Its actually the opposite, the gap between DLSS and FSR is bigger at lower resolutions than it is at 4k+. Thats because DLSS can still rely on ai model data that a lower resolution image cannot supply vs a higher base resolution where that data is likely more available. Its why when you have side by side tests, its less noticable as you get to 4k. Definitely one of the tests HUB does when conparing the scaling algorithms side by side.