Digital Image Synthesis

Digital Image Synthesis, Fall 2006

Forum

Jump to...

project #1
project #2
project #3
assignments

final project: PBRT+

Assigned: 2006/12/14
Proposal: 2007/1/4
Due: 2007/01/25

Project description

In this project, you are free to implement existing or novel algorithms to extend pbrt to produce a realistic image of a scene or object. For inspiration, you can either look at the results of Stanford's rendering competition, internet ray tracing competition or SIGGRAPH papers of recent years. I list some papers as options if you do not have a clear idea in your mind.

implement A Spatial Data Structure for Fast Poisson-Disk Sample Generation or Recursive Wang iles for Real-Time Blue Noise
implement Structured Importance Sampling of Environment Maps
implement Fast Hierarchical Importance Sampling with Blue Noise Properties
implement Efficient BRDF Importance Sampling Using A Factored Representation
implement Wavelet Importance Sampling: Efficiently Evaluating Products of Complex Functions
implement Metropolis Light Transport algorithm
speed up the final gathering of photon mapping by Christensen's method
implement Fast and Detailed Approximate Global Illumination by Irradiance Decomposition
implement Photon mapping approach for the volume integrator
extend pbrt to handle L-system, refer to papers from Dr. P's lab.
extend pbrt to handle hairy objects. Refer to the following papers: Rendering Fur with Three Dimensional Textures, Fake Fur Rendering and Light Scattering from Human Hair Fibers.
extend pbrt to handle subsurface scattering, please refer to the following papers: A Practical Model for Subsurface Light Transport, A Rapid Hierarchical Rendering Technique for Translucent Materials and Light Diffusion in Multi-Layered Translucent Materials.
extend pbrt so that it can handle large complex scene efficiently, for example you can use GPU to accelerate it or consider cache coherence. You can refer to Ray Engine, Ray Tracing on Programmable Graphics Hardare and Rendering Complex Scenes with Memory-Coherent Ray Tracing.

Please notice that the ultimate goal is to synthesize a realistic image. Hence, even if you choose to implement a paper or two, please create a scene to demonstrate your result clearly. If you want to do real-time rendering instead, it is fine as well.

Proposal

Please send me your proposal by the end of Jan. 4. You can work in pairs if you want, but your results must justify your team size. In the proposal, please specify your topic, what program or algorithm you plan to implement and references if any. It would help if you can attach some reference images you attempt to simulate.

Submission

You have to turn in every source file you have modified or created and a html report. The report could describe how you create your scene and image. It is fine to import other's objects, such as those available online, in your scene. For the format of the report, please refer to the reports of Stanford's rendering competition for examples. In addition, you have to create an artifact image using your own version of extended pbrt.