Ideas for papers and projects

CS 99D - The Science of Art
Winter Quarter, 1998
Marc Levoy
Handout #2

Here are some ideas for your final project. Some of the ideas in the writing category are simple enough to tackle for one of your short papers, in lieu of answering one of the questions handed out in class. The items about Vermeer and Masaccio might fall into this category. Obviously, I expect different levels of work if you use one of these for a short paper assignment versus using it for your final project.

Programming

• Write an artist's aerial perspective assistant. First, using a standard interactive graphics toolkit, implement a color picker for choosing foreground colors. Use the RGB, YIQ, HSV, or HLS system from Principles of Computer Graphics by Foley, van Dam, Feiner, and Hughes. Alternatively, invent your own system for picking colors. Next, provide a set of sliders for choosing the distance to the horizon, elevation above sea level, atmospheric humidity, the smog or dust content of the air, and any other parameters you can think of. Finally, implement a simple mathematical model of aerial perspective. Don't worry about getting the science right, as long as the results look plausible. Use your model to compute and display a sequence of colors that achieves the desired visual effect.
  
  
• Program a tone reproduction operator that squeezes the large dynamic range of a synthetically rendered image into the limited dynamic range of a particular artist's media. Use a canned renderer to generate the images, Try your operator on hard scenes, like a moonlit scene or a sunlit scene with deep shadows. For extra credit, implement an operator that employs local contrast effects to extend the dynamic range of the media, as artists do. See me for some references to the graphics literature.
  
  
• Write a Web tutorial suitable for teaching Stanford art students how to draw objects in correct perspective. Concentrate on architectural objects if you like. See me for some sample handbooks of perspective drawing.

Artistic

• Model and render the 3D scene corresponding to a famous painting. Use a commercially available modeling and rendering system such as SoftImage or Ray Dream Designer. For an extra kick, script, render, and record on video a flyaround of the scene. Alternatively, convert it into a form that can be navigated interactively on one of our high-performance graphics workstations. Beware: many paintings do not have a single, consistent 3D interpretation!
  
  
• Write a RenderMan shader that implements a nonphotorealistic rendering style. Although not really programming per se, this project does require some understanding of programming language constructs. For some fun, try to mimic the style of a famous artist. Can you draw pictures that exhibit Leonard's sfumato, or Durer's hatchings? How about Monet's brushstrokes? This sounds easy, but it's not. I can point you to some recent efforts in the graphics literature. This could easily turn into a senior project, or even a PhD thesis!

Writing

• The recent cleaning of the Sistine Chapel ceiling suggests that Michelangelo, rather than employing the Renaissance discovery that shadows should be dark, tending toward black, instead adhered to medieval tradition, in which shadows were fully saturated with color. If correct, then this radically changes our view of Michelangelo and his art. If, on the other hand, we have cleaned away Michelangelo's intended tones, as some critics claim, then we have ruined his masterpiece forever. Read one of the many books about the cleaning. Then read James Beck's scathing condemnation of it in Art Restoration: the Culture, the Business, and the Scandal. Who is right?
  
  
• Trace the scientific and artistic history of some optical theme that we did not cover in class, such as motion blur or depth of field. If you choose motion blur, don't forget to talk about its use in cartoon animation. If you choose depth of field, don't forget to talk about the invention of photography and its effect on our expectations about art. This topic will require some digging into the literature.