Nvidia has completed the EG3D neural network it developed with experts from Stanford University. Based on artificial intelligence, the technology amazes with its ability to produce highly realistic 3D models from 2D images.
Lenovo announces AI-powered ThinkCentre computer
The source code of EG3D, which has been worked on for more than a year, is finally available to everyone. Engineers say that the algorithm they created is faster and more accurate than their peers. The new neural network can be used in different projects in the future, including creating game characters.
Does not require very high computing power
In the video released by Nvidia, we see that EG3D has managed to render extremely realistic images of human and cat faces from different angles, including famous personalities. It is noteworthy that the resulting models have the correct geometry.
What’s even better is that it doesn’t require huge computing power. It can even be used on laptops with high-end graphics cards. Since Nvidia has made the source code of EG3D available, those who want to have the chance to test it. All necessary files and installation instructions are available on GitHub.
15049
