clock menu more-arrow no yes mobile

Filed under:

Nvidia’s performance-boosting DLSS tech is now easier to add to some games

Unreal Engine 4 plugin for DLSS could be a big deal

If you buy something from a Polygon link, Vox Media may earn a commission. See our ethics statement.

Three soldiers carry bags of cash in a screenshot from Call of Duty: Warzone
Nvidia announced in January that DLSS support is on the way to Call of Duty: Warzone.
Image: Infinity Ward/Activision
Samit Sarkar (he/him) is Polygon’s deputy managing editor. He has more than 15 years of experience covering video games, movies, television, and technology.

Nvidia and Epic Games have taken what could prove to be an important step in the long road to bringing ray-traced video game graphics to the masses.

Last week, Nvidia released a plugin for Epic’s Unreal Engine 4 that will enable developers to implement the chipmaker’s Deep Learning Super Sampling (DLSS) technology in their games much more easily than before. Until now, developers have had to work with Nvidia to add DLSS to their games. That’s one reason why the technology has been limited in its use — it’s available in only 25 or so games at the moment, although the makers of more than a dozen others have announced plans to add DLSS support.

“UE4 integration is a critical step to making a technology pervasive in games,” said Henry Lin, senior product manager at Nvidia, in an interview with Polygon. “We’ve worked closely with developers over the last year to bring DLSS technology to over 20 UE4-based games and creative apps via our DLSS UE4 Github integration. A UE4 marketplace plugin now accessible to all developers makes DLSS even simpler to integrate and will mean more DLSS games — from indies to AAA titles — for users to enjoy.”

Now that the plugin is available on the Unreal Engine Marketplace, developers making projects in Unreal Engine 4 can “start using DLSS right away,” said Lin. Adding DLSS support to a game will probably take a bit more work than simply flipping a switch, but Lin said that Nvidia has “been able to fine tune the integration to work well out of the box for the vast majority of content.”

In a statement to Polygon, an Epic Games spokesperson pointed out that the use cases for DLSS aren’t limited to video games, saying, “Professional workflows including virtual production, architectural visualization, product design, and animation will all benefit from the increased speed and fidelity.”

For the uninitiated, DLSS uses AI to accelerate graphics rendering through image reconstruction, maintaining high image quality while delivering improved frame rates. Put simply, the technology allows a video card to render a game at a lower internal resolution — reducing the load on the GPU — as it uses Nvidia’s AI algorithm to generate a reconstructed image that looks as good as (or, in some cases, better than) the game rendered at native resolution.

In games like Death Stranding and Fortnite, turning on DLSS can double or even triple performance, making it possible to play at frame rates above 60 frames per second in 4K resolution when that would be otherwise impossible. Using the performance mode of DLSS, a GPU would only have to render the game at 1080p — outputting a 4K image while rendering just one-fourth of the pixels on screen. This also allows players to turn up settings without dropping the frame rate too much.

DLSS is a proprietary Nvidia technology, and it relies on the company’s RTX line of graphics cards, since they contain the dedicated hardware — “tensor cores” on the silicon — that can handle the calculations for Nvidia’s AI algorithm. DLSS is essentially a must-have feature for games that support real-time ray tracing, since that rendering technique is very computationally intensive. Without DLSS, it would be difficult to run games with ray tracing at playable frame rates (especially on weaker GPUs like some of the first-generation RTX cards).

The initial version of DLSS, which launched in early 2019, required that the AI algorithm be trained separately with each game in order for it to work. Nvidia released DLSS 2.0 in the spring of 2020, and it offered a crucial upgrade: The AI was now being trained with generic images, making it faster and easier for developers to implement DLSS in their games. It’s likely that the new Unreal Engine plugin wouldn’t have been possible before DLSS 2.0.

Nvidia’s chief rival, AMD, is working on its own alternative to DLSS called FidelityFX Super Resolution. The technology wasn’t ready in time for the company’s RX 6000 line of GPUs, which launched last fall, and AMD hasn’t yet given a release window for it. But an important advantage that FidelityFX Super Resolution will have over DLSS is that AMD has promised to make it open rather than proprietary — the FreeSync to Nvidia’s G-Sync — and cross-platform. That means that the technology could also come to PlayStation 5 and Xbox Series X; after all, both consoles are powered by AMD Radeon RDNA 2 graphics, just like the RX 6000 GPUs.

Update (Feb. 19): We’ve added a comment from Epic Games.

The next level of puzzles.

Take a break from your day by playing a puzzle or two! We’ve got SpellTower, Typeshift, crosswords, and more.