Check out Stanford's 3D fax Project.
More about the Michaelangelo Project
Modeling from Laser Range Find Data
Whereas IBMR works from two dimensional images, laser range scanning techniques directly record 3D. A sensitive laser beam sweeps across the object to be modeled, say a building or a sculpture and the computer captures every data point recorded. The level of detail essentially depends on the resolution of the laser.
"Gathering range image data is similar to taking a photograph, only instead of recording a color for each pixel, we have a depth instead."
Brian Curless, currently Associate Professor of Computer Science and Engineering at the University of Washington, completed his doctoral dissertation at Stanford last year on laser range scanning research. "Gathering range image data is similar to taking a photograph, only instead of recording a color for each pixel, we have a depth instead," he notes. "We can then take multiple points of view and blend or merge them together into a single surface representation."
Replica of Happy Buddha generated from a digitized video image converted to a 3D model
Courtesy Brian Curless and Marc Levoy, Stanford University.
Next fall Curless will take his laser scanning know-how on the road to participate in the Michelangelo Project sponsored by Stanford University. Many of the great sculptor's statues in Florence and Rome will be scanned and recorded for posterity. "There are times when it's just not practical to make a physical copy of something (e. g., from a mold.) This way we can make a digital copy and disseminate that information to artists and scholars around the world," says Curless.
Relatively straightforward views can be scanned from a floor-mounted laser rangefinder. However, more intricate details, such as hidden cracks or crevices, can be measured with a smaller arm-mounted laser scanner. The laser armament can then be moved to a variety of positions which records x, y, and z coordinates of the laser's position along with its roll, pitch, and yaw orientation in space, which is needed to correlate the different views. Other applications include modeling objects too complex for standard techniques, for instance, a space alien. "Sculptors have traditionally mastered the 3-d interface of clay," continues Curless. "Shapes and forms which would be difficult to design with a computer keyboard and mouse, can be rather efficiently done in clay and then scanned, transforming it into something we can use for computer animation." The scale and level of resolution is limited largely by the narrowness of the laser beam, with the sizes of the object to be modeled ranging from tens of millimeters to building-sized structures.
And as far as accuracy goes, "Terminator 2" producers needed a computer model of the bad guy's face for one particular effect. Actor Robert Patrick's head was laser scanned and the model morphed into a blob of metallic goo. Laser scanning reduced the complexities of modeling a recognizable human face down to a trivial task. According to Curless "we are seeing more and more synthetic actors -- actually digital stuntmen -- doing things that wouldn't be possible for live actors." Industrial applications abound for laser scanning as well, particularly in the field of Computer Aided Drafting and Design (CADD).
Curless says "often times architectural plans are changed during construction and the CADD plans are not revised, what we call 'as-builts.' Laser scanning allows for a quick update of the computer plans. Or sometimes you might have a design that was created pre-CADD. For example the 777 aircraft was built entirely inside a computer. But if you need specifications for a B-52, none may exist in CADD and for purposes of machining spare parts, it's handy to have a digital blueprint." "And besides," says Curless, anticipating his trip to Italy, "we'll have a lot of pretty interesting 'as-builts' in Florence, with all the complex friezes and facades of the buildings."