Issue link: https://www.e-digitaleditions.com/i/433760
THE CAPITOL DOME 38 Fig. 14. Recreation drawing by the author showing the exterior of the roof of the South Wing. bits of information. For example, all of the design documents depicting the vault between the interior entablature and the exterior wall show a segment of an arch. In two detailed topographical drawings of the Capitol's ruins (1815), one by Latrobe himself and one by his stone carver Giovanni Andrei, essentially showing the as-built condition, the vault is depicted as a barrel vault of a five-foot diameter. Sometimes a dimen - sion noted on a drawing might be countermanded later in an informal letter, or a decision may be referred to obliquely, or a design changemay occur as two drawings jump in scale. In some cases I simply could not determine a detail or dimension, in which case I would design an element based on a similar - example by Latrobe, maintaining the spirit of his intentions. My actual synthesis began by creating a computer model of the architecture as pure geometry. Two things make these geo- metric pieces look real. One is the texture they have, and the other is the lighting of the computer scene. I created realistic textures using an imaging program and projected them onto the geometry. Projecting the image onto geometry is called texture mapping, which can make a simple gray cube look like a block of sandstone, for instance. Lighting is the crucial next step, as the lighting of a large interior space with windows, skylights, deflected light, and reflected light is quite complex. When a final scene involving geometry, texture mapping, bump mapping, and lighting is assembled, it's viewed through a software camera that has all the attributes of a real camera, either a still camera or a moving picture camera. e camera, following motion picture standards, can be tracked, panned, dollied, orbited, or zoomed, to achieve the desired composi - tion. Once a camera composition is established, and many set- tings are fine-tuned, the scene is rendered as a still image or as an animation—now all the data regarding geometry, material, and light is synthesized based on a camera angle and lens open- ing by the computer's CPU and the computer's memory. e rendering process is roughly equivalent to photographing the final scene, but neither the camera nor the scene actually exists. Camera renders for stills are usually set at larger resolutions, and renders for animation are set at video resolution, knowing that the effects of perception and motion blur will communi - cate a sequence of frames into believable motion. AU T HOR