You know the Japanese pop stars I’m talking about: they are mass-produced by Mitsubishi; confections of makeup and electronic processing that all look and sound the same except each new one is ten seconds more stylish than her predecessor, like Britney Spears without the staying power. I’m reading William Gibson’s Idoru and in it these marketing creations have reached their ultimate form, completely computer generated. I’m thinking we are damned close to being able to do that. The record companies could save a ton of money just on eye liner and a singer who has mastered the style could continue recording well into her dotage while getting a new face each week.
“Polar Express” showed that we aren’t quite there with creating CGI people but how far away are we? Checking out the geeks at the Renderosity site (first page worksafe but I’d be careful digging further) they still have problems with “dead” eyes, but those guys tend to concentrate on unrealistically large breasts and their women’s faces don’t get much attention once they are generically pretty and expressionless, usually looking lobotomized. And these are stills, not animations, and most CGI animations look pretty stilted. However, is ten years too long to wait before fully CGI idoru come on the market? Will it start, or has it already, with manga superstars?
Well, how different would such a wholly manufactured pop star be from the Archies or from Milli Vanilli?
Most dance pop music is put out by faceless, interchangeable “groups” of producers, session musicians and session vocalists. This isn’t new! Did anyone know or care who the “members” of C & C Music Factory were? Of course not.
They could do it now, but real people are cheaper and it’s considered less creepy to fantasize about. (Even though you have as much a chance of doing Yuna as Britney Spears.) If you had a Pop Star game where you could make your own and have them compete online, that might be worth something. (Random Event! Your idol O.D.s on pain killers. Return to start.)
Since they majorly advertised who really did the singing (in Japan, anyway), she’s different from Date Kyoko - or the Archies. (It’s Koda Kumi in the Japanese version, Jade from Sweetbox in English.)
The difference would be the degree of realism and the degree to which the public could be fooled and see them as real-life trend setters. Nobody thought the Archies were real, much less Milli Vanilli.
Aren’t Hillary Duff and Lindsey Lohan both CGI? I mean, their careers are both entirely manufactured by Disney INC., why would it be surprising if all their images were too?
At this point, we are capable of generating photorealistic people (see the Final Fantasy movive), but it’s prohibitively expensive for a couple of reasons:
There’s a lot of detail. More detail means more time is needed to render each frame, which limits the amount of work you can do.
Most of it is ad-hoc. No one knows how to put together a physics engine to realistically simulate light reflecting off human skin, so a lot of the work would have to be done frame by frame. That’s where the real cost comes from.
Synthesizing the human voice is another task entirely.
