A while back, I wrote about the problem with next-gen graphics. They’re improving so much that our brains now start comparing video game characters to actual people — but the uncanny valley effect makes them seem less real than ever.
I didn’t say this explicitly in that post, but I wondered how video games could possibly overcome this problem, given that even $200-million Hollywood blockbusters can’t give us fully realistic CGI characters.
Edward Jay Epstein’s latest Slate column gives one answer: Video games can’t overcome this problem — because movies are still years away from solving it. He writes,
I recently discussed this issue (the possibility of creating CGI movie stars) with Shuzo Shiota, who heads one of Japan’s most innovative digital-animation companies, Polygon Pictures. Even though his company had “digitally reincarnated” a deceased Japanese actor for a video game, he found that the obstacles to creating virtual actors are almost insurmountable. He pointed out that it takes a 90-man team of modelers, renderers, and animators, many of whom are versed in human anatomy, neurology, and kinetics, a day or more to produce just three seconds of highly realistic animation. And even that level of animation can fool an audience for only a limited time—”10 to 15 seconds”—before the illusion is shattered. At present, he explains, “The amount of information that the human expression, skin, and body … require is just too huge for CG animators.”
Reading that, Nintendo’s current strategy seems even smarter than it did before.
— March 8, 2006