NVIDIA-RTX/RTXNS

What it solves

RTX Neural Shading (RTXNS) provides a framework for developers to integrate machine learning into graphics applications. It simplifies the process of training neural networks to replace or enhance traditional graphics rendering components, such as shaders or textures, allowing for more advanced visual effects achieved through ML-driven inference.

How it works

RTXNS utilizes the Slang shading language and leverages GPU acceleration via the DirectX Preview Agility SDK or the Vulkan Cooperative Vector extension. It provides a set of examples and helper functions for building custom neural networks. Additionally, it integrates with SlangPy, allowing developers to rapidly prototype neural networks in Python before implementing them for high-performance inference within the graphics pipeline.

Who it’s for

Developers familiar with machine learning and neural networks who are building graphics applications on Windows or Linux and want to use ML to optimize or replace traditional shading and texturing techniques.

Highlights

• Supports both Windows and Linux platforms.
• Compatible with DirectX and Vulkan backends.
• Integrates with SlangPy for rapid Python-based neural network development.
• Includes a progression of examples from simple inference to training networks to represent shaders and textures.