IRA FLATOW, HOST:
Joining us now is our multimedia editor Flora Lichtman who is - welcome to SCIENCE FRIDAY, of course.
FLORA LICHTMAN, BYLINE: Thanks.
FLATOW: We'll have a backend. We have our Video Pick of the Week today, sort of an oldie goldie, right?
LICHTMAN: Yeah. But it's related. So we thought we would bring it up. It's a hairy quest scientific question. I'm just going to fill it with bad jokes. It's about the physics of Rapunzel's hair, is the topic of this week's video pick. And we talked with Kelly Ward, who's a senior software designer for Disney, and she had studied hair modeling in school. And I didn't even know this was a field you could study, but it just sort of gets to what we were talking about - become really good at something. And then she found out that Disney was making this movie about Rapunzel, and it was like the dream fairytale ending for a student of hair modeling.
FLATOW: So it was the film "Tangled" and she - and in your video pick, she talks to you about how she had to mathematically model the hair and different...
LICHTMAN: It's a - yeah, it's a really complicated question, it turns out. So we have over 100,000 strands of hair on our real heads. In this movie, they look at - they sort of boil it down to over 100, but even that is a really complicated question. I think Tony DeRose could probably talk about this better than I could. But, you know, it's the problem that they bump into each other and the static. I mean, how do you think about this problem of hair movement?
DR. TONY DEROSE: As you said, there are about 100,000 hairs on a human head. There are many more on an otter.
(SOUNDBITE OF LAUGHTER)
FLATOW: Is that why we don't see many otter movies?
DEROSE: That's why we don't see many otter movies - yeah. But if you can capture the essential motion of about 100 hairs or so or 200, then you can kind of do in-betweening for the hair and get the basic motion at dramatically lower simulation cost.
LICHTMAN: And, you know, I mean, that's the one thing that Kelly mentioned, that it takes a long time to render this kind of stuff, right? Can you give us a sense of how long it takes to actually output a shake of the head?
DEROSE: Well, the outputting isn't so time-consuming, but doing the light simulation of how light interacts with all the hair and how the hair shadows each other, that's extremely computationally intensive.
FLATOW: Give us - how long for a second or two of - Steve, how long would it take?
DR. STEVE SULLIVAN: It could be overnight for a single frame.
FLATOW: Overnight for one frame.
SULLIVAN: Yeah. Depending on the creature and what's going on. Another interesting aspect of this is why we have to get the physics right so it's kind of believable. We have to then put controls on top so animators can make the hair perform, because what you see in the Rapunzel footage is that hair is doing something that real hair couldn't do. It needs to express something about the story there. So a lot of the challenge for our technologists is to come up with interesting controls for the animators to create performances out of physically-based things.
LICHTMAN: Yes. This is exactly what Kelly Ward said. If you actually had Rapunzel's hair, you'd be dragging around 80 pounds.
LICHTMAN: You wouldn't get that.
FLATOW: I'm Ira Flatow, and this is SCIENCE FRIDAY from NPR. Who wants to talk about the uncanny valley? Anybody? Steve?
JAMES CAMERON: Oh, boy. The uncanny valley, that was the terrain that we had to get across in order for "Avatar" to be a success. You know, I always said, you know, we're going to try to catapult ourselves across the uncanny valley and land with our fingertips embedded in the opposite cliff, try to pull ourselves over. That was the challenge for the whole team on "Avatar."
FLATOW: And what was the big challenge? Why was it so changeling?
CAMERON: Well, all the main characters, you know, or most of the main characters in the film were going to be done completely with CG. So we had to really feel like in order for us to have the correct emotional impact on the audience, the audience had to believe that the creatures that they were seeing, which were largely humanoid, were real. They couldn't be thinking that it was good CG, or we'd be doomed. And, you know, so we had the problem the so-called deadeye syndrome that we had to get over unfortunately.
You know, Weta Digital at the time was one of the leaders in creating a very kind of liquid beautifully refractive eye. And I always felt that the problem was around the area of the mouth, the kind of neuromuscular interaction around the mouth that forms the expression is so complex. And I always felt that that's where most CG characters broke down, not around the eyes. So I actually encouraged them to spend a lot of time and energy in that area as well.
FLATOW: Yeah. Because if you don't get stuff like that right, it sort of comes out creepy looking, doesn't it?
(SOUNDBITE OF LAUGHTER)
FLATOW: I mean, the audience has a little bit of problem with it. It's almost real but not really real.
CAMERON: Right. Well, the uncanny valley refers to a dip in positive response. It's actually meant to be, you know, kind of an X-Y graphing. On the Y-axis, you've got positive response to what you're seeing. And on the X-axis, it's from left to right. It's as you're approaching looking human. And so there's a point where kind of animated animals and a talking moose and so on at the left side of the graph. They're all fine because they're humanoid to the extent that they're doing human behaviors, but they don't look human.
Then as you start to approach looking human with things like androids and, you know, synthetic rubber flesh on Japanese robotics experiments and even things like corpses where there's no movement, we start to get a very negative effect, so the curve dips. And then it comes up again sharply as you exactly approach the look of a true living human.
FLATOW: One last question before we have to go. Is the story still matter? The story, you know, does the animation and visual effects overwhelm the story so much that the story doesn't really matter anymore? Well, I'll ask Jim. Jim, what do you think?
CAMERON: Well, look, I mean, I think we all know the answer to that question. People respond to stories. And, you know, they respond to human interaction and to emotion. And they want to have an emotional response themselves. And so I think we've seen the big hollow kind of clanking but quite beautiful and quite difficult visual effects extravaganzas that didn't work. And I think we've seen films that were much more modest in their effects goals that did work, you know?
So I think step one is you've got to cast it right. You've got to have a great script. And you've got to be - yeah, and you've got to be telling a well-modulated human story throughout. Then you can start to layer on the visuals and the imaginative elements, you know, to any extent and certainly my stuff has been, you know, with an awful lot of that...
FLATOW: All right.
CAMERON: ...but you still have to have the basics first.
FLATOW: All right. Thank you very much for joining us, James Cameron, Tony DeRose, Steve Sullivan and Craig Barron, thanks to all for taking time to be with us today.
SULLIVAN: Thank you.
FLATOW: We're going to take a short break. We'll be right back after the break. Stay with us.
(SOUNDBITE OF APPLAUSE)
FLATOW: I'm Ira Flatow. This is SCIENCE FRIDAY from NPR. Transcript provided by NPR, Copyright National Public Radio.