Carson Conversations Forum | Roy Taylor
Roy Taylor, Corporate Vice President and Head of Alliances, Content and VR atRadeon Technologies Group AMD, speaks at the University of Nebraska-Lincoln's Johnny Carson Center for Emerging Media Arts. His talk: “GPU = EMA.”
icon search Searchable Transcript
Toggle between list and paragraph view.
[00:00:06.890]I have the dubious honor
[00:00:08.510]of being the person between you and lunch.
[00:00:10.540]So I'm going to try and stay punchy
[00:00:13.130]and I'm going to try and give you
[00:00:14.709]as much specific material as possible.
[00:00:16.910]This time for people
[00:00:18.050]who are putting together
[00:00:20.210]the new program you're looking at.
[00:00:22.930]So first of all, what do I consider
[00:00:24.590]to be emerging arts?
[00:00:26.140]well I actually like some
[00:00:27.140]of the speakers earlier
[00:00:28.140]that said that actually emerging arts
[00:00:29.820]have been going on forever,
[00:00:31.050]ever since the Lumiere brothers
[00:00:32.850]added a sprocket to the kinetoscope
[00:00:34.260]and turned it into the cinemagraph,
[00:00:35.500]that was emerging arts.
[00:00:36.949]So it's been going on all this time.
[00:00:38.949]So what I want to do here is to share
[00:00:41.010]with you some specific ideas that you
[00:00:44.391]might want to add into your curriculum.
[00:00:47.519]So the first thing I want to share with you
[00:00:50.039]is I'm chairman of the VR advisory group,
[00:00:54.809]And this group's pretty interesting
[00:00:55.920]because it's made up half between creatives
[00:00:57.819]and half between technology.
[00:00:59.699]And there recently was a furious debate,
[00:01:01.420]which seemed to me at the beginning
[00:01:02.620]to be not really that important.
[00:01:04.400]And the debate was is VR a platform,
[00:01:07.540]or is it an art form?
[00:01:09.450]And the debate was interesting
[00:01:11.119]because it turned out that we agreed
[00:01:13.640]unanimously actually it was both.
[00:01:15.480]And that was interesting because the games
[00:01:17.649]console is certainly not an art form.
[00:01:19.500]The games are an art form.
[00:01:20.580]The console is not.
[00:01:22.710]And yet for virtual realities
[00:01:24.649]and immersive technologies,
[00:01:26.640]the marriage between the technology
[00:01:28.710]and the art has to be very carefully entwined.
[00:01:31.899]And the reason for that is
[00:01:32.930]because the storytellers and the directors
[00:01:35.439]and the producers can't produce
[00:01:37.979]and tell all the stories that they want
[00:01:40.590]without the technology supporting it.
[00:01:43.039]They are closely matched.
[00:01:44.549]And so I thought that was important
[00:01:46.509]and it has a direct bearing
[00:01:47.859]on an emerging media conference.
[00:01:51.149]So a little quickly a little bit ...
[00:01:52.560]I'm not going to talk about me
[00:01:53.560]because there's a time
[00:01:54.789]and that's not my style,
[00:01:55.789]but a little bit about AMD.
[00:01:57.049]We make chips.
[00:01:58.259]We make microchips.
[00:01:59.259]We make processes, and our processes
[00:02:01.469]are in the PlayStation, and they're in XBox,
[00:02:05.840]and they're in the current PlayStation
[00:02:07.149]that runs VR, and they're going to be
[00:02:08.550]in the next Xbox, which runs VR.
[00:02:11.970]In fact last year 83% of all virtual reality
[00:02:15.391]ran on AMD-based products.
[00:02:18.420]So that's a little bit about the background,
[00:02:20.160]why I feel like we should be able
[00:02:21.410]to talk about this,
[00:02:22.410]but it occurs to me and I'm told
[00:02:25.043]that very often
[00:02:26.080]I may take for granted I assume
[00:02:27.390]that you know the things that we know.
[00:02:29.301]I actually liked that comment earlier
[00:02:31.670]about it shouldn't be in the inside.
[00:02:32.670]You've got to get on the outside.
[00:02:33.970]So the chips that we make relative
[00:02:36.861]to emerging media are used
[00:02:38.760]in three major ways.
