NVIDIA DLSS 2.0: A Big Leap In AI Rendering

By Andrew Burnes on March 23, 2020 | Featured Stories GeForce RTX GPUs NVIDIA DLSS NVIDIA RTX

Artificial intelligence is revolutionizing gaming - from in-game physics and animation simulation to real-time rendering and AI-assisted broadcasting features. With Deep Learning Super Sampling (DLSS), NVIDIA set out to redefine real-time rendering through AI-based super resolution - rendering fewer pixels and then using AI to construct sharp, higher resolution images. With our latest 2.0 version of DLSS, we’ve made big advances towards this vision.

Powered by dedicated AI processors on GeForce RTX GPUs called Tensor Cores, DLSS 2.0 is a new and improved deep learning neural network that boosts frame rates while generating beautiful, crisp game images. It gives gamers the performance headroom to maximize ray tracing settings and increase output resolutions.

 

DLSS 2.0 offers several key enhancements over the original version:

  • Superior Image Quality - DLSS 2.0 offers image quality comparable to native resolution while rendering only one quarter to one half of the pixels. It employs new temporal feedback techniques for sharper image details and improved stability from frame to frame.

  • Great Scaling Across All GeForce RTX GPUs and Resolutions - A new AI network more efficiently uses Tensor Cores to execute 2X faster than the original. This improves frame rates and eliminates previous limitations on which GPUs, settings, and resolutions could be enabled.

  • One Network For All Games - The original DLSS required training the AI network for each new game. DLSS 2.0 trains using non-game-specific content, delivering a generalized network that works across games. This means faster game integrations, and ultimately more DLSS games.

  • Customizable Options - DLSS 2.0 offers users 3 image quality modes - Quality, Balanced, Performance - that control the game’s internal rendering resolution, with Performance mode enabling up to 4X super resolution (i.e. 1080p → 4K). This means more user choice, and even bigger performance boosts.

Deliver Us the Moon - DLSS 2.0 substantially boosts performance while maintaining --and in some cases improving-- image quality.

NVIDIA DLSS 2.0 is already available in Deliver Us The Moon and Wolfenstein: Youngblood; launches in MechWarrior 5: Mercenaries today; and comes to Control via a game patch on March 26th. Also, DLSS 2.0 is now available to Unreal Engine 4 developers through the DLSS Developer Program that will accelerate deployment in one of the world’s most popular game engines. If you’re a developer, head here to learn more.

DLSS 2.0 - Under the Hood

Using our Neural Graphics Framework, NGX, the DLSS deep neural network is trained on a NVIDIA DGX-powered supercomputer.

DLSS 2.0 has two primary inputs into the AI network:

  1. Low resolution, aliased images rendered by the game engine
  2. Low resolution, motion vectors from the same images -- also generated by the game engine

Motion vectors tell us which direction objects in the scene are moving from frame to frame. We can apply these vectors to the previous high resolution output to estimate what the next frame will look like. We refer to this process as ‘temporal feedback,’ as it uses history to inform the future.

The NVIDIA DLSS 2.0 Architecture

A special type of AI network, called a convolutional autoencoder, takes the low resolution current frame, and the high resolution previous frame, to determine on a pixel-by-pixel basis how to generate a higher quality current frame.

During the training process, the output image is compared to an offline rendered, ultra-high quality 16K reference image, and the difference is communicated back into the network so that it can continue to learn and improve its results. This process is repeated tens of thousands of times on the supercomputer until the network reliably outputs high quality, high resolution images.

Once the network is trained, NGX delivers the AI model to your GeForce RTX PC or laptop via Game Ready Drivers and OTA updates. With Turing’s Tensor Cores delivering up to 110 teraflops of dedicated AI horsepower, the DLSS network can be run in real-time simultaneously with an intensive 3D game. This simply wasn’t possible before Turing and Tensor Cores.

New DLSS Announcements And Integrations

Since the publication of this article, we've been busy working on further improvements to DLSS, new DLSS features, and new integrations of DLSS into popular games:

DLSS 2.0 In Action

Let’s take a look at how DLSS 2.0 improves performance and image quality in each of the first four games to implement the technology: Control, Deliver Us The Moon, MechWarrior 5: Mercenaries, and Wolfenstein: Youngblood.

Control

Remedy Entertainment’s Control was one of 2019’s best single-player games, and also one of its most advanced, with a suite of ray-traced effects that took its graphical fidelity to new heights.  

Control originally launched with a prior version of DLSS that used an image processing algorithm to boost frame rates by over 70%. On March 26th, Remedy is releasing an update that adds new story-based downloadable content along with the latest AI-powered DLSS 2.0.

“With Control, we set out to create a visually stunning and immersive world,” said Mika Vehkala, Director of Technology at Remedy Entertainment. “Real-time ray tracing and NVIDIA DLSS made Control look amazing at launch, and upgrading to DLSS 2.0 made the game’s image quality better than ever.”

 

At 1920x1080 and 2560x1440, DLSS 2.0’s Quality mode boosts performance by up to 76% in our testing. And at 4K (3840x2160), Performance mode delivers gains of 2-3X, enabling even GeForce RTX 2060 gamers to run at max settings at a playable framerate.

While the original implementation of DLSS in Control boosted performance and provided great image quality across most scenes, it tended to struggle in areas with objects in motion. In contrast, DLSS 2.0 is able to handle these cases much more effectively as seen in the image of a spinning fan below.

