Magazine article American Cinematographer

Pygmalion Tries Again for Weird Science

Magazine article American Cinematographer

Pygmalion Tries Again for Weird Science

Article excerpt

Weird Science may have a slightly outrageous premise, but sometimes one must look beyond the surface to find the shine. The notion that a computer has the capacity to create a woman is not so far-fetched. Entire worlds are created by computers. Graphics and animation have become so advanced it is often difficult to distinguish between the real and the imagined.

"Lisa" was created for this film by two boys who put their computer to the test. But actually, she was born on video, by the special effects experts at Video image, who not only put their computer to the test to simulate a frill human body, but stretched their imaginations to devise a vehicle through which Lisa would travel from computer to reality without spilling disbelief all over the screen.

Richard Hollander and John Wash, technical and art directors (respectively) at Video Image, oversaw the Weird Science project. In creating the woman, they started out by documenting a mannequin originally to be used as Lisa 's prototype. Hollander, known for his work in such films as Blade Runner, 2010, Brainstorm, was responsible for designing a data base capable of producing this three-dimensional form. In order to accomplish the feat, he first had to obtain all the information required for translating a 3-D object into computer language, a process known as digitizing. According to Hollander:

"When you have a shape that can not be modeled as a regular solid (such as cubes, circles, etc) or if no one has plans that are cross sections of the object on an incremental basis then one of the best ways to get it into the computer is just to go and digitize the surface of it."

Before entering the computer world, Lisa had to exist in ours. Well, half of her did, anyway. After finding store-bought mannequins unrealistically proportioned, a full-scale model of half a woman was specially sculpted by Mike Hosch for the task of digitizing. "We took the liberty of stating that the left hand side was identical to the right hand side, only mirrored, and so it saved us time," Hollander states. "Most people are not built this way, but for the graphic case it wasn't necessary to have two distinct sides. The film wasn't based on that kind of realism."

After the model was completed, the labeling phase of the project was put into action. Numbered points were needed to feed the computer the data for digitizing. With assistant John DesJardin, Hollander began what would become a two-day, non-stop, polygon plotting orgy.

"We had the whole body smoothed so we could write on it," Hollander recalls, "We went in and marked points with a one-colored pencil, that represented vertices of polygons, all over her body. We put these points in such a way as to increase the number of points per square inch where it was very curvy, and decrease the number of points per square inch where it was not so curvy. More data is needed around the curves, but if it's a big, flat area, only three points spread apart are needed to define that area. Every time we made a point, we also numbered it ... a little small number, right on the model."

In order to conduct the process as efficiently as possible, the model was subdivided into sections, each section assigned points and corresponding numbers. The points, or vertices, were then connected into triangles, with each triangle receiving a number. When all labeling was completed, the model was taken to a milling machine, where a small metal pointer placed on the chuck of the mill acted as reference.

DesJardin moved the model in various positions, and used the digital readouts generated by the mill to get 3-D coordinates in space. For every vertex labeled on the model, there were three numbers (the x, y, and z coordinates) recorded for use in the data base. The model was positioned to move in all directions, allowing full control and the ability to obtain any desired point in 3D space.

All labeling information was then recorded into the database, which enabled the computer to "read" the three-dimensional object and translate it into a language the viewer could understand . …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.