This is the project accompanying the paper A Scalable and Production Ready Sky and Atmosphere Rendering Technique presented at EGSR 2020. It can be used to compare the new technique proposed in that paper to Bruneton et al. as well as to a real-time path tracer. The path tracer is using a simple and less efficient noise because I was not sure I could share noise code I had before. The technique is used to render sky and atmosphere in Unreal Engine.

Build the solution

1. Update git submodules (run git submodule update --init)
2. Open the solution
3. Make sure you select a windows SDK and platform toolset you have locally for both projects
4. In Visual Studio, change the Application project Working Directory from $(ProjectDir) to $(SolutionDir)
5. Select Application as the startup project, hit F5

Runtime keys:

• SHIFT + mouse to look around
• CTRL + mouse to move the sun around
• C to capture a screenshot
• T toggle between ray-marching and path tracing (disables the multiple scattering approximation when switch to path tracing)
• F5/F9 to save/load a state

Submodules

• imgui V1.62 supported
• tinyexr

About the code:

• The code of Eric Bruneton has been copied with authorization from his 2017 implementation.
• The code in this repository is provided in hope that later work using it to advance the state of the art will also be shared for every one to use.
• The code has been built from using the Dx11Base demo/test platform.
• RenderSkyRayMarching.hlsl contains the ray marcher building the different LUTs for the new technique.
• RenderSkyPathTracing.hlsl contains the basic volumetric path tracer. It is matching PBRT and Mitsuba output for other participating media tests. It could be improved by really following the Radiance transfert Equation path integral as a loop for each event: currently we only handle participating media and intersection with the planet as a special case.
• This code has been tested on Windows only with visual studio (no build script generation)

Thanks to Epic Games for allowing the release of this source code.

The code is provided as is and no support will be provided. I am still interested in any feedback, and we can discuss if you contact me via twitter.

Seb