NVIDIA Unveils RTX Kit Update with Revolutionary AI-Powered Neural Rendering Technologies

NVIDIA's RTX Kit for developers updated with game-changing AI-powered RTX Neural Shaders

It's been a gradual shift, but with the arrival of NVIDIA's GeForce RTX 50 Series, we're seeing a clear shift to AI-enhanced rendering becoming a major component of gaming's future. We're already seeing the benefits of this today with technologies like the AI-powered NVIDIA DLSS and Frame Generation, which is spilling over into the console space thanks to the PlayStation 5 Pro. AMD, too, is set to switch its FSR upscaling model to an AI-based one to match DLSS and Intel's XeSS. Microsoft's next Xbox console is rumored to heavily use ML or machine learning hardware and software. One catalyst for the change has been the arrival of cutting-edge real-time ray-tracing, as seen in 'Full Ray Tracing' titles like Cyberpunk 2077 and Alan Wake 2. With the arrival of the GeForce RTX 50 Series, NVIDIA is not only enhancing the AI capabilities of its next-gen RTX Blackwell architecture for new AI-enhanced rendering but is bringing its innovations to the software space. The company's RTX Kit for game developers has just been given a significant update and is packed with new neural rendering technologies. RTX Neural Shaders, RTX Neural Texture Compression, and RTX Mega Geometry are all about giving developers the tools to render complex scenes without relying on raw performance alone. With the aid of real-time AI models, game environments and characters will look better than ever. Technologies like RTX Neural Texture Compression use AI to compress textures to minimize VRAM utilization. At the same time, RTX Mega Geometry accelerates path tracing to create scenes with more detail that run faster than possible. RTX Mega Geometry is already available in Alan Wake 2, and based on our benchmarks, the game's impressive Path Tracing mode now runs 10-15% faster on GeForce RTX 40 Series and RTX 50 Series GPUs. The new RTX Kit adds a whole suite of neural rendering technologies for the Vulkan API, with DirectX support coming soon. To see what this technology brings, check out NVIDIA's Zorah demo from CES, built with RTX neural rendering tech running in real-time in Unreal Engine 5.

NVIDIA RTX Kit with Neural Shaders Is Out Now; Mega Geometry and Neural Texture Compression Demos Available

A few days after the originally intended launch window of late January, the NVIDIA RTX Kit has finally been released to the public for the most part. NVIDIA RTX Kit can be downloaded via GitHub, though each component also has a dedicated repository. This includes what was arguably the most exciting and future-facing technology: RTX Neural Shaders, now abbreviated as RTXNS. Here's the excerpt from its dedicated GitHub page: The sample is intended as a starting point for developers interested in bringing Machine Learning (ML) to their graphics applications. It provides a number of examples to help the reader understand how to train their own neural networks and then use those models to perform inference alongside their normal graphics rendering. RTXNS uses the Slang shading language and it utilises the Vulkan Cooperative Vectors extension to provide access to the GPUs ML acceleration. RTX Neural Materials is not yet available as part of the RTX Kit suite, but practically everything else is. For example, RTX Neural Texture Compression includes a demo sample that can be compiled to demonstrate the memory-saving technology. Compusemble did just that, finding massively reduced memory consumption (from 272 to 11.37 MB), although there is a performance cost associated when enabling the NTC AI inference model. Similarly, RTX Mega Geometry (RTXMG) can be tested by compiling a new sample. This is the only technology that's already been used in a game (Alan Wake 2), but the tech demo is nonetheless interesting. RTX Character Rendering (RTXCR) is also available, providing a sample of how to render path traced hair and skin. Hair is particularly interesting because, with the new GeForce RTX 50 graphics cards, NVIDIA introduced hardware support for a new hair data structure called Linear-Swept Sphere (LSS). According to NVIDIA, this delivers high performance in tracing and BVH updates while maintaining exceptionally low memory consumption. Compared to Disjoint Orthogonal Triangle Strips (DOTS), rendering animated hair on humans is about 2x faster with LSS and requires one-fifth of the VRAM. We'll see the first example of RTX Hair with the Indiana Jones and the Great Circle February update. The RTX Kit also includes one technology that NVIDIA had not previously discussed at CES 2025: RTX Texture Filtering (RTXTF). According to NVIDIA, RTXTF randomly samples textures after shading and filters difficult volumes, reducing artifacts and improving image quality. Lastly, NVIDIA also updated its Streamline SDK to version 2.7.2, adding support for DLSS 4 Multi Frame Generation and the new models for Ray Reconstruction, Frame Generation, and Super Resolution. This means we might soon see the first DLSS 4 mods.