Less Can Be More: A Footprint-Driven Heuristic to Skip Wasted Connections and Merges in Bidirectional Rendering

Implementation of the EGSR 2025 paper "Footprint-driven cost reduction for bidirectional rendering"

https://doi.org/10.2312/sr.20251180

Abstract

Bidirectional rendering algorithms can robustly render a wide range of scenes and light transport effects. Their robustness stems from the fact that they combine a huge number of sampling techniques: Paths traced from the camera are combined with paths traced from the lights by connecting or merging their vertices in all possible combinations. The flip side of this robustness is that efficiency suffers because most of these connections and merges are not useful – their samples will have a weight close to zero. Skipping these wasted computations is hence desirable. Prior work has attempted this via manual parameter tuning, by classifying materials as “specular”, “glossy”, or “diffuse”, or via costly data-driven adaptation. We, instead, propose a simple footprint-driven heuristic to selectively enable only the most impactful bidirectional techniques. Our heuristic is based only on readily available PDF values, does not require manual tuning, supports arbitrarily complex material systems, and does not require precomputation.

Dependencies

• .NET 9.0

Running

To run (and build) the default experiment use the following command-line in your favorite terminal:

cd Experiments
dotnet run -c Release

The executable allows the following arguments to be passed alongside it:

• --help To get the following help displayed in the terminal.
• --width [NW], --height [NH] Change the default (1280x720) image size to [NW]x[NH].
• --eq-sample [N] Conduct a equal-sample benchmark instead of an equal-time with [N] samples.
• --time [S] Change the default maximum rendertime (30s) to [S] seconds.
• --no-reference Do not try to compute a reference automatically.
• --scene Load a specific scene instead of the defaults. Can be specified multiple times.

To specify arguments for the application using dotnet use the following example:

cd Experiments
dotnet run -c Release -- --width 800 --height 600 --time 300 --no-reference --scene RGBSofa --scene StillLife

Results after a successful run can be found in the corresponding Experiments/Results*/ directory. The actual name of the result directory depends on the user arguments. Default is Experiments/ResultsEqTime30s/.

The environment variable SEESHARP_SCENE_DIRS can be used to add directories containing scenes.

Project Outline

This project is divided into multiple directories

• Experiments/ Contains the code to evaluate Ours against the Baseline and a PathTracer. Also contains the results after a successful run.
• Integrators/ Contains the reference implementation of our heuristic Ours.
• Scenes/ Contains scenes to evaluate against.

Scenes

This project ships the following scenes used for the evaluation in the paper:

CornellBox

DragonInIce

RGBSofa

RoughGlassesIndirect

StillLife

TargetPractice

VeachBidir

Funding

This project was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 540768663.

Bibtex

@inproceedings{Yazici2025,
    booktitle = {Eurographics Symposium on Rendering},
    editor    = {Wang, Beibei and Wilkie, Alexander},
    title     = {{Less can be more: A Footprint-driven Heuristic to skip Wasted Connections and Merges in Bidirectional Rendering}},
    author    = {Yazici, Ömercan and Grittmann, Pascal and Slusallek, Philipp},
    year      = {2025},
    publisher = {The Eurographics Association},
    ISSN      = {1727-3463},
    ISBN      = {978-3-03868-292-9},
    DOI       = {10.2312/sr.20251180}
}