Nvidia introduces Deep Learning Dynamic Super Resolution technology
Previously, AMD announced the launch of a technology called Radeon Super Resolution (RSR), which evolved on the basis of the current FidelityFX Super Resolution (FSR). However, compared to FSR, which requires game developers to manually add to the game, RSR is built into the driver level, thus bypassing the need to deploy in a specific location in the graphics pipeline. As a full-screen upgrade technology, players can use this technology after they have installed the RSR-containing driver and theoretically open the full-screen exclusive option in any game. AMD said the technology will appear with new drivers in the first quarter of 2022.
Today, NVIDIA announced the launch of Deep Learning Dynamic Super Resolution (DLDSR) technology. It’s an AI-powered Dynamic Super Resolution (DSR) technology that improves upon it by adding an AI layer powered by Tensor Cores, which also means the feature is exclusive to Nvidia GeForce RTX GPUs. The emergence of RSR technology and DLDSR technology has made the competition between AMD and NVIDIA in screen scaling technology even hotter.
Unlike DLSS, FSR, RSR, or future Intel XeSS technologies that render at low resolution and then increase to the original screen resolution for output, DLDSR technology renders at high resolution and then compresses to the original screen resolution for the output. This improves image quality with more detail and reduces jagged edges on objects, making the image look smoother and less glare. With the newly added AI layer, fewer pixels need to be input when rendering. It is said that the image quality of DLDSR 2.25X is comparable to DSR 4X, and the performance is higher.
Nvidia says DLDSR technology is available in most games on GeForce RTX GPUs and will be integrated into the GeForce Game Ready driver update on January 14.
