The image-quality battle brewing between AMD and Nvidia is heating up. And this week, the back-and-forth regarding Nvidia’s deep-learning super sampling (DLSS) and AMD’s FidelityFX Super Resolution (FSR) is taking on a human-versus-machine angle. To be more specific, AMD believes the humans responsible for FSR are sometimes more effective than the machine learning that Nvidia uses for DLSS.
AMD introduced FSR 2.0 last week, and it emphasized that it isn’t using any machine learning to accomplish its impressive upscaling results in games like Deathloop. Now, AMD is explaining a bit more about why it is using hand-coded algorithms as opposed to DLSS, which uses a method where a computer trains itself to recognize what a perfect frame should look like at the highest possible resolution.
“In any science, which includes software engineering, discoveries are made by analyzing data from experimentation, resulting in mathematical models that can explain the results,” reads an AMD blog post. “Broadly speaking, machine learning (ML) is an incredibly useful set of tools and techniques that can aid and accelerate this process. However, the results that ML achieves can sometimes not be the most optimal, lacking the spark of human imagination that can often lead to breakthroughs for complex problems.”
Basically, AMD thinks humans are already good at solving the problems that upscaling introduces. And it claims that it has more flexibility with prescribing solutions to certain scenarios than it would if it adopted the DLSS model.
GamesBeat Next 2023
Join the GamesBeat community in San Francisco this October 24-25. You’ll hear from the brightest minds within the gaming industry on latest developments and their take on the future of gaming.
“The FidelityFX Super Resolution 2.0 analytical approach can provide significant advantages compared to ML solutions, such as more control to cater to a range of different scenarios, and a better ability to optimize,” reads the blog.
DLSS is still the image-quality king — although that may not matter if FSR is on more hardware
Of course, AMD is slightly mischaracterizing how machine learning works. While Nvidia is certainly training computer models to recognize what games should look like, it is humans who are behind those models. It is also extremely difficult for AMD to argue with the results. DLSS upscaling is frequently indistinguishable from native 4K. Occasionally, it is even better than the real thing. And that is because it has access to models that provide more information than the original frame contains.
FSR 2.0 is certainly closing the gap. And I do believe that it is a fine solution for the problem of running high-fidelity games at acceptable framerates. But DLSS is the cutting-edge market leader for a reason.
Although, AMD doesn’t seem worried. Because while DLSS requires a modern Nvidia GPU, FSR works on almost any recent computer or console.
“Above all, not requiring dedicated ML hardware means that more platforms can benefit, and more gamers will be able to experience FSR 2.0,” claims AMD.
To combat this, Nvidia has launched a similar tech to FSR called Nvidia Image Scaling (NIS) that also works across devices and operates on the platform level. But it’s DLSS that Nvidia cares about — and it should. It’s a standout feature for Nvidia hardware. And the company needs developers to keep supporting it.
GamesBeat's creed when covering the game industry is "where passion meets business." What does this mean? We want to tell you how the news matters to you -- not just as a decision-maker at a game studio, but also as a fan of games. Whether you read our articles, listen to our podcasts, or watch our videos, GamesBeat will help you learn about the industry and enjoy engaging with it. Discover our Briefings.