Death Stranding came to PC yesterday, and as our regular readers will know, it comes with a major face lift at the graphic level and it also has technology support NVIDIA DLSS 2.0, a system of smart rescaling which reduces the plot load the GPU must face by decreasing the actual pixel count in each scene.
We will understand it better with an example. Imagine you start Death Stranding and set a resolution of 1,920 x 1,080 pixels. Well, NVIDIA’s DLSS 2.0 technology would start at half that native resolution, and would rescale combining different images in real time to create, thanks to them, the best possible image.
The result in the first implementations of this technology was not a good thing, but With DLSS 2.0, things have changed, and much. I had the opportunity to test the first integration of the DLSS system in games like Battlefield V and Metro Exodus, and I must admit that I did not like the end result. However, the arrival of DLSS 2.0 in Wolfenstein: Youngblood marked a huge turning point, since, as we told you at the time, this technology finally began to do its job.
NVIDIA DLSS 2.0: a deeper look
As we have told you in the previous paragraphs, DLSS technology was born as an intelligent rescaling technique that is supported by an artificial intelligence system. This system has been trained to develop a series of algorithms that allow create a high-quality image by combining multiple images.
What this technology does is therefore run a real-time image rebuild process resorting to a very particular rescaling. It has nothing to do with other known rescaling techniques, as NVIDIA DLSS is not limited to starting with a given pixel base and “stretching” it to fill in the blanks, then applying temporary edge smoothing to “fix” the result and make up the saw teeth. For example, PS4 Pro uses the “checkerboard rendering” technique, which renders only 50% of the pixels needed to create an image, and extrapolates the rest from the base pixels, applying a temporary reconstruction filter to improve the final quality. .
The objective of DLSS technology is very different, since it seeks create a perfect image starting from several images and selecting the best possible combination of those, all in real time. The chosen image will be rendered starting from a lower resolution, and rescaled until reaching the objective resolution, which is none other than the one we have selected in the configuration of each game.
NVIDIA promised that DLSS rescaling would be able to offer high image quality and greatly improve performance in games, a promise that the first generation of this technology could fulfill, but that has been realized with the arrival of a new generation, known as DLSS 2.0.
In order to activate DLSS technology we must have a graphics card based on the Turing architecture NVIDIA, but beware, only the models of the RTX 20 series because said technology runs through the tensor cores, specialized in carrying out workloads focused on artificial intelligence. This hardware acceleration through the tensor cores is essential, so if we have a GTX 16 or less (GTX 10, GTX 900 …) we will not be able to activate it. On the other hand, if you have a GeForce RXT 2060 or higher you can activate it without problem.
When we activate DLSS 2.0 technology we can choose between different configurations that prioritize performance or quality. The only difference between these modes is the base resolution from which the rendering process will be performed. Thus, in “performance” mode, 50% part of pixel count of desired resolution, and in «quality» mode the total count rises to 67%. So that we understand each other, if we configure a game to 1080p and activate DLSS 2.0 in quality mode, the base resolution will be 540p.
I know what you are thinking, that it doesn’t sound good, and that the pixel clipping seems too big for the final image quality to be good. As incredible as it may seem, and Death Stranding is an excellent example of the value that NVIDIA DLS 2.0 technology offers. Let’s see why.
Death Stranding with DLSS 2.0: Higher Image Quality and Performance
It’s that simple, when we activate DLSS 2.0 rescaling technology in Death Stranding we not only noticeably improve performance, we also achieve a clear improvement in image quality.
The performance increase varies depending on the graphics card that we use and the resolution. When we play in 1080p there is a less noticeable performance improvement because it is not a demanding resolution. Obviously also influenced by the fact that Death Stranding is quite well optimized. However, when we raise the resolution to 1440p we see a significant increase in performance, and when we reach 2160p (4K) the jump is so great that DLSS 2.0 technology virtually doubles performance.
Then I leave the testing equipment that I have used to analyze DLSS 2.0 technology in Death Stranding:
- Ryzen 7 1800X processor with eight cores and sixteen threads at 4 GHz.
- 32 GB (4 x 8 GB) of Corsair Dominator Platinum RGB DDR4 3,200 MHz RAM (CL16).
- GIGABYTE AORUS GA-AX370-GAMING 5 motherboard with BIOS F25.
- Corsair Hydro Series H100i RGB Platinum cooling system.
- GIGABYTE RTX 2080 Super OC graphics card (1,975 MHz GPU in turbo mode) and 8 GB of 15.5 GHz GDDR6.
- Samsung Evo 850 500 GB SSD (operating system).
- 960 GB Corsair Force Series MP510 NVMe PCIE SSD, where the game has been installed.
- 2TB Seagate SHDD with 8GB SSD as cache.
- Windows 10 Pro 64-bit.
- Corsair AX1000 80 Plus Titanium power supply with 80 Plus Titanium certification.
The performance improvement is clear, and forceful. Thanks to DLSS 2.0 technology we can play Death Stranding in maximum quality and 4K resolution keeping 60 FPS stable with an RTX 2060, and achieve 90 FPS stockings with a RTX 2080 Super keeping those settings, but what about image quality? I want to do a blind test with you, I am going to leave you a series of images and I want you to try to guess in which of them the DLSS 2.0 technology was activated and in which not. You can expand them by clicking on them.
Let’s see how many you have guessed right. All images named as Death Stranding 1 have DLSS 2.0 disabled, while images identified as Death Stranding 2 has it activated.
If you look closely, activating DLSS 2.0 technology in Death Stranding greatly reduces saw teeth (aliasing), which is especially evident over long distances. The sharpness of some surfaces and numerous details are also improved. The conclusion is very clear, NVIDIA has done an excellent job with DLSS 2.0 technology, and has achieved an important advance that, in my opinion, represents a more interesting value even than ray tracing, at least in its current state.