Inside USC’s Entertainment Tech Lab: Erik Weaver on AI, Virtual Production, and the Future of Story

Technically Creative by KoobrikLabs Podcast Transcript

Title:

Inside USC’s Entertainment Tech Lab: Erik Weaver on AI, Virtual Production, and the Future of Story

Orlando (00:05)
Well, thank you for being on the podcast, Erik. You’re Erik Weaver,
head of virtual and adaptive production at the Entertainment
Technology Center at USC. You, to me, are one of the interviews I’ve
been wanting to have for a really long time, because you’re like,
you’re the guy. You’re the guy engaging with as much tech as can
possibly thrown at you, be thrown at you all the time with the
specific remit of making sure that it’s being used in the best and
most interesting ways for
entertainment distribution, entertainment creation, know, creative
ideation. It’s like so it’s really excited to talk to you. Thank you
for coming on the podcast.
Erik Weaver (00:40)
You’re absolutely welcome and you’re correct. I do get to explore a
lot of really cool new and innovative ways to work with the whole
intent of making it practical and usable for the actual industry. I’m
not interested in just theory and other things. I believe in getting
my hands dirty.
Orlando (00:58)
Amazing. mean, and what a what a lucky position to be in. mean,
obviously to do with quite a lot of ⁓ good work on your end. But, you
know, what a great place to be. mean, USC, for people who don’t really
know USC is the kind of soul’s home and it’s where so many of the
industry leaders in entertainment were educated. And it’s got an
amazing creative arts program, film program, all that kind of stuff.
But can you first explain you are at the Entertainment Technology
Center at USC? Can you explain the Entertainment Technology Center and
what your role is there?
Erik Weaver (01:29)
Absolutely. So the ETC is an organization that was born about 33 years
ago at the best of George Lucas. And he kind of saw this whole VHS
versus beta war had just happened. And he thought, I don’t want that
to happen with digital cinema. And so he wrote letters to all the
studios and said, you know, we should come together on certain issues.
And so under the banner of SCA or School of Cinematic Arts at USC,
A nonprofit was formed. We are completely funded by the studios. So I
report to a board. That board is the CTOs of Disney, Warner,
Universal, Paramount, Sony, Epic Games, New Fox, and a couple others.
So we just get to look at what’s coming around the corner and share
our knowledge with the community.
Orlando (02:15)
Yeah, if people don’t remember the VHS beta wars, that the argument is
actually that the wrong, the wrong technology one, or when you look at
it through the lens now, what we realize is the superior technology
doesn’t always win. It’s usually actually the most user friendly
technology to a new generation of people using it. And beta stuck
around is for professionals for years, beta SPs and Digi betas were
the industry standard. ⁓ sorry, not to nerd out on the beta.

Yeah, if people don’t remember the VHS beta wars, that the argument is
actually that the wrong, the wrong technology one, or when you look at
it through the lens now, what we realize is the superior technology
doesn’t always win. It’s usually actually the most user friendly
technology to a new generation of people using it. And beta stuck
around is for professionals for years, beta SPs and Digi betas were
the industry standard. ⁓ sorry, not to nerd out on the beta.
Erik Weaver (02:23)
Correct.
Yes.
Mm-hmm.
Hahaha!
Orlando (02:42)
Betamax VHS Wars, but it was a, it’s really interesting and it’s
interesting that you guys kind of sprung out of that huge, you know,
that huge, you know, technology war that we, we look on fondly, but
most people won’t even remember.
Erik Weaver (02:51)
Yeah, we actually…
Yeah, we actually did the first draft of DCI, or Digital Cinema. So
what you had was a whole bunch of people coming together saying, this
is what the standard should be, that we’re commercial players. And the
studios had to kind of come together. And so we got about 80 % of the
way there. And then we handed it off to SMPTE, or the Society of
Motion Picture Technical Engineers.
Orlando (03:14)
Yeah. And so, I mean, you already mentioned your, was going to say
clients, but actually they’re board members because they’re, they’re,
you know, part of the, the, the support of the ETC. They’re some of
the largest entertainment producers in the world across many different
verticals. How do they engage with you?
Erik Weaver (03:29)
So they sit on a board, they guide our direction, they look at the
areas in which we’re doing research, and they have varying levels of
involvement based on the studio. So I run a grant now that every year,
and I am looking at what is the most cutting edge aspects of
production, primarily. And so right now that has just kind of shifted
from OSVP or Onset Virtual Production, where we did a tremendous
amount of work.
Orlando (03:47)
Mm-hmm.
Erik Weaver (03:56)

