NVIDIA DLSS 3.0 technology has generated great expectations, and it is perfectly understandable, since after the arrival of DLSS 2.0 we were able to confirm that the smart rescaling the green giant is finally able to do a good job.
A few days ago I had the opportunity to thoroughly test DLSS 2.0 technology in Death Stranding, and as we saw, it is fantastic. The level of sharpness it achieves is very goodIn fact, it reduces the problems of “aliasing” (saw teeth) in the medium and long distance, and achieves a fairly clean and sharp image. Gone are the problems that this technology gave in its first implementation in games like Battlefield V and Metro Exodus.
As you know, those of you who regularly read DLSS 2.0 technology render starting from a lower resolution (50% to 67% of total pixels), combine multiple images to create the highest quality image possible, and rescale until you reach the target resolution. This reduces the actual number of pixels, which contributes to improve performance even if we are using ray tracing, since having fewer pixels also reduces the work that RT cores must do.
Yes, it is very interesting, and as I said at the time it seems to me one of the most important values of the GeForce RTX 20 series. However, this technology presents a major problem, and that is that must be supported by developers in each specific game, that is, it does not have a “universal” support, as is the case, for example, with other technologies such as TAA, a reality that could completely change with the arrival of NVIDIA DLSS 3.0 technology.
What NVIDIA DLSS 3.0 Promises and Why It Matters
Its base will not change, that is, it will continue to be an intelligent rescaling that will allow games to be moved at a relatively low resolution but maintaining a quality practically identical to that which we would obtain with a higher resolution. For example, a game configured in 1080p and with DLSS activated in performance mode would have 540p as actual resolution, and from there it would rescale to 1080p. The result, both in terms of sharpness and image quality, would be practically the same as we would get with native 1080p.
The main improvement that the green giant plans to introduce with the launch of NVIDIA DLSS 3.0 is the support in any game with TAA. In effect, this means that such technology could work without problem in any title that integrates TAA, which would greatly contribute to its takeoff, although developers will still have to make a little effort to implement it, and it will be essential to install new Game Ready controllers with the arrival of each new NVIDIA DLSS 3.0 compatible game.
You can also expect an improvement in image quality with that new generation of smart rescaling. What we don’t know is if NVIDIA DLSS 3.0 will be exclusive to the new 30 series RTX Or will it work on the RTX 20 series as well? Both possibilities are open, but it is clear that limiting it to the next generation will not sit well with those who trusted the Turing architecture and bought a GeForce RTX 20 series.
In the accompanying image we can also see two other important innovations that will come with the NVIDIA 30 series RTX, NVCache and memory compression through tensor cores. The first is a technology that aims to balance and optimize the use of RAM, VRAM and SSD storage to reduce loading times in games, and the second promises to intelligently compress certain elements, such as textures, to reduce the consumption of graphic memory between 20% and 40%.
I remind you that the presentation of the new GeForce RTX 30 graphics cards will take place between September and October of this year, and that those in green are likely to take advantage to announce NVIDIA DLSS 3.0 as well.