIDIA presents an unprecedented technology that should simplify the rendering of 3D objects.
Extract 3D objects from 2D images
In order to highlight its new technology, NVIDIA has married the setting of the CVPR – New Orleans therefore – to its most formidable musical expression, jazz, in these terms:
” Jazz is synonymous with improvisation, and NVIDIA is paying homage to the genre with AI research that could one day allow creators to improvise with 3D objects created in just the amount of time it takes to hold a jam session. “.
Vice-president of graphic research at NVIDIA, David Luebke is not stingy with superlatives to illustrate the technique which aims to create 3D objects from 3D images. He’s talking about ” Holy Grail that unifies computer vision and computer graphics “.
David Luebke continues by specifying that ” the NVIDIA 3D MoMa rendering pipeline uses modern AI mechanics and the raw computing power of NVIDIA GPUs to quickly produce 3D objects that creators can import, edit, and extend without limitation “.
One hour to generate a 3D model
Through an article published on its blog, NVIDIA explains that ” to be most useful to an artist or engineer, a 3D object must be in a form that can be inserted into widely used tools such as game engines, 3D modelers, and movie renderers “.
The most expert among you probably already know this, but this form is nothing other than what is called a meshin other words, ” a triangular mesh with textured materials “. We can also say that it is the ” common language used by 3D tools “.
NVIDIA rightly points out that the creation of 3D objects was done using “ complex photogrammetry techniques that require a lot of time and manual effort “. Of course, that was before NVIDIA 3D MoMa, which is able to ” generate triangular mesh models in one hour of computation on a single NVIDIA Tensor Core GPU “.
The American company drives the point home by explaining that ” the pipeline result is directly compatible with the 3D graphics engines and modeling tools creators already use “. The video above allows you to see all this in a less abstract way.
On the same subject :
But what does Tolstoy’s “War and Peace” do in NVIDIA’s CUDA-related files?
Source : NVIDIA