sc94597 said:
What you don't seem to be getting is that using DLSS is not something that is done because it is "necessary to reach 1080p" but because it is superior to running at "native" 1080p with some inferior temporal method like basic TAA. In modern games, there really isn't such a thing as not using some form of temporal AA as the games are designed for it, so "native 1080p" rarely is "native" across the board of the rendering pipeline. Often you have noisy assets or effects at much lower resolutions than would be adequate without the temporal AA passes, and the TAA is required to make them look coherent without dithering effects and other noise artifacts. It also makes the game look very blurry. So you typically have two options: 1. run the game at your native target, get worse image quality because basic TAA is crappy, but you need a temporal solution of some kind. 2. run the game with DLSS and get better image quality. With the cost that you have to lower internal resolution because DLSS consumes resources. It is pretty obvious which one developers are going to choose. The one that gives the better result. Also this is the same reason why Series S uses FSR 2 in many games, often with a lower internal resolution than PS4 Pro or One X. Even FSR 2 is better than basic TAA. Notice that many of the examples you provided are 7th Generation and early 8th Generation titles before TAA became the standard. Also notice that Switch 1 games (which aren't using DLSS most of the time) and early Nintendo Switch 2 games tend to run above 1080p on Switch 2, and Switch 1 games fall in between the 7th and 8th Generation in render techniques. |
Native 1080P resolution with good TAA should get you better results then DlSS upscaled from low resolutions, it really depends on the TAA.
Last edited by redkong - on 14 August 2025