Erik Weaver (03:56)
over to ⁓ AI and how does AI really set a foundational future in a
studio system? A lot of the stuff that you’re seeing online right now
is kind of small, really good creatives of one or two people executing
short form pieces. But what we really need to understand is how does
this actually evolve and how do we have professionals on a team?
Orlando (04:22)
Yeah. Yeah, because I mean, you know, specificity, you know, any
professionals are constantly seeking creative excellence and
specificity. And, I mean, things like character permanence that are
still things that people are struggling with, you know, it’s funny
because everybody kind of said when Nana Banana came out and it’s a
little bit like the VHS beta debate is, you know, everybody said when
Nana Banana came out, it’s an Adobe killer. And I’m like, well, then
you don’t understand Adobe because Adobe Photoshop.
Erik Weaver (04:47)
You
Orlando (04:48)
offers a level of specificity around images that yes, now consumers
can play, but I actually look at those new consumers playing at a
slightly more professional level as just more people being slowly
onboarded towards Adobe. know, the degree of professionalism I don’t
think will change. I just think more people will…
have ideas above their station and that’s a good thing for creativity.
We want people to have ideas above their station because it means
they’re reaching. It means they’re trying to get to the next level.
Erik Weaver (05:17)
Yeah, absolutely. We really want to be pushing things forward. And
what you’re going to see is everything’s kind of multimodal as we go
into the future. So you’re going to be using multiple models across
the board because some have different strengths and weaknesses than
others. And or you have a base of expertise that feels more
comfortable with a tool or a different tool.
Orlando (05:38)
Yeah, it’s interesting because I think that, you know, if the last 10
years was defined by disruption to distribution, it seems like now
what we’re talking about is disruption to production. And what’s
interesting is that the disruption to distribution, nobody really
noticed professional. mean, like, obviously, you know, the upper
echelons of companies noticed and, but there was a sort of…
shrugged off as well, you know. I mean, we call it AI slop now. Back
then, we called it useless and pointless empty calories of creator
content on YouTube. You know, like that was that was what we were
calling it then. But nobody really recognized the district, this
disruption that was coming until Netflix and Netflix was slow. They
were using, you know, the big studios, previous work and just
licensing it very standard like like any television distributor would.

shrugged off as well, you know. I mean, we call it AI slop now. Back
then, we called it useless and pointless empty calories of creator
content on YouTube. You know, like that was that was what we were
calling it then. But nobody really recognized the district, this
disruption that was coming until Netflix and Netflix was slow. They
were using, you know, the big studios, previous work and just
licensing it very standard like like any television distributor would.
Erik Weaver (06:11)
Hehehehehe
Orlando (06:29)
And then suddenly they started creating their own content. was like, ⁓
I see something different is happening here. And so it kind of all
came in through the back door and very slowly. It seems like this
transition to generative AI, people are much more aware of much more
afraid of right out of the gate. Are you seeing that as well?
Erik Weaver (06:46)
absolutely see a fear, I’m seeing it kind of shift here. So if you
take a couple years back, you look at kind of the shock and awe that
some of the open AI stuff did, especially around Sora, where there was
a television academy event and people were literally booing folks on
the stage around us. Because in that shock and awe, they thought my
job is done, my career is over, there is no future. And that’s
absolutely not the truth. We are using
like Clarence Major, ⁓ a production designer on our thing. We’re
using, you know, Roberto Schaeffer, a DP to help our framing, our
visuals, or everything. So all these roles are still absolutely
critical in professional grade work when you’re doing it. And so the
year later, I spoke at the most recent Television Academy AI event,
and people are now much more embracing. And they really are seeing how
their jobs will still be there. But secondarily, they’re seeing it as
this whole tool in a new tool set. And when you would have started
with something like virtual production or onset virtual production
with LED volumes, originally we just called that a tool in the
toolbox. But now anything that’s not practically shot, we’re
considering virtual production. So AI, hybrid,
all these other forms, it’s almost a whole new toolbox to solve
problems, you know? And if we, the community, do not embrace these new
tools and this new toolbox, then we’re gonna lose much of the world to
the Silicon Valley types or the types that don’t come from more or
less the same soul as us, where we just love the heart and soul of
filmmaking.
Orlando (08:32)
Yeah, because ironically with those Sora videos, yeah, they were using
AI, but they were probably doing the most amount of work that will,
you know, that could possibly be done. There was so much Sora gets,
you know, a lot right, but there’s a lot it gets wrong. And you had
to, especially those first videos that came out, it was traditional
VFX workflows and pipelines. I mean, it was, you know, yes, that was
used as a tool, but like, I mean, you’re talking about, you know, huge
amounts of

