If we look back and evaluate the last five years of innovation in the graphics sector, we realize that NVIDIA DLSS was, without a doubt, one of the most important advancesand has remained today one of the most important technologies with the greatest impact thanks to the numerous evolutions that the green giant is introducing.

This development left us with three main aspects of this hardware-accelerated artificial intelligence applied to games with NVIDIA DLSS technology at the moment. Super resolutionwhich is a method of image reconstruction and rescaling, the frame generationwhich uses AI and algorithms to create a completely independent interframe for every two traditionally rendered frames and flash reconstructionwhich acts as an advanced noise reducer capable of improving graphics quality and performance thanks to AI.

We can be sure that over time, NVIDIA DLSS technology will continue to change and expand its scope with the introduction of new technologies that will improve the performance, graphic quality and realism of games. Some of these technologies will be exclusive to new generations of graphics cards arriving in the coming years, while others may retain full compatibility with previous generations.

Along these lines, Bryan Catanzaro, VP of Applied Deep Learning Research at NVIDIA, commented on something that I honestly find very interesting, and that in his opinion, a very advanced version of DLSS could end. taking care of the whole rendering problem under the neural artificial intelligence system. Yes, that means everything related to game rendering would be done by DLSS technology.

The executive director of NVIDIA referred to a technical demo they showed in 2018, when they managed to represent the entire world using a neural network. In this demo, the game’s graphics engine was used to generate the information necessary from all the things and elements of the game, that is, as input information, and the neural network used all this information to solve the entire rendering process. The truth is that the result they obtained was very good for the time, as you can see in the attached video, and this is exactly the direction Catanzaro is pointing.

To make this leap, advanced tools will be needed, as it won’t be as simple as writing a paragraph for the AI ​​to create a game on its own at the level of Cyberpunk 2077. In his opinion, with an advanced version of NVIDIA DLSS (he cites version 10 as an example), they could we would achieve an advanced fully neural rendering system that works with the game’s graphics engine and that allows us create more compelling, complex and rich games.

At the time of this writing, I can confirm that NVIDIA is working on new advancements of great importance, such as radial caching and AI-powered texture compressiontwo keys that only confirm the growing importance that AI has in the world of video games.