Digital Diva.(the making of a virtual chanteuse)(Technology Information)

A scientist's surprising alter ego

Ray Kurzweil is a world-renowned scientist, inventor, and techno-evangelist. But one role he has never played--until now--is that of female rock star. Using motion capture and numerous other computer graphics technologies, Kurzweil is able to become Ramona, a real-time virtual chanteuse who helps deliver his technological message.

Ramona's genesis began at last year's Siggraph conference, where Kurzweil was introduced to the latest performance animation, digital puppetry, and 3D human scanning and modeling technologies. Kurzweil pondered how these tools could be used to present his concepts of "life in virtual reality" in future speaking engagements. Eventually, he decided to create a traveling road show starring a 3D singing, dancing, female alter ego with realistic moves and facial expressions. This real-time computer-generated performance, complete with backup dancers, would be projected onto a large screen behind Kurzweil as he performed.

Helping with the high-profile project, which was developed in Kurzweil's lab outside of Boston, was a team of engineers and artists, as well as a number of companies. Among these were Kleiser-Walczak Construction Company (KWCC), known for its "synthespian" film and television characters, motion-capture vendor Ascension Technologies, graphics card maker 3Dlabs, system developer Compaq, animation software vendor Kaydara, and 3D scanning system manufacturer Cyberware. In all, more than 50 people collaborated in the development of Ramona.

Kurzweil found among his own staff a woman to serve as the basis of Ramona's form. She was digitally scanned at Cyberware in Monterey, California, and the resulting data was imported into Kaydara's Filmbox, then tweaked and animated. A skintight wardrobe was chosen for the photorealistic Ramona, both to emphasize her proportions, and also because tight clothing presented less of an animation challenge than loose-fitting garments.

Using a speech-input feature in Filmbox, the team could control real-time facial animation data, allowing Kurzweil to perform Ramona's singing while on stage. To accomplish this, many 3D facial expressions from Ramona's live model were acquired during the scanning ...