Yeah, because ironically with those Sora videos, yeah, they were using
AI, but they were probably doing the most amount of work that will,
you know, that could possibly be done. There was so much Sora gets,
you know, a lot right, but there’s a lot it gets wrong. And you had
to, especially those first videos that came out, it was traditional
VFX workflows and pipelines. I mean, it was, you know, yes, that was
used as a tool, but like, I mean, you’re talking about, you know, huge
amounts of
Erik Weaver (08:55)
Yeah.
And still when you’re using AI, it’s a little bit like a wild horse
without a bit. You can absolutely get to where you want to go. ⁓ But
sometimes it kind of takes you in these weird journeys, right? ⁓ And
it does need some corralling. It does need some VFX, some traditional
tools. We’re using a lot of Blender and Nuke and other things.
Orlando (09:06)
Right, yeah.
Mmm.
Mm. Yeah.
Erik Weaver (09:25)
in finishing some of these shots right now.
Orlando (09:28)
Yeah, and now there’s just new tools like training loras becomes a
significant part of an already complex VFX pipeline. I always feel
like it used to be that producers, especially VFX and post people,
them getting a new tool or a new piece of tech was, I mean, it’s
every, yeah, so exciting. Yeah.
Erik Weaver (09:48)
Exciting. It was exciting and fairly often, you know, and in this
situation, though, I kind of feel like we had moved from with OSVP to
a F1 car and with AI, the progression is more like a rocket ship. It’s
like you almost can’t possibly know everything involving and changing
any particular time. But yeah, there’s a lot of a lot of change.
Orlando (10:05)
Yeah.
Erik Weaver (10:14)
happening right now and and you know you just got to start finding the
people that you respect or trust so let’s go back to this Laura thing
so I actually really love some of these conversations so when you’re
looking at a Laura model you’re training a thing a character an
environment a background and that’s off of a larger video model right
so for example in the movie or short we’re doing right now called the
bends

