In 2018, NVIDIA introduced a breathtaking feature called DLSS, which stands for Deep Learning Super Sampling. It’s a fancy way to make video games look nicer. As time has passed, DLSS has improved with each new version, and NVIDIA thinks that in the future, DLSS 10 could make graphics look great by using artificial intelligence.
Not long ago, there was a talk about AI in video games arranged by Digital Foundry. They talked to Bryan Catanzaro from NVIDIA, who’s a big deal and knows a bunch about deep learning. They asked him what he thinks will happen with DLSS in the future and how machines that learn can make things even cooler.
DLSS 10 can render the game environment entirely by a neural network
In 2018, during the NeurIPS conference, Bryan Catanzaro and his team at NVIDIA put together an impressive demonstration. They showed a virtual world rendered entirely by a neural network but with a unique twist: a game engine drove the whole process.
For simplicity, the game engine provided information about the positions of objects in the virtual world. This data was then fed into a neural network which took responsibility for rendering the entire scene. The neural network handled every aspect of the graphics rendering process. Achieving real-time performance with this innovative approach in 2018 was a visionary achievement for Catanzaro and his team.
Catanzaro mentioned that the quality of the rendered game’s visuals at that time didn’t compare to something like Cyberpunk 2077. However, he believes that, in the long run, this is the direction the graphics industry will move towards.
NVIDIA’s DLSS technology has come a long way since its introduction with the RTX 20-series GPUs. Initially, people were unsure about the benefits of including tensor cores in gaming GPUs. The first ray-tracing games and the first version of DLSS didn’t impress everyone. But DLSS 2.X made significant improvements to the technology, making it more valuable and practical. As a result, it became more popular and even inspired similar technologies such as FSR2 and XeSS, which were later introduced by other companies.
Generative AI will become increasingly prominent in the graphics field. The rationale behind this shift is consistent with the broader application of AI in various fields. It’s about leveraging the ability of AI to learn complex functions by analyzing vast datasets, rather than trying to construct algorithms from scratch
In about ten years, DLSS might be able to replace the way games are usually made to look good. NVIDIA is also working on new tricks like radial caching and texture compression that could be added to DLSS to do even more of the work. But to make this happen, they might need to put more special cores in their GPUs.
Can DLSS 10 replace GPUs’ significance in the Gaming Industry?
DLSS, or Deep Learning Super Sampling, acts as a helper for your computer’s graphics card during gaming, employing clever techniques to enhance game visuals without putting excessive strain on your computer.
In the future, DLSS may become even smarter, enabling it to enhance game visuals without the need for an ultra-powerful graphics card. Nevertheless, it won’t completely supplant the graphics card because the graphics card remains the primary driver of superior game visuals. DLSS functions as a supportive companion, easing the workload of the graphics card rather than replacing it.
Regarding the future of GPUs in light of this advanced technology, DLSS 10 could have the potential to greatly enhance gaming graphics while still relying on GPUs. It would collaborate with GPUs to optimize their performance, potentially achieving fantastic visuals with reduced power consumption. However, GPUs will continue to be indispensable for the heavy-duty tasks of rendering images and running games. Thus, while DLSS 10 can enhance GPU performance, it won’t replace GPUs in the future of gaming.