Why Nvidia DLSS is perhaps the most exciting feature for PC gaming
With DLSS, RTX graphics cards from Nvidia offer an exclusive feature that is super exciting for gamers: it conjures up performance without the graphics suffering noticeably.
This article was created in cooperation with Nvidia and Omen.
PC games are not only becoming more and more complex in terms of play, but also technically and graphically. If you start a current AAA game on your computer, you'll hope for a smooth frame rate in addition to a high resolution and many details. If modern light and reflection technologies are then used, even expensive graphics cards quickly reach their performance limits.
The DLSS technical process developed by Nvidia, which is available for GeForce RTX graphics cards, can help. DLSS is a graphics option that can be used to significantly improve the performance of a game – without any noticeable loss of quality in the display.
What is DLSS?
The abbreviation DLSS stands for Deep Learning Super Sampling. It describes a process in which the game graphics are rendered at a lower resolution and then upscaled to the full monitor resolution.
Nvidia combines the two technologies Deep Learning and Super Sampling. The latter is a technique in which the game graphics are calculated in a higher resolution than the monitor can display. The image is then scaled back down to the resolution of the monitor. This process ensures that edges are smoothed and color transitions look cleaner than if the image were rendered natively at full screen resolution.
Nvidia uses this super sampling for DLSS by feeding an artificial intelligence on a supercomputer with an extremely high-resolution version of the game graphics. The image material used has a resolution that is up to 64 times higher than that which is later to be seen on the monitor. Based on these high-resolution images, the AI uses deep learning to learn how to recreate a version of the original image that is as detailed as possible from lower-resolution image material so that no details are lost.
With DLSS, the graphics of games can be upscaled using AI models. Image: © Nvidia 2022
The AI training data obtained in this way is then packed into the Nvidia graphics drivers and delivered to the GeForce RTX graphics cards. Using the previously developed data, they can carry out an upscaling in which almost no details are lost. The upscaling process itself is then called DLSS.
What does DLSS do in a game?
If you enable the DLSS graphics option in a compatible game, the rendering resolution of the game will be lowered. Typically, three DLSS modes are available:
- quality
- Balanced
- performance
The three modes represent different render resolutions that DLSS works with. For example, if you select the "DLSS quality" option with a monitor resolution of 2,560 x 1,440 pixels, the game will be rendered in a resolution of 1,707 x 960 pixels and then scaled up to the full display resolution using the AI-calculated models.
If you select the "DLSS Balanced" option, the render resolution is 1,485 x 834 pixels and with "DLSS Performance" it is even just 1,280 x 720 pixels.
The differences of DLSS modes at 2x zoom. Image: © Techland 2022
This is what the scene looks like in the game itself. Image: © Techland 2022
Of course, this has a slight impact on the graphics quality. While the "DLSS quality" mode often shows no difference to the native resolution with the naked eye, small losses in graphic quality can be seen in the balanced mode and slightly larger losses in the graphic quality in the performance mode. In most cases, however, the differences can only be seen if you look very closely in the detail area, as our example from "Dying Light 2" shows.
The leaps in performance that contrast with this are enormous. While "Dying Light 2" ran on our test system, an Omen 17 notebook with an Intel Core i9-11900H and GeForce RTX 3080 , in a native resolution of 2,560 x 1,440 pixels with 25 to 30 frames per second, the DLSS quality mode a jump to 40 to 50 FPS. In balanced mode, the system already achieved a smooth 60 to 70 frames per second, in DLSS performance mode the performance sometimes scratched the 100 FPS mark – with minimal losses in image quality.
HP Omen 17-ck0390ng Buy it now at
How can I use DLSS?
You can only use DLSS with Nvidia's GeForce RTX graphics cards. This applies to all models that have been on the market since August 2018, including older cards such as the RTX 2060, RTX 2070 and RTX 2080 as well as GeForce RTX graphics chips in notebooks. It is important that the graphics card always runs with the latest driver, because support for DLSS depends on the driver.
Older GeForce RTX graphics cards also offer full support for DLSS. Image: © Youtube / Nvidia 2018
In addition, the respective game itself must support DLSS. The feature doesn't just work like that, but has to be actively programmed by the developers. The list of DLSS-enabled games already includes more than 150 titles and is constantly growing. You can find a complete overview on the Nvidia website .
Does DLSS also have disadvantages?
While the technology is impressive, DLSS isn't perfect. From time to time graphic errors can occur in the display. This is very rare, but it does happen. Nvidia is working on training the AI algorithm better and better.
Are there alternatives to DLSS?
Upscaling processes are not uncommon in the gaming world. In many cases, the next-gen consoles PlayStation 5 and Xbox Series X also depend on using a form of upscaling in order to be able to display smooth gameplay on a 4K television.
In addition to DLSS, numerous other upscaling methods are available on the PC. FSR (FidelityFX Super Resolution) and NIS (Nvidia Image Scaling) are quite current methods, which, however, do not use data that has been pre-calculated by an AI.
"Dying Light 2": three upscaling methods in comparison. Image: © Techland 2022
Upscaling methods have been used for a long time. Missing pixels are artificially added to an image rendered at a lower resolution. An algorithm scans the existing image information and attempts to display the added pixels in such a way that a coherent overall picture emerges.
Depending on how big the difference between the rendered resolution and the monitor resolution is and how well the algorithm works, there may be a greater or lesser loss of quality. Objects in the game world are then blurry and details can no longer be seen clearly. This is why these methods can sometimes produce a good upscaling image – but none that is able to reconstruct image information at the level of DLSS. This can be clearly seen in our comparison image from the game "Dying Light 2" in the comparison between the native resolution and the scaled-up image.
Conclusion: DLSS is an asset to any game
With DLSS, the performance of games can be massively improved without the display quality of the graphics suffering noticeably. This allows the feature to use high graphics settings and complex effects such as ray tracing in games without the result degenerating into a stuttering orgy
The only downside is that not all games support DLSS. Here the developers are asked to make greater use of the advantages of the RTX platform for their games.