Video games have become so realistic that we often wonder how their creators manage to do it. Now Nvidia has an answer. The company has created an artificial intelligence tool that is capable of transforming videos into virtual landscapes.
Nvidia gave a short presentation at NeurIPS AI in Montreal. The company used its own supercomputer, the DXG-1, equipped with Tensor Core graphics cards, during the test. With this setup, Nvidia was able to put the theory into practice.
The research team created a neural network that "maps" what it sees and creates it in Unreal Engine 4.After the 3D mapping is done, the same neural network also creates the images or color combinations present in the original video.
“Nvidia has been creating graphics for 25 years. But this is the first time that something like this has been done with the help of the neural network that our team created.” In addition, a company representative added that with this new technology, developers will be able to create new content at a significantly lower cost than they have had to date.
Finally, the news is very positive news for game developers and creators of virtual reality hardware. However, this new technology requires supercomputing power to run. In addition, it is still in the testing phase, so its operation is not in its final stage. We do not know when (and if) Nvidia's service will be available for public use.
You can read more here.
