The original stills (after JPEG compression) can be retrieved by clicking on the corresponding thumbnail.
David submitted a 12 frame sequence of still images showing the growth of a city from 1890 through 2000. The city starts as a simple tipi in 1890 (first frame), a simple house in 1900 (second frame), and then grows into an ever-larger arrangement of skyscrapers (frames 3 to 12).
The primary feature of David's submission is the effective use of procedural modeling: all the scenes were completely modeled with a hierarchical algorithm. To illustrate the complexity of the resulting models, the scene for the year 2000 (frame 12) consists of 98,723 triangles in 4,000 meshes, and it took 10 minutes to render on a Sparc20 with 4 by 4 jittered stochastic antialiasing.
The scenes are lit by the sun, with the city at 45 degrees north of the equator, and sunrise at 6:00am and sunset at 6:00pm. The skyscraper sequence (1910 to 2000; frames 3 to 12) starts with a 5:00am sun in 1910 (frame 3) and advances through the day to a 7:00pm sun in 2000 (frame 12). The windows of the skyscrapers emit yellow light, and reflect pure blue; thus, at night, the windows show a distinct yellow compared to the darker walls, as if everyone were working late on their CS 248 project. As the day progresses, the blue light cancels the yellow to yield lighter gray, blending in with the brighter walls.
The city is set on a field modeled by a mesh of shaded green triangles on top of a flat (painted) green background. At some hours of the day, this yields a perfect color match; at others, shading modifies the color of the green triangles significantly. Not rendering triangles to the edge of the screen has the artistic effect of focusing attention on the city instead of the bland background. At late afternoon or night, it also offers the impression of shadows from the skyscrapers.
Texture is applied on the mountains in the background. The small green box on the side of the mountain is actually the house of 1900 (second frame).
Another aspect of David's renderer, which is not reflected by the frames above, is its ability to describe a selected object. That is, when the user clicks on the picture, the renderer responds with a description of the picked object, such as "Alps," "Intersection of Street 3 and Avenue 4," or "Skyscraper, Story 5, Right side, Window."