[MUSIC] Hello, I'd like to welcome Dr. Harry Brenton, director of BespokeVR and immersive user experience design. Harry, you've got lots of experience designing virtual characters for VR. What do you have to think about when you're designing a character? >> Well, there's quite a well known theory called the uncanny valley. Which basically says that as a character gets closer and closer to a human being, you're fine, you're fine, you're fine and suddenly there's a sharp drop-off, where it gets really eerie to interact and to look at the character. And that's one of the most kind of well known, it's not a design rule, and there's still quite a lot in the research that's not known about it. But that's something which designers certainly avoid. So what people tend to do is, dial down the levels of realism. So that they avoid that kind of uncanny zone. But that raises a really important issue about what level of realism you're going for. And then how you match that realism to the way the character behaves. And it's really a balance, like a lot of things in design. It's a balance between matching the way a character looks and the way a character behaves. Whether that's, an avatar or a non-player character. >> So if a character looks very realistic, but actually their movement is a bit clunky and weird, that can cause problems. >> Yes, it can create a mismatch. Also, if you've got an avatar and you look in a mirror and the avatar's got eyes. But you don't have eye tracking, it's always going to look weird, because the eyes are just going to kind of look dead. And you can put scripts on the eyes to make them blink and things. But it's always going to look a little bit weird. And if you look at the oculus avatars that were released, what they do is they have sunglasses on. So they get around the problem of not tracking the eyes by blocking the eyes with sunglasses. And they only show the hands and the neck and the head movements. So they're only representing bits of the avatar that receive direct motion capture. And if you look at some of the interviews with the people that designed them, they're very careful to directly reflect the motion capture data. So they don't add animations or blinks. They just show the animation that's captured. >> So having a more simplified abstract representation of a person can actually make for more realistic experience? >> Absolutely, and I think there are a lot of the lessons from comic strips. And I'm thinking, there's a great theorist called Scott McCloud, is actually by stripping back the levels of realism. You can say, quite sophisticated things. Because the realism doesn't get in the way of say, some movement and some of the tricks of animators of kind of a stylization and exaggeration can come into play. So even though you're making it look more simple on one level, it doesn't mean that you can't have a sophisticated kind of character or interaction. >> So maybe it's time to talk a little bit about interaction. How do you design an interaction with a virtual character? How do you design for social VR? >> I think two of the most important things there are space and body language. So the rule of thumb is we care very little about things that are far away. And we care a lot, and there's lot of our brains that are devoted to thinking about things that are close to us. So how you position a character and how they move in and out of your personal space is very important. And I mentioned a while ago about about the robot dog in the lab. And what's nice there is that the dog goes out when you throw the stick. And then he comes back in again. And there's something inherently pleasing about that kind of exiting and coming back into your personal space. Body language is a lot more complicated than it seems at first. So if I'm kind of talking to you and I'm just looking up there, that's a really strong cue to you that I'm not listening to you or that I'm behaving inappropriately. And the second I do that, that kind of breaks it. So you have to get the body language right. But you have to continually get the body language right. And that's more of a challenge than you think. >> So you've talked about the Oculus avatars and multiplayer social VR. How is working with non-player characters different from working with avatars and multiplayer environments? >> Well, I think multiplayer, so I'm thinking some of the early multiplayer games and experiences like Toy Box, they kind of make it look easy. Because you've got another human there to do that moment by moment reaction. They sell the illusion that this is a plausible human being there. But actually teaching a non-player character those rules, it works in some contexts. So again, with a dog fetching a stick I can create some simple rules, which say fetch the stick, start begging. But when you want to try and simulate slightly more complex social interactions, those rules can break down. Because essentially, the rules hit the limit of what the coder or the designer think people behave like. But that's not necessarily the way that people actually behave. And again, body language is complicated. Because there are things we do that we're not really aware of, that happen subconsciously. And it's very hard to teach a computer subconscious rules. >> So is that one of the big challenges we have as VR designers? It is, I don't think that users will phrase it in those terms. But they'll just see something as odd, or something that doesn't feel quite right. And again, going back to the idea of the uncanny valley, there's something that is just a little bit eerie or a little bit off center about the interaction. But as a designer, having some kind of understanding and these are just theories. The uncanny valley is a theory. But it's a good way to kind of hang your thoughts about it as a designer on. Because then you can test different approaches. And I think we're still quite far off from I think interactions with photo-realistic virtual humans. But I think in the next ten years for sure, we'll get a lot closer to that. >> So that's one of the really exciting things we can look forward to both as users and designers of VR, is really creating fantastic social experiences. >> Yeah, absolutely, and I think specifically, better non-player characters. Because as the massive multiplayer VR world kick in, and I think there's going to be a lot of activity on that next year. People are going to be much more aware of when non-player characters aren't working well. And so that's a really interesting area of research, I think. >> Okay, great well, thanks a lot, Harry, for your time and sharing your experience with us. And I'm sure our learners have learned a lot. So thank you very much. >> Pleasure. [MUSIC]
