Users of modern video cards Radeon The effectiveness of AMD’s FSR 4 intelligent scaling technology on the RDNA 4 graphics architecture is being actively discussed. The discussion stemmed from complaints from RX 9060 XT graphics card owners about insufficient image sharpness and texture blurring in fast-paced scenes. Some players are declaring their willingness to switch platforms in favor of competing solutions due to the more stable performance of anti-aliasing and ray-tracing algorithms in similar gaming scenarios. At the same time, experienced community members note that the overall perception of the final image depends heavily on the monitor’s native resolution, and in 4K mode, upscaling artifacts become virtually unnoticeable.
During a detailed review of the updated AMD FSR 4.1, opinions within the gaming community were divided. Some users point to a noticeable improvement in rendering fine details and distant objects, bringing image quality closer to the level of current versions of NVIDIA DLSS technology. Other gamers criticize the update for introducing excessive sharpness on object edges and some image instability during sharp virtual camera movements. Independent benchmarks conducted by specialized technical channels confirm that in static scenes, the algorithms from both manufacturers produce virtually identical results, with the main differences appearing in complex dynamic conditions.
A key argument in the discussion was the hardware specificity of processing scaling algorithms on new graphics chips. On the RDNA 4 architecture, thanks to the use of dedicated matrix units, the upscaler pass execution time is only 0.6 to 1.2 ms in separate 1080p and 1440p screen sizes. Players emphasize that visual noise in some graphically intensive games is often caused by the lack of full integration of ray-tracing systems by game developers, rather than by defects in the upscaler itself. Further development of the software ecosystem should minimize blurring artifacts without impacting overall graphics card performance.