The modders are doing an excellent job with Starfield and it’s thanks to them that we can enjoy the technology Intel XeSS, NVIDIA DLSS Super Resolution and NVIDIA DLSS 3, which adds frame generation if we have a compatible graphics card (GeForce RTX 40).

The truth is, at this point, I don’t think there’s anything more to say. It is evident that Bethesda did not include these technologies in Starfield since AMD’s sponsorship prevented it, there is no other explanation that makes the slightest sense, especially seeing how easy it was for modders to implement them.

It is important to note that there was already a mod that added DLSS 3 support and frame generation in Starfield, but it was developed by DarkSoul and was only accessible by paying. This new new mod was developed by LukeFZ and is completely free. As we can read in the news feed, the mod replaces FSR 2 and works so well that It is capable of converting an average speed of 56 FPS with a latency of 16 ms to 110 FPS with an average latency of 8 ms.

However, it is a mod that is still in development and therefore It represents certain failures (“bugs”) that will be resolved over time. Looking at the list of confirmed bugs, we can clearly say that this mod has not reached the state of absolute maturity and that it has room for improvement, but what can I say, it is free and fulfills its goal of adding frame generation to the game.

The frame generation is a huge improvement over Starfield as it is a very poorly optimized game that has everything heavy reliance on single-threaded CPU performance and does not scale very well when using reconstruction and scaling technologies. In these cases, the performance improvement we can achieve is quite discrete, even when using performance mode.

With frame generation, a noticeable increase in frames per second is achieved, and the impact this technology has on the game’s fluidity is very large. As our regular readers already know, this is because this technology does not depend on the CPU, because it creates fully independent frames using intermediate frame information, artificial intelligence, motion vectors and hardware acceleration.