With Intel’s announcement of ExtraSS technology that will be integrated into XeSSrepresents the definitive confirmation of the success that at the time represented NVIDIA’s commitment to image generation technologies through artificial intelligence in DLSS 3. Recall that NVIDIA presented this DLSS evolution together with the GeForce RTX 40 generation last September and that one of its main innovations was just this, to combat the CPU bottleneck problem by making the graphics adapter able to produce “average” frames thanks to AI.

The first reaction in this regard came a few weeks later, in November of last year, when AMD announced FSR 3 and AMD HYPER-X took the opportunity to tell us about Fluid Motion Frames technology for the first time. They didn’t make it clear then. but everything pointed (as it was subsequently confirmed) to proprietary AI image generation technology. Although the announcement was over a year ago, it wasn’t until last August that we finally had confirmation of when it would start rolling out to users.

Obviously, NVIDIA knew how to get ahead and lead the way in this way, with the type of technology solution at least as important as AI-based rescaling, a path that opened up with DLSS and that AMD tackled with FSR and Intel with XeSS. And again, the same thing happened with the generation of AI-based frames, and some “ultras” indulged in mocking and underestimating its huge potential, until even its reference manufacturer decided to bet on this technology.

Although its role in the sector of dedicated graphics adapters is still quite limited, we cannot say the same if we talk about integrated ones, present in a large part of APUs that leave their production nodes. That is, the announcement ExtraSS, Intel’s AI frame generation technology that will be integrated into XeSSas the technology company announced during SIGGRAPH Asia 2023, which is taking place these days in Sydney, Australia.

«Introducing ExtraSS, a new framework that combines spatial supersampling and frame extrapolation to improve real-time rendering performance. By integrating these techniques, our approach achieves a balance between performance and qualitygenerating temporally stable, high-quality, high-resolution results.«We can read in the statement issued in this regard by Intel, in which we also find examples of its operation and effects.