Images captured at 1080p using DLSS 2.0’s Quality Mode

The latest version of DLSS also improves image quality where the details are a bit more subtle. The following images show additional examples of improvements from the original version of DLSS to the updated version of DLSS 2.0.

Images captured at 1440p using DLSS 2.0’s Quality Mode

Elsewhere, text on seals and other detail is similarly improved, enhancing image quality and fidelity.

Images captured at 1440p using DLSS 2.0’s Quality Mode

And flat textured game elements, such as posters and the map below, also see noticeable improvements to clarity and sharpness

Images captured at 1080p using DLSS 2.0’s Quality Mode

To get DLSS 2.0 for Control, download and install our newest Game Ready Driver, and on March 26th download a new Control game update. Enable DLSS in-game, select a display resolution, and then a DLSS Rendering Resolution, equivalent to the Quality, Balanced and Performance modes seen in other DLSS 2.0 titles.

Deliver Us The Moon

KeokeN Interactive’s space and Moon-based adventure, Deliver Us The Moon, added ray-traced effects and DLSS 2.0 late last year, boosting its beauty and performance.

 

“We never expected the image quality in Deliver Us The Moon to increase when DLSS was enabled, but that is exactly what we experienced,” said Koen Deetman, CEO and Game Director at KeokeN Interactive. “This coupled with real-time ray tracing, and the huge performance boost from DLSS, gives our players the ultimate Deliver Us The Moon experience without any compromises.”

Deliver Us The Moon Adds Stunning Ray-Traced Effects and NVIDIA DLSS

Performance leaps by 60% in a single bound with NVIDIA DLSS 2.0 enabled in Deliver Us The Moon

Throughout the game, DLSS 2.0 delivers image quality on par with native resolution gameplay utilizing Deliver Us The Moon’s built-in Temporal Anti-Aliasing technique. Under closer examination, DLSS 2.0 even shows subtle improvements, with better fidelity on railings and other fine detail:

Deliver Us The Moon Adds Stunning Ray-Traced Effects and NVIDIA DLSS

In other instances, DLSS 2.0 delivers clear improvements, such as on intricate cyclone fencing, which boasts better image quality and superior temporal stability (clarity while the player’s view is moving).

Deliver Us The Moon Adds Stunning Ray-Traced Effects and NVIDIA DLSS

And on the game’s many computer displays and surfaces with miniscule text, DLSS 2.0 shows more detail at a greater level of clarity.

NVIDIA DLSS In Deliver Us The Moon: Fortuna

MechWarrior 5: Mercenaries

Piranha Games’ MechWarrior 5: Mercenaries is the latest game to add NVIDIA DLSS 2.0, and as in other DLSS-enhanced titles, performance improves dramatically, with gains of up to 75% recorded when using Quality mode:

“NVIDIA DLSS 2.0 basically gives our players a free performance boost, without sacrificing image quality” said Russ Bullock, President at Piranha Games. “It was also super easy to implement with NVIDIA’s new SDK, so it was a no brainer for us to add it to MechWarrior 5.”

 

Image quality is also improved with NVIDIA DLSS 2.0 in a number of cases. In the comparison below, captured at 1440p using Quality mode, DLSS adds sharp detail to the intricate parts of the ‘mech, showing the wear and tear. In addition, DLSS also stabilizes the cyclone fencing on top of the building, for better fidelity during gameplay.

In this example, captured at 1080p with Quality Mode, DLSS maintains sharp, clear image quality while boosting frame rates. Note also the fine line details on the barn and the increased fidelity of the fields of grain.

And finally, in this example captured at 1080p using Quality mode, DLSS restores the guard railing at the top of the wall, and sharpens smaller details both on the wall and the areas on either side of the gate.

Wolfenstein: Youngblood

Bethesda and Machine Games’ Wolfenstein: Youngblood is the first Vulkan-powered title to add ray tracing and DLSS 2.0.

 

Out of the gate Wolfenstein: Youngblood performs extremely well, and with DLSS 2.0 it gets even faster:

“It was critically important to us that our game standout both visually and from a performance perspective,” said Jim Kjellin CTO at MachineGames, developer of Wolfenstein: Youngblood. “Adding NVIDIA DLSS enabled us to get the best of both worlds, maximum performance with incredible image quality."

DLSS maintains detail, while also improving image quality in certain spots. In the example below, grating on the right side of the screen, and the pipes on the building in the center, are more defined and detailed with DLSS 2.0 enabled, compared to Wolfenstein: Youngblood’s highest-quality TAA mode:

Clear improvements are evident in other locations, too. For example, our second comparison sees detail increase significantly on the building to the right, and the antennas on the roof become clearer and more visible, improving the quality of the scene.

DLSS Continues to Learn

With the Turing architecture, we set out to change gaming with two big leaps in graphics: real-time ray tracing and NVIDIA DLSS. Ray tracing brings next-generation realism, while DLSS boosts framerates to help you enjoy those stunning ray-traced visuals.

With NVIDIA DLSS 2.0, we’ve made big leaps in image quality and performance, all while paving the way for easier integration so gamers can enjoy the technology in more and more games. And thanks to the power of AI, our DLSS network will continue to grow and improve over time.

We hope you get a chance to try out our latest work in MechWarrior 5 and Control this week, and stay tuned to GeForce.com for more news on upcoming DLSS games and advancements.

Comments