Accelerated Photon Mapping for Hardware-based Ray Tracing (using Falcor 5.2)

A Progressive Photon Mapper that uses the ray tracing hardware for distribution and collection of photons. Our Photon Mapper uses NVIDIA's Falcor 5.2 framework. More details on how the Progressive Photon Mapper works can be found in our paper published in the Journal of Computer Graphic Techniques here.

Prerequisites

• Windows 10 version 20H2 (October 2020 Update) or newer, OS build revision .789 or newer
• A GPU which supports DirectX Raytracing, such as the NVIDIA Titan V or GeForce RTX
• NVIDIA driver 466.11 or newer

Additionally needed for the precompiled version:

• Microsoft Visual C++ Redistributable x64 ( Most windows programs need these, so they are probably already installed. If the applications do not start, try installing these.)

Additionally needed for building the code:

• Visual Studio 2019 or 2022
• Windows 10 SDK (10.0.19041.0) for Windows 10, version 2004

Setup

This section can be skipped if using the precompiled files (Release Build).

The setup is identical to Falcor 5.2. For the sake of simplicity, we only describe how to set up Falcor with Visual Studio 2019 and 2022. For other build options and additional information, see the Building Falcor section here.

Visual Studio

You can set up a native Visual Studio solution by running setup_vs2019.bat (or setup_vs2022.bat, same process) after cloning this repository. The solution files are written to build/windows-vs2019-d3d12 and the binary output is located in build/windows-vs2019-d3d12/bin.

Alternatively, you can use CMake. Run the setup.bat and then use the CMake Presets windows-vs2019-d3d12 or windows-vs2022-d3d12 with the command cmake --preset <preset name>. You can find more information for generating the build files using CMake under the Building Falcor/Configure Presets section here.

Build in Visual Studio with Build -> Build Solution for the correct build order. Otherwise, multiple builds may be necessary.

Loading in Falcor Scripts and Scenes

• Start the renderer Mogwai:
  • Build Solution with Build -> Build Solution in Release or Debug configuration. This step is skipped for the precompiled version.
  • Run RunMogwai.bat, which loads in the RTPM render pass and the caustic glass scene.
    • Alternatively, start Mogwai via Visual Studio or from FalcorData/Mogwai.exe for the precompiled version. In this case, a photon map script and scene need to be loaded separately.
• Load a render pass:
  • Load a render pass in Mogwai under File -> Load Script. You can find the photon map scripts in the PhotonMapPasses folder.
• Load a scene:
  • Load a scene in Mogwai under File -> Load Scene.
  • Most scenes from the paper can be found in the Scenes folder. Load in a .pyscene file to get the same results as in the paper.
    • For information about settings used in the paper, see Scenes\SceneSettings.csv.
    • The Amazon Lumberyard Bistro scene needs to be downloaded separately (here). Our test scenes are marked with an RTPM prefix. For more information see Scenes/Bistro_v5_2/BISTRO_README.txt.
  • All Falcor-supported scenes with emissive lights and analytic spot/point lights are supported. Environment maps are currently not supported.

Examples

• Example renders can be found under the ExampleImages folder.
• A demo video showing the photon mapper and some functions can be found under: https://youtu.be/PVgip3L8q-I

Render Passes

Graph	Description
VBufferPM	A modified V-Buffer that traces the path until it hits a diffuse surface.
RTPhotonMapper	The ray tracing hardware-based progressive photon mapper. Photons are distributed through the scene with ray tracing. An acceleration structure is built with the distributed photons that are then collected with an infinite small ray. The user controls the number of photons and needs to ensure that the photon buffer is big enough. You can access additional information for each UI variable by hovering over the question mark on the right side of the UI variable.
HashPPM	An alternative implementation of the RTPhotonMapper using a hash grid for collection. The photons are still distributed with ray tracing but are now stored in a hash map. Like with the RTPhotonMapper, the user has to ensure that the photon buffer is big enough. For additional information, hover over the question mark on the right side of the UI variable.
StochHashPPM	An alternative implementation of the RTPhotonMapper using a stochastic hash grid for collection. Photons are distributed via a ray tracing shader and are stored in a hash grid. On collision, the photon is randomly overwritten. For additional information, hover over the question mark on the right side of the UI variable.

Source Code

We recommend Visual Studio 2019 or 2022 for navigating the source code. You can find the code for our Photon Mappers in the respective folder under Source/RenderPasses. For more information about how to use Falcor, see the getting started or the full documentation index