[00:02:40.440]The first way is that
[00:02:41.810]in the content creation process
[00:02:43.370]if you're doing pre-visualization
[00:02:45.380]or post production or even virtual production
[00:02:48.030]using a workstation
[00:02:49.310]like this one here from HP.
[00:02:51.350]And we make the processes
[00:02:52.390]that go inside of those boxes,
[00:02:54.060]both the central processing unit,
[00:02:56.630]the CPU and the GPU.
[00:02:58.770]The second thing we're doing is we're inventing.
[00:03:00.340]We're coming up with new ideas
[00:03:02.680]on how to create emerging media.
[00:03:06.550]Two years ago we invented
[00:03:07.550]something called liquid VR.
[00:03:09.400]You will recall that it wasn't so long ago
[00:03:11.870]it was impossible talk about virtual reality
[00:03:13.540]without mentioning motion sickness.
[00:03:16.210]And so we had a brilliant young engineer
[00:03:18.040]called Leila Mar who came up
[00:03:19.540]with something called last latch,
[00:03:20.810]an asynchronous compute, and we put that
[00:03:22.830]in something called liquid VR,
[00:03:24.230]which was immediately adopted
[00:03:25.290]by a Valve and by Oculus.
[00:03:27.700]And today we don't talk about motion sickness,
[00:03:29.430]though that hasn't completely gone away.
[00:03:31.200]And then the last thing we do
[00:03:32.870]is we power the computers
[00:03:34.260]that actually experience technology.
[00:03:35.590]And here's an interesting observation.
[00:03:37.430]The last hundred and twenty two years,
[00:03:39.580]technology has being used
[00:03:40.830]to create the content.
[00:03:42.330]It hasn't been used to consume the content.
[00:03:44.920]The consumption of content has involved
[00:03:46.670]a rather comfortable chair
[00:03:48.240]or sometimes not so comfortable
[00:03:49.980]and a large screen
[00:03:50.990]or a screen in the living room.
[00:03:52.750]And yet now for emerging technologies,
[00:03:54.280]actually you need technology to consume,
[00:03:57.670]as well as to create.
[00:03:59.960]So why do I believe the emerging technologies
[00:04:03.110]and VR are so important to storytelling process
[00:04:05.880]and to the film industry?
[00:04:07.370]I believe that that is the case because
[00:04:12.890]at its best a wonderful piece of movie
[00:04:15.170]or television takes us away from ourselves.
[00:04:18.610]That for the period of the hour
[00:04:20.259]or the two hours
[00:04:21.259]we were watching that,
[00:04:22.259]we are not in our living room.
[00:04:23.260]We are not in a cinema.
[00:04:24.610]We are totally taken outside
[00:04:26.250]of ourselves and away.
[00:04:28.520]And that I believe is why
[00:04:29.640]it's so universally popular,
[00:04:31.170]regardless of age
[00:04:32.170]or culture or background or anything else.
[00:04:34.950]So I was very interested about this piece
[00:04:37.610]of research, which was done
[00:04:39.270]I think by Virginia Tech.
[00:04:41.640]And what they did was they wanted
[00:04:42.720]to measure that immersion.
[00:04:44.100]So here's my first kind of request
[00:04:45.790]as you're thinking about the curricula.
[00:04:47.770]I'd love to learn more about this,
[00:04:48.850]and here's what they did.
[00:04:50.180]They took two sets of students,
[00:04:51.970]and they put them into this room.
[00:04:54.350]Now one set of students were looking
[00:04:55.530]through a traditional monitor,
[00:04:56.530]and the other set of students
[00:04:57.530]were in virtual reality
[00:04:58.860]and then what they did
[00:04:59.860]was they asked them
[00:05:00.860]to identify patterns sequences.
[00:05:03.060]And the set in virtual reality
[00:05:04.150]were dramatically faster than the ones
[00:05:05.910]who were looking at a traditional monitor.
[00:05:07.500]In fact not, only that, but hours
[00:05:10.380]and days and weeks afterwards,
[00:05:11.380]their cognitive recall was also better.
[00:05:14.400]And the cognitive performance continued
[00:05:16.320]to be better.
[00:05:17.720]So we know then that virtuality does this.
[00:05:19.780]I'd love to learn about some more research.
[00:05:21.180]That's a little bit of science.
[00:05:22.800]Now I want to show you an issue in video.
[00:05:24.100]And can we make sure we have the audio.