Orlando (10:14)
Yeah.
Erik Weaver (10:42)
we had to create ⁓ a blobfish because it’s about a blobfish and
traditional models don’t have that blobfish. Well, we have a problem
here because that data doesn’t exist. So we had to go to research
institutes, find images of the fish, take those and use Meshi to
convert them into three-dimensional models, clean them up, write a
script to reposition them and light them. And I’ll share a video with
you here. And
And then ⁓ train the LoRa. And another trick about training a LoRa
that’s going to be in a small data set or a private one is that you
need a large VRAM footprint. So we had to burst to run pod or mass
compute to get 100 gigs of VRAM for these training models. Some of the
new VO3 and other things will allow you to train on set, but
The challenge here is that as you shift into professional workflows or
characters or people, SAG does not allow you to let that data slip
into a larger data set. So you must have it in a controlled
environment and you must delete it at the end, which is a critical
component.
Orlando (11:51)
Yeah. yeah, I was, I was talking to somebody the other day who kind of
said, you know, actually in this world access to, know, a, a, you
know, a, a constantly evolving with the infrastructure of an AWS or
whatever else that becomes such a huge part of what you’re doing is
understanding what the usage is going to be, what that cost is going
to be that actually, you know, yes, things are a lot cheaper.
Erik Weaver (12:09)
Yes.
Orlando (12:14)
But that doesn’t mean they’re free and they still get very expensive
as these high-end use cases.
Erik Weaver (12:17)
Absolutely not free.
Yeah, in our situation, we had an amazing architect named Con Wilms
from formerly of AWS, who helped build us a zero trust network that
was all on virtual machines. So when the artists would log in, they
would be able to have access to their LoRa models and the different
tools they needed to use and any other data they needed so they could
drop things in really quick. So it’s pretty cool. And you’ll get to
see that.

share the clip with you of that as well.
Orlando (12:48)
Yeah, well, we’ll definitely, I mean, we’ll put the clip up for the
moment. I’ll just, I’ll just kind of say about it. What I also think
is fascinating about, about the bends is obviously this is a blob
fish, but blob fish, whole point of it is that it changes dramatically
over the course. So you need character permanence without too much
character permanence you need, because you’re, you’re seeing this,
this thing turn, turn from a, a, you know, very structured fish.
Erik Weaver (13:01)
Yes.
Correct.
Orlando (13:12)
into just kind of a gel, honestly, on the surface. Sorry not to spoil
it, I think, you know, and it’s really remarkable what you guys have
been able to do with it.
Erik Weaver (13:16)
Yes.
Yeah, no, no, and that’s something that some people might not get
in watching the film originally, but yes, it’s about a transformative
journey of this fish. Right. And so we had to create multiple models,
models at different levels of depth in the sea so that it does look
different at different times. And you might be wondering why, but you
the nail on the head there.
Orlando (13:25)
You’re right.
Alright.
Yeah.
Yeah.
well, it’s very, it’s very cool. How does that workflow look like?
mean, I imagine that you’re still engaging. You’re you start with a
traditional VFX workflow and animation workflow, and then you kind of
look for the moments you can inject AI to reduce cost or reduce
timelines. I mean, that’s that’s the long and short of it to take
something incredibly complex and reduce it down to its L L L C D.
Erik Weaver (14:02)

No, we actually
So we actually started backwards. So ⁓ we did a much more traditional
pipeline in which we did storyboards. Then we built ⁓ a custom shot
grid integration. And then we did what’s called a genjam. And in the
genjam, we got everyone around a table, including AI artist, the
consulting DP, the writer.
Orlando (14:09)
Okay.
Erik Weaver (14:27)
Kayan Kamm who did the AI infrastructure and tech piece and the
ShotGrid team and we basically broke it down. And in that breakdown,
it was eight scenes across 86 shots. And so each person was handed
their components and I’ll actually show you a video here so you can
see exactly how they did that or you can see the character consistency
to the models. So now.
Orlando (14:51)
Great.
Erik Weaver (14:52)
When they each log in they have access to multiple models. They have
all the loras They have all the pre-chain data and they try and get it
as close to the previs or animatic slash Story that they have right
and so you’re going AI first and then from there we’re doing a
layering of things on top of that which are things like ⁓ debris or
silt
a couple other things for image quality. So we’ve got, then we had
that whole layer of aspects. The last part we have is in this
particular film, we’re getting very sophisticated. One of the major
holdbacks I constantly hear from the studios are the quality isn’t
good enough, right? So what you’re traditionally getting out is
roughly 720p at 8-bit color.
That’s not acceptable if you’re gonna go into any kind of real post
workflow. So we have created an entire pipeline or workflow simply to
get us to 32-bit EXRs. And it is not a simple one button solves
everything. It is step, step, step, step, process, step kind of aspect
to get to that in deliverable.
Orlando (15:58)
Yeah, kind of similar to
the, you know, I mean, I didn’t like a lot of these, the 3D
conversions that were hot for a minute. wasn’t just press a button.
There was still a bit of Roto to do and a bit of things that that
required attention. I still I think I never saw a truly good one. No,
But yeah, I mean, even, you know, uprising, you know, something like I
know.

