Awesome
DirectX Raytracing: A Path Tracer
A DXR path tracer with OptiX denoising. 5 months worth of research, trial & error as part of a project to learn and understand DirectX Raytracing & raytracing concepts.
- Progressive Monte Carlo pathtracing
- Native DirectX Raytracing
- DXR Fallback Layer
- OptiX deep-learning denoiser
- Anti-aliasing with various sampling patterns
- Reflection
- Refraction
- Material picking & editing
- ImGui
Performance
Specifications:
- NVIDIA GeForce RTX 2080 Founders Edition (non-overclocked)
- Intel Core i6-6600K at 4.1 GHz
- 8 GB RAM at 2133 MHz
- 256 GB NVMe M.2 Samsung SSD
- All tests ran at 720p (1280 x 720)
Test case: Cornell Box (Release mode)
| Bounces | Frame Time |
|---|---|
| 0 | ~3.6ms |
| 1 | ~4.0ms |
| 2 | ~4.3ms |
| 3 | ~4.9ms |
| 4 | ~5.6ms |
| 5 | ~6.3ms |
| 10 | ~10ms |
| 15 | ~12.9ms |
Test case: Crytek Sponza (Release mode)
| Bounces | Frame Time |
|---|---|
| 0 | ~5.1ms |
| 1 | ~6.3ms |
| 2 | ~8.4ms |
| 3 | ~10.0ms |
| 4 | ~11.4ms |
| 5 | ~13.6ms |
| 10 | ~22.7ms |
| 15 | ~31.8ms |
Building the project
- Clone the project
- Download the project's dependencies from here!
- Run CMake on the project.
- Configure for Visual Studio 2017 x64.
Prerequisites to compile & run
- Must be on Windows 10 October 2018 update (RS5 | v1809) or newer
- Must have Windows 10 SDK version 10.0.17763.0 or newer (prefer using that exact version!)
- To run native DXR: NVIDIA Turing or Volta GPU or newer.
- To run fallback: Second generation Maxwell GPU or newer.
- Visual Studio 2017 v15.8.6+.