NVIDIA through its research department presented a work generated entirely by AI
The department called Deep Learning of NVIDIA through its VP Bryan Catanzaro has stated that for NVIDIA: "It is a new way of rendering video content using deep learning, Obviously NVIDIA is very focused on the generation of graphics and AI could revolutionize this field" .
The AI has generated by itself a video of a driver that runs through the city where all the landscape and environment was made by the AI. This thanks to the GPU Titan V and its 640 "tensor cores", designed to multiply its performance in the area of deep learning and artificial intelligence.
To create this demo researchers have used the open system pix2pix, a tool based on neural networks that has also been applied in another type of graphics generation.
BitLife: How to use the heirlooms - tips and tricksRE Legend: How To Fish - tips and tricks
The first thing in the whole process was to collect the data, necessary for the training of artificial intelligence, so that later the system would be in charge of translating them into the language of animated graphics
This technology is in a completely experimental phase and it would still be decades before it could really be applied in the development of a videogame. But from the company they describe it as a process similar to that experienced in ray tracing, a technique that has recently been in fashion but was first proposed in 1968 and a few years later, in 1979, perfected to give it greater realism. A long journey that could also be applied to the generation of graphics through AI.