the, you know, I mean, I didn’t like a lot of these, the 3D
conversions that were hot for a minute. wasn’t just press a button.
There was still a bit of Roto to do and a bit of things that that
required attention. I still I think I never saw a truly good one. No,
But yeah, I mean, even, you know, uprising, you know, something like I
know.
Erik Weaver (16:06)
Yeah.
I wasn’t a big fan either. I wasn’t
Orlando (16:24)
Back to the Future is, you know, they’re doing a thing in IMAX. You
know, mean, people would be shocked to understand how tricky that is.

I mean, it’s not going to the full 70 millimeter, but it is being up-
res’ed, and there’s a lot of work that goes into just taking that and

blowing it up like that. There’s a lot of hand work that needs to be
done.
Erik Weaver (16:32)
Yes.
Yeah, it’s actually quite amazing if you look at the Wizard of Oz
pipeline to the sphere. It literally took Ben Grossman’s team a year
to execute that. A very hard work. So yeah, it is not a simple thing.
It is a whole workflow unto itself.
Orlando (16:49)
Yeah.
Mm. Yeah.
Yeah. I mean, so I mean, in terms of like, you know, you know, your
overarching mission, virtual and adaptive production. mean, I imagine
that do do just simple production workflow things fall into that does,
you know, budgeting or or, you know, machine learning on variance
between expected bid budget versus actual and.
know, ⁓ pipeline work and scheduling and day of days, you know, and
things like that. Does that level of stuff come into it or is it only
when a production moves into a, you know, virtual environment or VFX
environment or, know, or, or, yeah, yeah, sorry. I didn’t really end
that question. It just kind of filtered out.
Erik Weaver (17:38)
So when we do these films, do…
No problem. So when we do these things, that is a of a little bit of a
challenge for us. I don’t have a pipeline team per se. So I have to
build everything from scratch. So we do a lot of software integrations
and custom API builds between things. I just released the Europa
workflow, which I’ll share with you, which is quite sophisticated.
It’s a cloud first.

shot at Sony Stage 7. These pipelines get exceptionally complex. We do
evaluate everything from some aspects of budgeting. We don’t get into
the nitty gritty there, but we do use all the traditional tools
wherever possible and we build continuity between the tools. We also
really care about metadata nowadays. Analyzing the data and metadata
is…
Rebel Fleet helped us on the Europa project in which we grabbed every
single bit of metadata we possibly could and then fed it into a master
system. then instead of doing a sidecar or embedded file, we now have
a database that can only send or will only send the data that’s needed
to the different departments as it’s needed. For example, on that one,
we were also using
anamorphics, which is very difficult on an LED screen. So we had quick
lenses that were custom mapped. And with all that data, we were able
to fix that in VFX.
Orlando (18:59)
Yeah.
It’s funny, even with all of these tools, my face still lights up the
most with talk of anamorphic lenses and in-camera VFX and, you know,
and somebody make, you know, you know, people making a model of a head
and, you know, hitting it with a hammer and blood flying everywhere.
Like that’s the, you know, that’s the fun stuff. And I mean, do you,
you know, is, is it part of your mission at all to be
Erik Weaver (19:12)
Ha ha!
Yeah.
Orlando (19:30)
integrating with old technologies. I only say, you know, I don’t mean
that to sound pejorative, but I mean, these are the technologies that
DPs are most comfortable with and that production designers are most
comfortable with is part of the remit to make sure you’re still
speaking that technological language and bringing those tools that
they know very well up. Or is it kind of a listen in many cases, you
got to forget what you knew and talk about what you can know now.
Like how does, how do you balance that?
Erik Weaver (20:01)
So yes, so we are really trying to bridge the gap of tomorrow. And in
that, a lot of it is bringing traditional things like scripting, sync
on set, some of these other tools, ShotGrid, like I said, that create
traditional framework or structures around things that people can
classically understand. If you find something that’s radically new and
completely revamped something, sometimes we consider it, sometimes we
don’t.

