DLSS 5 for Sim Racing: Games, Mods and Support

NVIDIA DLSS 5 is now officially available, but sim racing is already taking the technology beyond its initial supported games. No major racing simulator currently offers native DLSS 5 support, yet community tools and mods are already getting Neural Rendering working experimentally in titles including Automobilista 2 and Assetto Corsa.

NVIDIA launched DLSS 5 on September 3, 2026, with NBA 2K27 becoming the first officially supported game. Its headline feature is 3D-Guided Neural Rendering, a new stage in the graphics pipeline designed to enhance lighting and material detail while remaining anchored to the original image produced by the game engine.

For sim racers, however, the most interesting developments are currently happening outside official developer support. Community projects are finding ways to feed the information required by DLSS 5 into games that were never designed around the technology, opening the door to experiments in both modern and older racing simulators.

Assetto Corsa is becoming DLSS 5 test bed

Assetto Corsa is another obvious candidate for this type of experimentation. More than a decade after its original release, Custom Shaders Patch continues to extend the simulator’s graphics pipeline with rendering features that were never part of Kunos Simulazioni’s original engine.

Recent CSP preview development has already introduced modern upscaling functionality, while community users are now experimenting with DLSS 5 Neural Rendering alongside those graphics modifications. The result is a particularly flexible environment for testing how NVIDIA’s new rendering model behaves in a racing simulator.

As with AMS2, this should not be confused with official support. Kunos has not announced DLSS 5 for the original Assetto Corsa, and the Neural Rendering experiments currently circulating within the community depend on external components and custom configurations.

Early testing also exposes some of the problems that matter specifically in racing games. Ghosting and disocclusion can become visible around wheels, trackside barriers, vegetation and other objects moving quickly relative to the camera. These situations are considerably more common in a racing simulator than in the relatively controlled scenes used to demonstrate many graphics technologies.

Assetto Corsa can therefore be described as having experimental community DLSS 5 implementations, but not native DLSS 5 support.

DLSS 5 is already working in Automobilista 2

Automobilista 2 has become one of the clearest examples of what the community is already doing with DLSS 5.

A guide published on the Reiza Studios forum on August 31 demonstrated Neural Rendering running inside AMS2 before NVIDIA had even released DLSS 5 commercially. The simulator has no native DLSS implementation, so the process relies on several external components to create the rendering information needed by NVIDIA’s Neural Rendering system.

The current AMS2 method uses ReShade with full add-on support alongside DLSS5-Feeder, DLSS 5 Bridge and the RenoDX DLSS tools. Together, they create a pathway that allows the Neural Rendering component to process a game that does not normally make DLSS calls.

The results can be impressive, particularly in the way cars interact with track lighting and shadows. Users have highlighted improvements to the perceived grounding of vehicles, wheel lighting and general scene depth, but this comes with a considerable performance cost. Reports from the community include major frame-rate reductions even on very powerful GeForce RTX hardware.

There are also visual artifacts, and Reiza Studios has not announced or endorsed DLSS 5 support. The current solution should therefore be treated strictly as an experimental community implementation, rather than a new graphics feature available in Automobilista 2 itself.

The full installation discussion and ongoing testing can be followed through the AMS2 DLSS 5 community guide.

The tools making unofficial DLSS 5 possible

A key part of the story is DLSS5-Feeder, a community project designed specifically for DirectX 11 games that do not contain an existing DLSS implementation.

DLSS 5 Neural Rendering normally works by connecting to DLSS calls already being made by the game. If a title does not use DLSS, those calls do not exist. DLSS5-Feeder creates a synthetic DLAA-style evaluation using information available through ReShade, including the current frame, depth buffer and estimated motion vectors.

That information is processed through a private DirectX 12 device before the Neural Rendering result is copied back into the final frame. In practical terms, it creates the missing connection that allows DLSS 5 experiments to take place in games that would otherwise provide Neural Rendering with nothing to process.

Motion estimation is particularly important in this process. Some current configurations use ReshadeMotionEstimation to generate motion information when accurate engine motion vectors are not directly available.

Another important project is DLSS 5 Bridge. This ReShade add-on can mirror DLSS information from DirectX 11 or Vulkan into a private DirectX 12 session, providing the environment required by the Neural Rendering add-on.

The bridge can also create a substitute DLSS contract for games without DLSS by using ReShade depth information and NVIDIA Optical Flow. Its developer explicitly notes that this route provides lower-quality input than using a game’s own DLSS data, something that becomes important when judging visual artifacts or image stability.

For games that already have compatible temporal upscaling, OptiScaler_DLSSNR takes another approach. The experimental OptiScaler fork adds NVIDIA Neural Rendering as an extra processing pass after the existing upscaler output.

Its developer makes an important distinction: NVIDIA has not documented or officially supported this integration method. The project directly drives the Neural Rendering model, making it useful for experimentation but fundamentally different from a developer integrating DLSS 5 through NVIDIA’s official tools.

There are also projects attempting to make this increasingly complicated ecosystem easier to use. 1-Click-DLSS5 combines several installation routes into a single utility, including feeder-based configurations, OptiScaler methods and direct injection depending on the target game.

Meanwhile, the RenoDX DLSS Installer packages ReShade, the RenoDX DLSS add-on and updated Streamline/DLSS components into a more straightforward installation process.

These tools are developing extremely quickly, which also means compatibility, installation steps and required components can change from one release to another.

Official DLSS 5 support is very different from a mod

The fact that Neural Rendering can be made to operate in an unsupported simulator does not mean a community implementation can reproduce everything available to developers using DLSS 5 natively.