[00:05:26.600]We've all seen scary movies,
[00:05:28.740]and we've seen some scary television.
[00:05:30.610]And we sometimes jumped,
[00:05:32.310]but I want you to look at the reaction
[00:05:35.070]to the virtual reality experience
[00:05:36.620]for paranormal activity.
[00:05:38.450]So let's just roll this video.
[00:05:41.000]This was put together
[00:05:42.000]by my friends at VR works.
[00:05:43.700]They put this on location at AMC theaters
[00:05:46.030]across America to coincide with the film.
[00:06:17.411](various people screaming)
[00:06:32.900]Now here's the interesting thing.
[00:06:34.060]I know I'm being filmed I am determined
[00:06:37.490]that I'm not going to jump
[00:06:39.090]because I'm going to be filmed,
[00:06:41.120]and I know you're going to see it.
[00:06:42.840]And yet still the sense of immersion
[00:06:44.400]was so strong that I still couldn't stop myself.
[00:06:46.920]And that's what I think is exciting
[00:06:48.730]about this medium.
[00:06:50.290]But, I don't believe yet that we really
[00:06:53.470]understand the medium very well.
[00:06:55.400]And one of the reasons for that is
[00:06:57.500]we always look at new technology
[00:06:59.590]through the prism of the old one.
[00:07:02.760]And that's why I believe the first movies
[00:07:04.120]look like theater plays
[00:07:05.130]and the first VR looks like movies because
[00:07:06.970]that's what we understand.
[00:07:08.430]And so I want to share
[00:07:09.880]this observation with you.
[00:07:12.080]We have in virtual reality the ability
[00:07:14.440]to completely reimagine
[00:07:15.830]what I call the Z or depth.
[00:07:18.099]So we know that we have in this medium
[00:07:20.590]the ability to zoom far far out
[00:07:22.460]or zoom far close in.
[00:07:24.730]And yet I don't believe
[00:07:25.730]we've taken advantage of that
[00:07:26.730]or even beginning to understand it properly
[00:07:29.169]because quite frankly how could we?
[00:07:32.250]We can only see it amongst the prism
[00:07:34.100]of the one that we know.
[00:07:35.100]So we show you a couple examples here.
[00:07:37.790]This is Hololens from Microsoft
[00:07:39.980]using augmented reality to really understand
[00:07:42.740]about the way the blood system flows,
[00:07:45.120]and I thought this is really interesting,
[00:07:46.540]but yet they're not going far enough
[00:07:49.180]because actually you can use this medium
[00:07:51.590]to go down to subatomic level
[00:07:53.840]to really understand the way
[00:07:55.010]our DNA strands work.
[00:07:56.230]And I think we're going to be able
[00:07:58.210]to understand the body in completely new ways
[00:08:00.130]we haven't thought or before.
[00:08:01.300]Let me give you one other example.
[00:08:03.410]It's a popular piece of augmented reality from Epson,
[00:08:06.210]which can allow relatively modestly trained
[00:08:09.520]engineers to do some sophisticated work
[00:08:11.560]using augmented reality.
[00:08:13.470]And yet why couldn't they actually design
[00:08:16.080]or build a Swiss watch?
[00:08:18.300]because we don't think in these terms.
[00:08:20.669]How could we?
[00:08:22.190]And I think that one of the things
[00:08:23.420]that lot of things I know
[00:08:25.420]that I love about students.
[00:08:26.420]So they can ask me what's
[00:08:27.860]more important than VR?
[00:08:28.880]I say students because they're not inhibited.
[00:08:32.599]They're not held back by the constraints
[00:08:34.080]of thinking that we had.
[00:08:35.899]And as they use the medium more,
[00:08:37.430]they're going to find new ways
[00:08:38.639]to explore in it.
[00:08:39.719]Now the other thing I wanted to do
[00:08:42.169]is to share another observation for you.
[00:08:43.610]Something else I'd like you to consider
[00:08:44.610]for your curriculum.
[00:08:46.410]The human visual system is very very clever.
[00:08:50.339]And it's much cleverer than any system
[00:08:52.220]we've designed for a camera.
[00:08:54.300]And the reason for that is quite frankly
[00:08:55.999]our very survival depended on it.
[00:08:58.889]If you can't quickly identify that
[00:09:00.980]as a tiger running towards you
[00:09:02.949]in your environment you might actually
[00:09:05.730]be caught by the tiger and die.
