Nvidia admits that DLSS needs to be improved



When Nvidia first announced the RTX 20 series, raytracing was only one of the highly sought features. The other was of course DLSS, which uses autoprocessor cores on the graphics processor to analyze the images in play and provide better anti-aliasing compared to other methods. Unfortunately, the performance of DLSS has yielded questionable results in some cases, which Nvidia hopes to solve soon.

For the moment, DLSS support needs to be added on a case-by-case basis, with Nvidia's algorithm analyzing each title at a specific resolution to produce the best possible results. The disadvantage is that DLSS works better with a 4K resolution, because Tensor cores have more pixels to analyze, which gives a better result.

Battlefield V has recently received its DLSS update, which claims to bring performance improvements of up to 40% to 4K and 1440p. Unfortunately, those who play 1080p end up with an image that may look worse than you would get with TAA.

In an updated DLSS FAQ, Nvidia recognized the problems with DLSS at lower resolutions and its lack of support for popular ultra-wide resolutions. According to Nvidia, the company's next goal will be "training to improve image quality at 1920 × 1080 as well as for ultra-wide screens (eg 3 440 × 1440)". Nvidia also admits that performance with DLSS at some resolutions is not what it should be.

Metro Exodus is another game configured to support DLSS. An update will be available soon to improve sharpness and image quality in all resolutions.

KitGuru says: Although Ray Tracing is nice, I was more excited for DLSS after Nvidia's RTX revelation. Taking advantage of machine learning to enhance anti-aliasing seems to be a very smart way of incorporating AI-based features into the game space. Unfortunately, DLSS did not make the best start, many wondering about the quality of the results. Hopefully, over time, these problems will be solved and we will begin to see expanded support delivered more quickly.


Source link