Very amazing, but Ira is bald
wonder how much overhead realistic hair would add.
Accurately modeling and rendering the ocean is highly complex from a computational perspective – but it’s considerably harder to portray accurately an artificial simulation of the human face, according to Nvidia CEO Jen-Hsun Huang in his opening keynote address at the GTC 2013 conference in San Jose. Human faces contain a …
Technically, a FLOP is a floating (point) operation, the plural of which is FLOPs. It is legitimate to say "To compute this thing will take 2 TFLOPs" meaning 2*10^12 floating point operations. It's one of those things that does require a bit of context to parse.
That said, I agree that in the context used, TFLOPs means "teraFLOPs per second", but the other usage does appear in the literature.
Early days still, its good but not yet totally convincing. But I suspect you will see these happen in our lifetimes:
Hollywood "stars" will be scanned and sampled, and contracts will mention not just "use of likeness" but animated likenesses etc.
Thirty years from now Justin Beiber et al will be 'virtually' appearing to promote products, appealing to their aging demographics sense of nostalgia for the 2010s. They will have signed away their likeness rights early in their careers while up-and-coming.
Future movies can have fun with mashups, casting actors from different generations together.
Whole body versions of similar tech will be used for porn.
It will become possible to fake compromising footage of famous people, and this will be used for counter propoganda.
I think most of us figure Hollywood will get the most use out of it at first. At the very least, when it comes to final render, they don't have to worry so much about realtime rendering; it's the facial accuracy they'll want more than anything. This level of facial detail is perhaps still rather too complex for use at the consumer level, but it shows a tantalizing hint of the future--say, 7-10 years down the road.
This post has been deleted by its author