Built on Direct3D 12, the project uses a custom native UnityRHI plugin to integrate DLSS Neural Rendering into Unity as a URP post-processing effect. Its parameters can be adjusted through the Volume system.
The project includes:
UnityRHI C# and native plugins
A DLSS Neural Rendering Renderer Feature
A prebuilt Windows x64 native release, so no native compilation is required
Since I don’t have access to a GeForce RTX 50-series GPU, the project has currently only been tested on RTX 30-series GPUs.
This project was developed with extensive use of AI-assisted programming.
Project repository:
Prebuilt release:
nvngx_dlssnr.dll is not included in the release. Please obtain it separately and place it in the appropriate directory.
The project is still under active development. Feedback, testing, discussions, and issue reports are all welcome!
DLSSNR does not support HDR input, so the image must first be converted to SDR by the post-processing tonemapper before it can be used as input. This is also the generally recommended setup.
If you want to run NR before upscaling for performance reasons, you’ll have to disable HDR. This is mainly a limitation of Unity’s rendering pipeline, where post-processing must occur after upscaling.
NVIDIA’s official DLL only supports RTX 50-series GPUs. If you’re using an RTX 20-, 30-, or 40-series GPU, you’ll need a third-party DLL. So far, I’ve only tested the integration on RTX 30- and 40-series GPUs.
