If you're planning to work on graphics as an independent study or
independent study transforming into a thesis with me this semester at
Williams, you need to get my approval of your syllabus before the
semester starts. If you don't have my approval, I will not agree
to the independent study. I require this because the syllabus is a
contract between us. Without it, we don't have an agreement of what
is expected of you, which puts you and your grade at great risk. I don't let students
take on risks like that!
You know what a general course syllabus looks like
(e.g., http://graphics.cs.williams.edu/courses/cs371/f10/schedule.html). This
doesn't have to have that level of granularity. It should be a two-page description showing me that you've thought about this enough to have a reasonable plan and understand the commitment of your project. This should include:
- What is your project?
e.g., "A new BRDF model for fruit"
- What is the prior work and state of the art?
- Give a one-paragraph summary of the most recent paper(s). You can usually find this out by finding SOME paper and then following the "cited by" references forward in time in the ACM digital library.
- List scientific citations for what you think are key research and survey papers, and books and chapters. The related work section of the most recent paper will help you here.
- Your best estimate of current industry practice, and how you determined that. SIGGRAPH courses and "talks" and Eurographics "State of the Art" (STARS) reports often discuss this, e.g., "Advances in Real-Time Rendering 2011"
- How will you evaluate your algorithm/system/theory?
- For absolute error
e.g., "I'll 3D scan fruits, place the real ones in a Cornell box, and then measure the difference between photographs and rendered images."
- Against prior art
e.g., "I will reproduce Fig. 7 and Fig. 9 from [Handel09] using data from the Matusik database, and then compare on time by running their code and mine on the same machine."
- A very conservative schedule, specifying for each week:
By conservative, I mean that you should probably quadruple your
time estimates for how long you think it will take to accomplish
tasks. When you work on a project for a regular course, the
professor has already laid out how you will work through each
aspect and ensured that you knew enough material to solve the problem.
In independent work, none of that is done for you. Commit to the
least you think you can get me to approve...I'm not going to complain if you overdeliver,
but my hands are tied if you don't produce what you said you would.
- What you will read (and how you'll update an annotated bibliography or other material so that I know that you did it)
- What you will write
- What you will code (and what you'll show me to prove that it works)
- The criteria I should use to evaluate your work (note that I have to make this evaluation based almost exclusively on what you present at a weekly meeting)
- Resources that you need
- e.g.,- Computers, libraries, compilers, data, equipment, physical space, disk space, my time, permissions from various campus bodies
- If working with other students, explain the division of responsibility and how I should evaluate each.
- What you will submit at the end of the semester. For the thesis, note that there are some required documents.