Practical Path Guiding for Efficient Light-Transport Simulation

Implementation of "Practical Path Guiding for Efficient Light-Transport Simulation 2017" by Thomas Müller, Markus Gross, and Jan Novák in Mitsuba3. This is done as an exercise in the Computer Graphics Lab, University of Bonn. Please read the full report lab_report.pdf for full detail.

Package dependencies

• Mitsuba3
• matplotlib
• pandas
• progressbar

File description

Rendering

• main.py : render a scene using the path guiding algorithm.
• path_tracing_render.py : render a scene using path tracing with NEE.
• repeat_high_spp_renderer.py : repeatedly render the same scene from main.py using same SD-tree data that has been generated. For averaging performance result.

Analysis

• tree_plotter.py : visualize quadtree radiance at a given position.
• performance_plot.py : plot variance and mean square error against budget. For comparison performance, you will have to run path_tracing_render.py and repeat_high_spp_renderer.py first.

Qualititve comparision