COS 526 - Advanced Computer Graphics	Fall 2018

Assignment 3 - Photon Mapping

In this assignment you will implement photon mapping, an algorithm for image synthesis with global illumination. Part I of the assignment is to implement a basic system, while part II involves implementing an extension of your choosing, and producing a realistic rendering of some real-world object or scene.

I. Implementing a Photon Mapper

• Photon emission:Implement code to emit photons in random directions from every light source in a scene. The total number of photons emitted for each light source should be proportional to the power of the light source (so that each photon carries approximately equal power), and the distribution of photons should be proportional to the power in each direction -- e.g., for spot lights (section 2.1.1 in Jensen01).
• Photon scattering: Trace photons via reflections and transmissions through the scene. At each ray-surface intersection, randomly generate a secondary ray along a direction of diffuse reflection, specular reflection, transmission, or absorption with probability proportional to kd, ks, kt, and (1 - kd+ks+kt), respectively (section 2.1.2 in Jensen01).
• Russian Roulette: At each surface intersection, terminate rays with probability p (e.g., p=0.5) and multiply the power of surviving rays by 1.0/p (section 2.1.2 in Jensen01). See section 8.5 of these siggraph course notes for details.
• Photon storage: Store photon-surface intersections in a kd-tree, retaining the position, incident direction, and power of each photon. (section 2.1.3 in Jensen01). You can use your code from assignment 2, or the R3Kdtree class in R3Shapes to implement this feature.
• BRDF importance sampling: Select the directions of reflected and transmitted rays with probabilities proportional to the Phong BRDF of the scattering surface. See Jason Lawrence's notes for details.
• Multiple photon maps: Implement separate photon maps for global (L{S|D}*D) and caustic (LS+D) ray paths (section 2.1.5 in Jensen01).

Rendering:

• Camera ray tracing: Generate a ray(s) from the camera eye point through each pixel. Trace them through the scene with reflections and transmissions at surface intersections -- i.e., at each ray-surface intersection, randomly generate a secondary ray along a direction of diffuse reflection, specular reflection, transmission, or absorption using importance sampling. (section 2.4 in Jensen01).
• Radiance estimation: Use the kd-tree to find the N closest photons for each ray-surface intersection. Estimate the radiance traveling along the ray towards the camera from the power of those photons. (section 2.3.1 in Jensen01).
• Pixel integration: Trace multiple rays per pixel and average the radiance computed for all rays to estimate the radiance to store in the output image for each pixel. Compare the results with different numbers of rays per pixel (N).

Visualization: (helpful for debugging!)

• Photon map visualization: Visualize photons stored in your photon map(s) -- e.g., show positions, normals, and powers of photons.
• Ray tracing visualization: Visualize ray paths traced from the camera -- e.g., show line segments between the camera and successive surface intersections for a random sampling of rays.

Getting started:

To get started, download cos526_photon.zip (courtesy Tom Funkhouser). This contains some scenes to get you started, as well as C++ code providing the basic infrastructre for reading scenes, computing ray intersections, etc. It also provides a simple program (scnview) for viewing scenes using OpenGL. You will probably need to augment this program to include command line arguments of your own to turn on and off specific features and/or provide parameters for specific applications.

The skeleton code is able to read scenes in a simple scene file format, This format was created to provide the features required by this assignment -- a simple scene graph along with materials, lights, and cameras. We provide several scenes in the input subdirectory of the zip file that you can use to test basic functionality of your program. However, these scenes are not enough to test all the interesting lighting effects your program should demonstrate. So, you should design your own scenes for testing/demonstration and include them in your writeup file.

If you are using a programming language other than C++, feel free to use the provided code as a starting point, and translate to your languange of choice. Also feel free to use any ray-tracing code you may already have as reference when implementing your photon mapper.

Useful resources:

• Global Illumination using Photon Maps, Henrik Wann Jensen, 1996.
• Realistic Image Synthesis Using Photon Mapping, Henrik Wann Jensen, Second Edition, 2012.
• Notes by Zack Waters, WPI.

Example solutions from previous years:

• Photon Mapping Assignment Writeup, Nik Sigatapu, COS 526, Fall 2014.
• Photon Mapping Assignment Writeup, Maciej Halber, COS 526, Fall 2014.
• Photon Mapping Assignment Writeup, Brian Matejek, COS 526, Fall 2014.
• Photon Mapping Assignment Writeup, Nora Willet, COS 526, Fall 2014.
• Photon Mapping Assignment Writeup, Abuhair Saparov, COS 526, Fall 2012.
• Photon Mapping Assignment Writeup, Edward Zhang, COS 526, Fall 2012.
• Photon Mapping Assignment Writeup, Ohad Fried, COS 526, Fall 2012.

Submitting Part I

Part I of this assignment is due Friday, December 14.

Please submit a single .zip file containing your code and a writeup describing your implementation and showcasing some pretty pictures!

The Dropbox link to submit the assignment is here.

II. Implementing Advanced Effects

• Filtering: Implement disk, cone, and/or Gaussian filtering methods for weighting photons during radiance estimation and compare the results of different methods. (section 2.3.2 in Jensen01).
• Participating medium: Implement volume photon maps that represent scattering of participating medium (e.g., fog) (sections 2.1.4 and 2.3.3 in Jensen01).
• Projection maps: Use projection maps to sample photons leaving light sources in the directions of scene geometry (section 2.1.1 in Jensen01).
• Something else: Extend your photon mapper to handle subsurface scattering, motion blur, or another feature described in Jensen01.

Rendering a realistic object:

• Pick a real-world object or scene, and use your system to produce a rendering or short animation that is as photorealistic as possible. See this page for inspiration. Note that most of your effort should be spent on modeling optical phenomena, not on geometric modeling.

Submitting Part II

Part II of this assignment is due on Dean's date, Tuesday, Jan 15. Per university policy, no extensions of this deadline will be available.

Please submit a single .zip file containing your code and a writeup describing your implementation and showcasing some pretty pictures!

The Dropbox link to submit the assignment is here.