A reference implementaion of the paper "Revisiting Controlled Mixture Sampling for Rendering Applications" by Qingqin Hua, Pascal Grittmann and Philipp Slusallek, SIGGRAPH 2023.

The implementaion contains our path tracing application (Section 6 and 7): Control variate formulate with per-light sampling technique and Bsdf sampling technique in global illumination.

The project is based on SeeSharp v1.9.0 and .NET 7.0

Simply run

cd ControlledMixtureSampling
dotnet run -c Release

The code generates both equal-Spp results and equal-time results for our per-light path tracing results. You can view the results under ControlledMixtureSampling\bin\Release\net7.0\Results\PT.

We tested our code on Windows (Visual Studio 2022 Win64). It should be runnable on x86-64 Windows, Linux, and macOS by easily run the same commands above. Please follow SeeSharp for detailed compliation instructions.

MakeFigures provides two python scripts to generate figures.

• MakeFigures\MakeFigure8.py generates Fig. 8 in the paper.
• MakeFigures\MakeSupplemental.py generates supplemental for different spatial grid resolutions. You can also download the supplemental here.

Below shows an equal-time result of ours in the setting of different grid resolution in full GI.

Presentation in Mandrain with slides in English