Nvidia Image Scaling “resurrects” the old GTXs

Nvidia DLSS is a technology that has allowed the two generations of RTX video cards to achieve previously unthinkable results: the improvements to the image and to the framerate that the AI ​​algorithm brings allow you to play at higher resolutions even with GPUs not designed for that purpose, giving new life to the less powerful models of the last generation, such as the RTX 2060 or the RTX 2070. However, DLSS is only available for the last two generations of video cards and only for those of the RTX range: Nvidia has therefore decided to develop a new solution for all its GPUs (from the GTX 900 series onwards), which also works on non-RTX models.

The technology is called Nvidia Image Scaler (NIS) and is in effect a spatial upscaler , which will be available for all games and will also integrate a sharpening algorithm. NIS is easily accessible from Nvidia's control panel or GeForce Experience, includes several settings that can be activated via Nvidia's in-game overlay, and, according to the company, offers image quality comparable to that of the FSR.

Below you can see an image created by Nvidia, where different images of a detail of Necromunda are shown: Hired Gun. FSR and NIS have a very similar final rendering of the image, also allowing to reach practically the same framerate. In both cases the output is slightly worse than native 4K, but the use of these spatial upscalers offers a 40% performance improvement. DLSS, on the other hand, plays a game of its own: even in performance mode, one of those that most penalizes image quality, the final result is better than the native one and the FPS are more than doubled.

Nvidia has made Image Scaling technology available as an open source SDK on GitHub, allowing all developers to integrate it into their games. But what is the main difference between a spatial upscaling technology like NIS and DLSS? The first uses only the current frame to reconstruct the image, while the second uses the previous, current and next frame. The use of three frames allows the DLSS to maintain an image quality equal to or even higher than that which occurs at native resolution, while the technique used by NIS (and all spatial upscalers) inevitably leads to a loss of quality. , in favor of performance.

To help users compare the outputs of different technologies such as FSR, NIS and DLSS, Nvidia today also introduces ICAT, a tool designed specifically for this task. The software allows you to insert up to four screenshots or videos and compare them side by side or with a slider, which allows you to switch from one image to another. The input videos can be synchronized and zooms occur simultaneously on all videos, thus facilitating the comparison process. ICAT is available for download on Nvidia's website and the company suggests using an external capture card to capture video to feed to the software.