[00:09:07.869]So a human visual system designed like that.
[00:09:10.459]We can take in a huge amount of data
[00:09:12.819]that no camera can replicate.
[00:09:15.579]Yet in virtual reality we can.
[00:09:17.300]We can replicate those environments.
[00:09:19.600]This is a scene from Battle of the Bastards
[00:09:21.570]from Game of Thrones.
[00:09:22.570]And it's an enormously complex scene to film.
[00:09:25.920]In fact the scene cost
[00:09:27.709]more than 10 million dollars.
[00:09:29.540]And then the reason for that
[00:09:30.920]is the director has to simultaneously do
[00:09:33.089]some different things.
[00:09:34.209]First of all he has to continue to keep us
[00:09:36.249]connected to the characters
[00:09:37.329]that we care about in the scene.
[00:09:38.839]At the same time he has to show you
[00:09:40.559]that they're all about to be killed
[00:09:41.749]because they're surrounded.
[00:09:43.180]And then he also needs to be able
[00:09:44.399]to pan out and show you
[00:09:45.399]that you're about to be rescued.
[00:09:47.269]Very very complex.
[00:09:48.269]A large number of camera angles.
[00:09:50.130]In virtual reality we will be able
[00:09:51.490]to take that all in
[00:09:53.439]in a very very natural way.
[00:09:55.180]So I think when we're teaching
[00:09:56.180]about emerging technologies
[00:09:57.410]and immersive technologies
[00:10:00.800]with the students,
[00:10:02.899]understanding the human visual system
[00:10:04.550]and the opportunities
[00:10:05.550]that virtual reality gives you is very important.
[00:10:08.040]And I do work with with Dos Chapman
[00:10:11.050]and NFTS and I am very encouraged
[00:10:13.139]by the students who are completely unafraid
[00:10:14.920]of not having control of the camera
[00:10:16.929]to drive direction.
[00:10:17.929]I think that's really fantastic.
[00:10:20.050]The other opportunity, and this came up
[00:10:21.769]at dinner last night, is audio.
[00:10:22.910]I'll be a little bit controversial to people
[00:10:25.910]who work in audio and I apologize in advance.
[00:10:27.899]I don't think we're well served
[00:10:29.100]by audio in entertainment.
[00:10:31.199]And the reason for that, it's not their fault,
[00:10:33.579]is because the people with the money,
[00:10:35.480]you know the people don't like
[00:10:36.480]who control films,
[00:10:37.480]they have to constrain the budget based
[00:10:40.990]on how the experience is going to be enjoyed.
[00:10:43.600]And they have no control
[00:10:44.929]over the whether the film is being watched
[00:10:47.040]in a two buck chuck, twenty seat cinema
[00:10:48.999]or in an IMAX.
[00:10:49.999]And at home quite frankly most television
[00:10:52.420]audio systems are not very good.
[00:10:53.600]The speakers aren't that great.
[00:10:54.879]So why would you invest
[00:10:56.170]in really really great audio?
[00:10:58.370]And so for that reason we don't get whispering.
[00:11:00.070]We don't get lots of ambient sound.
[00:11:01.540]We get very big environmental effects,
[00:11:03.440]but we don't get very much occlusion.
[00:11:07.279]In virtual reality and immersive technologies,
[00:11:09.519]we know exactly where the speakers are.
[00:11:10.960]They're right on your head.
[00:11:12.670]And so we'll be able to do some
[00:11:13.670]really really cool things.
[00:11:15.100]One of the questions I was asked:
[00:11:17.550]what are the areas that AMD would invest in?
[00:11:19.499]well one area we would invest in
[00:11:20.920]is any students or any body of work,
[00:11:23.420]which makes more better use
[00:11:24.420]of audio in immersive technologies.
[00:11:26.140]Now this is a pretty complex slide,
[00:11:28.619]but it can contains
[00:11:29.619]some really important points.
[00:11:32.870]I know we're tight on time.
[00:11:33.870]I'm not going to go through it all,
[00:11:34.870]but I want to share with you something
[00:11:35.870]that I discovered recently,
[00:11:37.690]which is really really a big deal,
[00:11:39.600]and that is virtual reality
[00:11:41.911]is having an impact,
[00:11:44.800]an unintended impact of changing
[00:11:47.910]the post-production and pre-production
[00:11:49.560]and digital pipeline of the film industry.
