NVIDIA RTXGI 2.0 Now Available: The Next Frontier of Ray Traced Visuals With Neural Radiance Cache, Spatial Hash Radiance Cache & Dynamic Diffuse Global Illumination Support

NVIDIA has updated its ray tracing global illumination SDK to the latest feature set in the RTXGI 2.0 update, offering support for new technologies.

At GDC 2024, NVIDIA is announcing its brand new RTXGI 2.0 update which adds a range of new technologies that the company has previously teased.

The new feature set makes it easier for developers to harness fast and efficient radiance caching techniques for faster performance in the most demanding applications such as games. The best part is that the solutions provided by NVIDIA under its RTXGI SDK are accessible for any ray-tracing-capable GPU and further take advantage of AI-accelerated algorithms such as the Tensor core architecture featured on NVIDIA's RTX GPUs. The full SDK is available on Github at this link.

The three major features being introduced in NVIDIA's RTXGI 2.0 update are:

First up, let's talk about NRC or Neural Radiance Cache which was proposed as a technique back at Siggraph 2021. The fundamental aim of this technique is to make advanced ray tracing and path tracing effects faster on NVIDIA GPUs by utilizing AI-accelerated hardware such as Tensor Core technology.

The Neural Radiance Cache is an AI technique aimed at improving signal quality and potentially performance in the context of pathtracing. The NRC operates in world space and predicts radiance at any point in the virtual world using path-traced live-trained data.

via NVIDIA

The algorithm works by training the radiance cache about a scene in real time without assuming lighting, geometry, and materials. AI does the main work here by training the radiance cache in real-time for improved performance and accurate GI (Global Illumination). Following is a before and after comparison showcasing how NRC enhances the signal quality of the radiance data:

NVIDIA Neural Radiance Cache Disabled:

NVIDIA Neural Radiance Cache Enabled:

NVIDIA states that the NRC library is currently available in experimental mode and is being actively developed.

Next up, we have SHaRC or Spatial Hash Radiance Cache which is built upon Spatial hashing data structures and provides a fast plus scalable global illumination technique for path tracing. SHaRC works very similar to NRC but it doesn't leverage AI or neural networks and works with any DirectX or Vulkan ray-tracing GPU.

Spatial Hash Radiance Cache (SHaRC) is a technique aimed at improving signal quality and performance in the context of path tracing. The SHaRC operates in world space and provides a radiance value at any hit point.

via NVIDIA

Following is a scene in Cyberpunk 2077 with SHaRC enabled & disabled. The advantage shown here is that with SHaRC, the GI lighting covers a much larger area and works accurately while having a negligible impact on performance.

NVIDIA Spatial Hash Radiance Cache Disabled:

NVIDIA Spatial Hash Radiance Cache Enabled:

Lastly, we have the DDGI or Dynamic Diffuse Global Illumination technique which offers accelerated global illumination in Path Tracing by providing infinite bounce lighting in real-time. DDGI works by delivering multi-bounce indirect lighting without offline light maps or light baking. The following comparison made using Unreal Engine shows DDGI providing more lighting detail and adding realism to the scene.

NVIDIA Dynamic Diffuse Global Illumination Disabled:

NVIDIA Dynamic Diffuse Global Illumination Enabled:

NVIDIA is really on top of the world at the moment. Not only is the company fueling the AI segment with innovations but it is also accelerating the graphics space. Since the introduction of RTX, the green team has taken charge of accelerating the era of ray tracing and path tracing to the next level. They are the leading force for the next generation of visuals and these new techniques in the RTXGI 2.0 update are going to help developers offer the most lifelike and immersive visuals to gamers and consumers.