According to NVIDIA’s DLSS 5 documentation, 3D-Guided Neural Rendering works from the game engine’s rendered frame while using additional information such as motion vectors, surface albedo, lighting and surface normals to guide the final result.

Developers can also tune how aggressively Neural Rendering changes the scene and control its behaviour using masks. That could be especially useful in a racing simulator, where the materials and lighting requirements of a car body are very different from those of tyres, glass, cockpit fabrics, asphalt or vegetation.

A community solution may not have direct access to all of that information. When engine-generated motion vectors or other buffers are unavailable, tools have to reconstruct or approximate them. That is why an unofficial implementation may technically work while still producing ghosting, unstable edges or incorrect changes to parts of the image.

This distinction will remain important as more DLSS 5 mods appear. A game running Neural Rendering through ReShade or OptiScaler is not equivalent to the developer adding native DLSS 5 support.

Why DLSS 5 could matter for sim racing

DLSS 5 should not be understood as another version of DLSS Super Resolution. The new 3D-Guided Neural Rendering system is an additional graphics stage focused primarily on the appearance of lighting and materials rather than simply reconstructing an image from a lower rendering resolution.

NVIDIA says the system can enhance characteristics including contact shadows, material response, light transmission and other lighting effects while keeping the underlying engine-generated frame as its foundation.

That makes racing games an interesting potential use case. A modern simulator has highly reflective car paint, carbon fibre, glass, rubber, metal, cockpit fabrics and complex track surfaces all visible at the same time. Rain, night racing, headlights and changing environmental conditions add another layer of difficult lighting interactions.

Wet asphalt is an obvious example. A native implementation could potentially use Neural Rendering to improve how reflected light interacts with the road while separately handling car paint, glass and trackside materials.

The same applies inside the cockpit, where leather, Alcantara, carbon fibre, displays and exterior lighting can all occupy a relatively small part of the frame but contribute significantly to visual realism.

DLSS 5 does not necessarily mean more FPS

This is one of the most important differences for sim racers to understand. DLSS has become closely associated with improving performance, but the Neural Rendering component introduced with DLSS 5 requires additional GPU processing.

DLSS 5 can operate alongside technologies such as Super Resolution and Frame Generation, but those remain separate parts of NVIDIA’s DLSS suite. Neural Rendering itself is primarily adding visual processing rather than acting as a conventional performance multiplier.

That distinction matters considerably in racing simulators. Many sim racers target 120Hz, 144Hz or higher refresh rates, while VR and triple-screen configurations already place unusually heavy loads on both the GPU and CPU.

An effect that looks excellent during a replay may be less attractive if it significantly reduces the underlying frame rate while driving. The early Automobilista 2 experiments already demonstrate this trade-off, with some users reporting substantial performance losses after enabling Neural Rendering.

Native implementations may perform differently, particularly when developers can provide better engine data and optimise exactly where DLSS 5 is applied. Until those arrive, performance results from current community mods should not be treated as representative of what an officially integrated racing simulator could eventually achieve.

Which racing sims currently support DLSS 5?

No major racing simulator has announced official native DLSS 5 3D-Guided Neural Rendering support.

Several titles already use other NVIDIA DLSS technologies, while Automobilista 2 and Assetto Corsa are currently the most visible examples of community experimentation.

Last updated: September 6, 2026. Community DLSS 5 refers to unofficial mods and experimental implementations. It does not indicate support from NVIDIA or the simulator developer.

Assetto Corsa EVO is one of the obvious titles to watch because Kunos has already moved its NVIDIA implementation to DLSS 4.5. That does not make the simulator compatible with DLSS 5 Neural Rendering, however, and there has been no announcement from Kunos or NVIDIA confirming an upgrade.

The same distinction should be applied to every racing title. Existing DLSS Super Resolution, Frame Generation or DLSS 4.5 support should never be interpreted as confirmation that DLSS 5 Neural Rendering is available.

Be careful using DLSS 5 mods online

There is also an important multiplayer consideration. Most of the current methods rely on ReShade add-ons, injected DLLs or other components that interact directly with the game’s rendering process.

That can create problems with anti-cheat systems regardless of whether the modification provides any competitive advantage.

The OptiScaler project warns against using its modifications in online games because anti-cheat software can potentially detect the injected components. The RenoDX installer similarly describes its ReShade build as intended for single-player games and warns that multiplayer use may carry a ban risk.

For that reason, users should not assume that an unofficial DLSS 5 implementation is safe for iRacing, ranked competitions or another anti-cheat-protected environment.

Until a particular simulator or anti-cheat provider explicitly confirms compatibility, community Neural Rendering experiments are better treated as offline graphics modifications.

What comes next for DLSS 5 in sim racing?

DLSS 5 is still at the very beginning of its commercial rollout. NBA 2K27 currently provides the first official implementation, while NVIDIA says additional games will adopt the technology over time.

No major racing simulator has joined that list yet, but the community has already demonstrated that waiting for official integration is not the only route.

Projects such as DLSS5-Feeder, DLSS 5 Bridge and OptiScaler_DLSSNR could expand that experimentation to significantly more racing titles. The bigger question is no longer simply whether DLSS 5 can run in a simulator, but whether it can deliver stable image quality and acceptable performance while cars, cameras and environments are moving at racing speeds.

VR and triple-screen performance will be particularly important, as will latency and compatibility with competitive online environments.

For now, the situation is straightforward: DLSS 5 has already reached sim racing through community tools, but native support from racing game developers has yet to begin.

DLSS 5 tools and resources

The projects below are some of the main tools currently involved in community DLSS 5 experimentation. They are evolving rapidly and should not be considered official NVIDIA software unless specifically stated otherwise.

Compatibility, installation methods and hardware requirements can change quickly as these community projects are updated.

See you on the track!