[00:11:52.819]And this industry is worth billions
[00:11:54.110]and billions of dollars.
[00:11:55.759]And what I have seen is
[00:11:57.369]it wasn't intentional.
[00:11:59.180]So let me explain for a second.
[00:12:00.800]The first virtuality we started out,
[00:12:02.260]and by the way Hollywood
[00:12:03.540]makes much more VR
[00:12:04.540]than the games industry so far.
[00:12:05.779]None of the big games publishers
[00:12:07.489]like Electronic Arts have produced
[00:12:09.509]any triple-a titles, and yet Hollywood
[00:12:11.709]has reduced up to 300 pieces
[00:12:14.050]of VR connected narrative.
[00:12:15.360]Mainly as promotional pieces
[00:12:17.119]or sometimes unkindly called PR for VR,
[00:12:19.279]but they produce the most.
[00:12:22.070]Now here's the interesting thing
[00:12:24.869]I started out doing 360.
[00:12:26.477]360 VR's great,
[00:12:28.067]and it's time is going to come back right now,
[00:12:29.350]but right now I see it dying on the vine.
[00:12:31.160]And the reason it's dying on the vine is because
[00:12:32.850]you can look around and that's great,
[00:12:34.559]but you can't interact.
[00:12:36.290]And when you've been able to interact,
[00:12:37.869]you don't like being held down like this.
[00:12:39.620]You want to interact.
[00:12:41.259]So movie industry then says we want
[00:12:43.709]to use game engines.
[00:12:45.939]Then something interesting starts happening.
[00:12:47.910]They start using game engines
[00:12:49.170]and then they realize
[00:12:50.959]the quality of this stuff is pretty damn good.
[00:12:54.269]It's not yet at a quality
[00:12:56.709]that the James Cameron's going
[00:12:58.050]to really start using the game.
[00:13:00.429]but it's getting close.
[00:13:01.429]And it's close enough that you can see
[00:13:02.529]that by the time your students graduate
[00:13:06.100]we will start to see film made
[00:13:07.699]using real-time renderers.
[00:13:09.499]And that's huge.
[00:13:11.410]That's really huge.
[00:13:12.410]If you're a wedder, okay,
[00:13:13.480]if your frame store, if you're Deluxe
[00:13:15.899]or Technicolor, that's huge.
[00:13:18.170]It's going to have profound effects
[00:13:20.160]because it's going to make film cheaper
[00:13:22.319]and faster to make and add special effects,
[00:13:24.899]but there are some challenges.
[00:13:25.899]And that brings me onto something else
[00:13:27.009]I wanted to share with you all today.
[00:13:28.170]I'm so excited about it coming along.
[00:13:30.100]We started out with doing post-production
[00:13:32.300]around kind of 1990.
[00:13:34.580]And that's a very well-established industry.
[00:13:36.339]Then around 1996 we started
[00:13:37.379]to use game engines to do pre-visualization.
[00:13:40.370]You could save money and time on set by ...
[00:13:42.629]You know all this.
[00:13:43.759]Then we moved into virtual production.
[00:13:45.009]So that's a digital pipeline.
[00:13:46.220]Okay so, so far so good.
[00:13:48.899]However, how do we get actors
[00:13:50.980]and actresses and beautiful assets
[00:13:53.860]into that pipeline?
[00:13:55.230]And so I'm very very excited
[00:13:57.429]about volumetric capture
[00:13:59.299]and in particular light-filled capture.
[00:14:01.490]And I think these are going
[00:14:02.730]to be explosive areas.
[00:14:05.089]The other thing is
[00:14:06.089]if we're going to build virtual reality worlds,
[00:14:08.579]unless and until we work out how to build
[00:14:11.179]them with millions of people in them,
[00:14:16.019]Right now if you go and play Call of Duty
[00:14:17.549]or a battlefield game, it's about 120 players.
[00:14:20.689]That's all we can get
[00:14:21.689]on a server together at one time,
[00:14:22.899]but we're going to create real worlds,
[00:14:24.370]we need hundreds of millions of people,
[00:14:26.290]billions of entities.
[00:14:28.110]So we're going to need AI.
[00:14:29.110]So I want to show you a couple of areas,
[00:14:31.929]which I think could be really terrific areas
[00:14:33.790]for you to explore from school.
