Ris vs dlss: which image rescaling technology is better?

Today we will talk about the comparison between RIS vs DLSS , two technologies related to the image of AMD and Nvidia , respectively. It is true that this second has received more attention from a large part of the public, but we must not underestimate the Radeon Image Sharpening . Although their implementations are different, what interests us is that their tasks are similar.

In case you were wondering, the main image of the article is a comparison of images of Halo 2 vs Halo 2 Remastered. The visual improvement is not due to either of the two softwares, but it seems somewhat related to us, since both technologies regenerate and improve frames.

Index of contents

Technologies of rescaling and image retouching: RIS vs DLSS

Let's start by defining where the limits of what we are talking about are, right? In the RIS vs DLSS comparison there are many things to consider, but what interests us most is the purpose of both programs.

What is clear to us is that both Radeon Image Sharpening and Deep Learning Super Sampling are rescaling and image enhancement technologies . However, each one has a different implementation.

Both technologies “reduce” the size of the frame to be rendered and then improve the image quality so that this change is not noticeable.

• The first step ensures that both the graphics and the processor can work with much less workload. After all, rendering an image at 1080p is a much lighter job than rendering it at 4K . The second step is an algorithm that 'regenerates' the image so that it doesn't look 1080p, but 4K, for example. With more or less success, both algorithms do this hard work and (or not) fool our eyes.

If the job is done well, the user enjoys higher fps on par with identical image quality. In the worst case we will see miscalculations, strange artifacts and other small bugs.

But as the some wise men say 'the devil is in the details' . Just like the wings of a bat and the wings of a bird, RIS vs DLSS are technologies whose tasks mostly converge, but whose ways of achieving it diverge. For this reason, we will talk individually about each implementation below.

AMD's Solution: Radeon Image Sharpening

The technology that AMD brings to the playing field is quite interesting. It is implemented alongside the open source tool AMD Fidelity FX , which means that any video game with this pack installed will enjoy AMD RIS .

The main section of the Radeon Image Sharpening is the adaptive contrast tuning algorithm . It has a strange name, but it comes to tell us that it retouches and improves the images closest to the camera while hardly retouching the backgrounds. The improvement is noticeable in some textures and the overall image quality is excellent.

However, this functionality can be combined with rescaling to maximize the power of our components. In some titles like Fornite we can reduce the resolution to project natively.

In our window (1920 × 1080, for example) we can have an in-game resolution of 100% (1920 × 1080) or 50% (960 × 540) . The reduction of pixels makes the work much less hard and that we can get more fps, but in exchange the image is compromised.

For this reason, mixing the visual retouching section together with a scaled down image can considerably improve the gaming experience.

Another point to note is that this technology is only available for Navi and Polaris graphics, although not in all titles. We can only activate these features in video games with Fidelity FX and APIs DirectX 9 (Navi only), DirectX 12 or Vulkan .

It is not the best there is, but the important thing is that it is oriented for the future. The next step the red team wants to take is to offer support for DirectX 11 .

Nvidia's Solution : Deep Learning Super Sampling

The solution Nvidia has come up with is somewhat different. It was announced, tested, and released some time before its competition, but that doesn't make it more dated. In fact, we would say that it is the opposite.

Deep Learning Super Sampling is a technology that uses the new system that uses Artificial Intelligence cores from Nvidia RTX graphics. The reason is quite clear: DLSS uses an algorithm based on the work of an AI that is learning. However, it is not exactly the same algorithm as that of Radeon Image Sharpening .

In the case of DLSS , a supercomputer is trained to resize images.

• At first you are given thousands of frames with and without antialiasing and asked to learn how to find the differences, then you are given a set of images at medium or low resolution to be resized at high resolution. The images are compared and if the result is similar, the algorithm is improving. However, if it has serious bugs, the researchers correct it and try to make the machine generate new rules to do it better.

This process is repeated thousands or millions of times over days or months to train the AI.

It highlights that while RIS makes changes to improve the image and rescals images in the background, here it is just the other way around. In addition, the use of Neural Networks allows this process to continually evolve, making DLSS work better and better.

Here's a video where they compare a classic image-processing algorithm against an AI- based testing algorithm:

However, it has the disadvantage that we only have this technology in Nvidia RTX graphics. By needing the RT cores, no other graphics can offer this functionality.

Furthermore, to introduce this software we cannot simply implement a tool, as in the competition. In the case of DLSS, each study must implement it "manually" in their code and for each graphics engine there are several differences. For this reason, DLSS is not so easy to implement.

RIS vs DLSS:

Therefore, the most obvious conclusion that we can offer you is that both technologies achieve similar things, but their tasks are not so similar.

The downside is that the two are limited to their brands, so it doesn't seem like we will be able to see a combination of both in the near future. Despite this, use the platform you use, you will have a good technology to lean on.

Today, the world of components is stirring and that is good for users.

• The CPUs have experienced a great launch that has destabilized the great Intel . On the other hand, AMD is going with a safe step in the field of graphics. Also, the blue team is preparing its discrete graphics, so nobody knows what will happen.

Who knows, maybe in the future we can see RIS vs. DLSS vs. Intel Technology . Or perhaps we can see a combination of the two or three technologies because the competition takes another tint.

Be that as it may, here we have shown you the majority of differences between these two incredible technologies. We hope that you have understood it easily and that you have learned something new. Furthermore, we encourage you to read and search for information on these topics, since these new technologies are based on very interesting ideas.

And you, do you think that Intel will establish itself as the third competition in integrated graphics? Which technology do you think is better RIS vs DLSS ? Share your ideas in the comment box.

AMD RISNvidia DLSS SourceNvidia DLSS FAQ