|
Comp 238 is UNC's advanced image synthesis course. It covers advanced
topics in rendering, including stochastic ray tracing, procedural shading,
radiosity, and image-based rendering. Topics from the current research
literature are also covered. Class assignments appear below. Click the
links for more information.
|
|
|
Programming Assignment 1: "Classic" Ray Tracer
Implement a "classic" ray tracer like the one from Turner Whitted's 1980 paper,
"An Improved Illumination Model for Shaded Display." It should support spheres
and planes as primitives, as well as point light sources. The surface model should
be the one from the paper. Specifically, it should implement recursive reflections
and refractions.
|
|
|
Programming Assignment 2: Stochastic Ray Tracer
Implement a stochastic (i.e. distributed or distribution) ray tracer, similar to
the one described in "Distributed Ray Tracing," the 1984 paper by Cook et al.
It should include antialiasing, soft shadows, image texturing, and blurred reflection.
Extra credit for motion blur and depth of field. It should support triangles and spheres.
|
|
|
Programming Assignment 3: Light Field Viewer
Implement a light field viewer. It should have a movable focal plane,
at the very least one that is parallel to the front plane and slides back
and forth. Begin with a dataset made with your ray tracer.
|
|
|
Programming Assignment 4: Procedural Shading
Experiment with procedural shading. This assignment is very open-ended. Exercise your
creativity. You can do offline rendering, or hardware-accelerated rendering. Possible
renderers include PRMan, BMRT, ATI 9700 hardware, 3DLabs P10 using OpenGL 2.0, and
NVIDIA's Cg on the NV30 simulator.
|
|
|
Final Project: Light Field Reconstruction
Common light field rendering methods such as quadrilinear interpolation of
a "two-plane" light field are prone to aliasing unless the sampling
density of the camera surface is relatively high (increasing the size of the
light field) or unless the source images are filtered
(increasing blurriness in the reconstructed images). This project attempts
to reduce aliasing when the light field is undersampled.
My write-up later became an EGSR paper.
|