[00:14:35.600]So first thing is light-filled.
[00:14:36.730]So Mr. Paul Nuevic
[00:14:37.730]is a wonderful wonderful man at USC
[00:14:39.799]and now works at Google.
[00:14:41.490]He has done some great great work
[00:14:44.009]on light fields.
[00:14:45.129]And what a light field is
[00:14:46.860]is literally this cage, as you can see
[00:14:48.259]just covered inside with LEDs,
[00:14:51.940]and you take an actor or an actress
[00:14:53.449]and you capture them.
[00:14:54.610]Now all of that is offline today.
[00:14:57.269]They're captured completely lifelike.
[00:14:59.920]Perfectly great for films for offline renderers,
[00:15:02.509]but not yet anywhere where we could do it
[00:15:04.220]in a real-time rendering environment.
[00:15:06.029]But we're going to get there.
[00:15:08.380]And actor and actresses,
[00:15:09.380]which are captured now.
[00:15:10.440]Obama was captured.
[00:15:12.249]At some point we're going to take those assets
[00:15:13.989]and we're going to be able to render them
[00:15:15.180]in real-time and animate them.
[00:15:17.910]Tons of work to do here.
[00:15:19.819]Massive computational advances needed,
[00:15:22.480]but the competition between ourselves,
[00:15:24.339]and our two competitors Intel and Nvidia ...
[00:15:26.420]Boy we really hate each other.
[00:15:28.009]We're going to compete so hard and fast
[00:15:31.749]that we're going to get here
[00:15:32.769]much much faster than you think.
[00:15:34.690]So I'm very excited for any research
[00:15:36.600]into light fields.
[00:15:38.399]Now I want to show you this.
[00:15:39.399]This is a piece of content made
[00:15:40.839]by Remington Scott.
[00:15:42.119]And Remington did all of the animation
[00:15:44.959]for Gollum in Lord of the Rings.
[00:15:47.249]And MacInnis, John MacInnis.
[00:15:48.509]He wrote the script for call of duty.
[00:15:50.769]They use volumetric capture for this.
[00:15:52.860]It is a VR piece of content.
[00:15:54.749]And what I want to draw your attention to
[00:15:56.480]is the quality animation
[00:15:57.730]because this is in a game engine.
[00:16:00.069]This is in Unreal Engine 4.
[00:16:02.589]And you can't see it so well here.
[00:16:04.889]It's running 720p,
[00:16:05.889]but the quality of the animation
[00:16:07.970]and of the piece in a game engine
[00:16:10.059]I think it's really really terrific.
[00:16:12.059]I'd love to see more work done here too,
[00:16:14.850]but now let me talk about something else.
[00:16:19.440]I'm going refer to scripts,
[00:16:20.439]but the scripts I talk about
[00:16:21.600]aren't the same scripts
[00:16:22.600]that you talk about.
[00:16:23.939]When I refer to a script,
[00:16:24.970]it's a piece of code.
[00:16:25.970]It's hundreds of lines of code.
[00:16:27.369]In fact actually often it's millions
[00:16:28.529]of lines of code.
[00:16:30.209]Now here's something interesting as well.
[00:16:34.199]for the game engines
[00:16:36.009]and virtual reality,
[00:16:37.369]but I think it's going to have
[00:16:38.500]a massive massive impact
[00:16:39.540]and it's this.
[00:16:41.139]If you want a character to move in a game,
[00:16:43.779]you have sometimes hundreds,
[00:16:46.059]millions of lines of command code
[00:16:47.769]and that's what I call a script,
[00:16:49.139]which says you know the character model
[00:16:50.649]has to move like this and then move like this
[00:16:52.889]and oftentimes those characters
[00:16:53.959]don't look very realistic
[00:16:55.920]because it's very very difficult
[00:16:57.470]to make these kind of scripts
[00:16:58.559]and command lines
[00:16:59.559]and make them look really really good.
[00:17:01.980]But what's happening now
[00:17:04.250]is deep neural networks
[00:17:06.240]are being used to create AI
[00:17:13.500]that's not only very very natural,
[00:17:16.289]but think about this.
[00:17:19.130]Right now if you're Dreamworks
[00:17:20.709]and you're making How to Train Your Dragon,
[00:17:23.029]there's an enormous amount of work,
[00:17:25.199]which goes into character movement,
[00:17:26.910]making it look realistic
[00:17:28.339]so that we can enjoy the character
[00:17:29.760]and we don't think ...
[00:17:31.570]That's been going on ever since Disney
[00:17:33.260]came up with Bambi.
[00:17:34.960]It's very expensive.
[00:17:37.020]Now how about now if the director
[00:17:38.850]can simply say to Kung Fu Panda,
[00:17:41.190]they can talk to him
[00:17:42.860]and just say walk up the stairs
[00:17:44.419]or climb that rope.
[00:17:46.830]With AI this will happen.
[00:17:50.429]We will direct characters,
[00:17:52.480]and the director will speak to them
[00:17:54.309]as though there are a natural human being.
[00:17:56.789]Again, a ton of work.
[00:17:59.149]Tons of computation to go in here,
[00:18:01.549]but artificial intelligence holds huge promise
[00:18:05.590]for the making of film
[00:18:06.840]and television in the future.
[00:18:08.059]There's another really really great area.
[00:18:10.570]And now let's talk about
[00:18:11.750]some of that difficulty.
[00:18:12.750]And again I'm going to keep speeding up.
[00:18:14.799]This is the opening scene from Gladiator.
[00:18:16.669]I'm sure everybody has seen it,
[00:18:19.260]and this scene is particularly interesting
[00:18:20.690]to me because of course it was very exciting.
[00:18:23.299]Thousands of Romans,
[00:18:24.299]thousands of Goths in a forest.
[00:18:26.600]So lots and lots and lots of trees.
[00:18:28.870]And then as the battle starts,
[00:18:30.240]they fire those arrows
[00:18:31.240]and then they fire the fireballs and so on.
[00:18:33.019]Now here's the interesting thing today.
[00:18:34.790]This could not be made in VR today.
[00:18:38.050]Utterly utterly impossible.
[00:18:40.960]We have what I call the the rock-paper-scissors.
[00:18:44.190]And this is where I'm going to
[00:18:45.190]circle back to the beginning
[00:18:46.680]and give you the complete
[00:18:47.830]kind of touch points in all
[00:18:49.120]of my speech to you today.
[00:18:51.420]If you wanted to make that you just couldn't.
[00:18:53.610]And so what would happen is
[00:18:56.889]the producers and directors are going
[00:18:57.889]to sit down and go you can't have the fireballs
[00:19:00.409]or you can do it, but you can't have the trees.
[00:19:04.830]So you can have high quality visuals,
[00:19:07.370]lower frame rates, less entities,
[00:19:09.730]or more entities, but it's Rock-Paper-Scissors.
[00:19:11.769]You can't have all of those things.
[00:19:15.429]Now if you're making a piece of content,
[00:19:17.180]that's planning to come out
[00:19:18.240]in three to five years,
[00:19:20.020]you can't do that within the parameters
[00:19:22.660]of current constraints
[00:19:23.740]because things are going
[00:19:24.960]to evolve so quickly.
[00:19:28.170]And so my final point on that theme,
[00:19:30.710]and my conclusion,
[00:19:31.710]I've gone as fast as I could for you, is this.
[00:19:34.779]It is entirely likely that the most important
[00:19:38.860]technological innovations to take place
[00:19:41.190]in the next thirty years
[00:19:42.919]have not only not yet been invented,
[00:19:45.240]they haven't even been thought of.
[00:19:47.690]And one of the things,
[00:19:49.580]even though I'm in my 50s now,
[00:19:51.750]that keeps me as animated as you can see
[00:19:53.929]and as excited as I am
[00:19:55.929]is the fact that if we think for a second
[00:19:59.720]that all the great brilliant ideas have
[00:20:03.360]happened, they most certainly have not.
[00:20:05.570]There is another Google, another Facebook,
[00:20:10.110]There is everything to be invented.
[00:20:12.890]And the reason I gave up my weekend
[00:20:15.019]to come and be with you ...
[00:20:16.019]First of all it's a great honor.
[00:20:17.230]Thank you for inviting me.
[00:20:18.350]It's because you hold the key.
[00:20:21.760]Your students are going to invent
[00:20:23.409]the techniques and the products
[00:20:25.690]and the things that are going
[00:20:26.690]to make all this come true.
[00:20:29.880]So thank you very much, thank you.
Log in to post comments