So yes, so we are really trying to bridge the gap of tomorrow. And in
that, a lot of it is bringing traditional things like scripting, sync
on set, some of these other tools, ShotGrid, like I said, that create
traditional framework or structures around things that people can
classically understand. If you find something that’s radically new and
completely revamped something, sometimes we consider it, sometimes we
don’t.
Orlando (20:13)
Mm-hmm.
Erik Weaver (20:26)
But if you think about a tool like, for example, Unreal, Unreal is a
fairly new tool into the environment. It’s only been really there in
the last six, seven years, possibly, for this field. But sometimes
there’s something innovative enough that you do kind of have to
embrace it. If you look at a piece like we’re working on with this
area called Chickaboom, we tried to look more at ⁓ utility AI than
traditional generative AI in that point. was an animation piece.
And so what you’re also seeing is a huge transformation in the tools
currently in the pipeline and how they work. So finding the best
different aspects of these new tools that will be able to replace or
slightly change the pipeline to make them more efficient or effective
is something we’re stretching.
Orlando (21:16)
Right. mean, and, you know, talking a little bit about, there was so
much you spoke about Unreal Engine there and, obviously the
Mandalorian came out. That was, you know, a great, you know, you know,
showcase for what the volume could do. It was very exciting. Everybody
was like, this is the future of production. And then Generative AI
comes out. Now everybody’s kind of forgotten about the volume even as
a tool.
And, you know, colloquially, I certainly don’t mean that producers
have forgotten about the volume in a practical way, but are you
finding that the generative AI is pulling a lot too much focus away
from that? Because that seemed all by itself like a absolute game
changer.
Erik Weaver (21:56)
So it is shifting a lot of the focus, absolutely taking the shine
away. But what you find if you’re really looking at professional
workflows, there’s a bunch of tools and lessons that are critical that
we learned from virtual production or on-set virtual production, like
the VAD, for example. I can now create ⁓ VAD scenes or virtual art
department scenes that may be gray boxed or simple. I can be using a
mixture of
different AI tools to build that or simplify that. I can then
retexture with AI. I can then shoot. can convert my plates through
basically a restyle or restyling the videos. So there’s a lot of
things that you can still use within that toolbox and processes that
will help benefit the overall production.

different AI tools to build that or simplify that. I can then
retexture with AI. I can then shoot. can convert my plates through
basically a restyle or restyling the videos. So there’s a lot of
things that you can still use within that toolbox and processes that
will help benefit the overall production.
Orlando (22:47)
Yeah, mean, so the I mean, so what do you think that like, you know,
it’s funny because it feels like, you know, the disruption is
happening at every single level. You know, everybody’s kind of very
focused on the disruption happening at the level of the armchair
creator, you know, that everybody’s like, ⁓ somebody’s going to make a
movie in their dorm room. And it’s like, I’m sure somebody will, you
know, like.
Its level of quality will be debatable and it will improve over time
and they’ll get experience and all that kind of stuff. But obviously,
you you’ve got many stratas of filmmaking. know, there’s there’s
obviously independent filmmakers, sort of, you know, people like Sean
Baker, you know, when he did Tangerine just on his iPhone, you know,
he’s he’s demonstrated clearly that this is a professional grade
camera and it can do what you need it to do.
Erik Weaver (23:19)
Mm-hmm.
Yeah?
Yeah?
Orlando (23:33)
Then you’ve got, you know, obviously indie filmmakers who are always
trying to do more with less. Then you’ve got the, you know, the studio
level. It seems like just everybody’s going to be kind of leveled up
in their quality and there’s still going to be an appetite for
entertainment at each one of those quality levels.
Erik Weaver (23:50)
I completely agree with you. you know, I mean, yes, does it give a
little bit more power or empower a user economy? Absolutely. You know,
the MCNs have existed for a while. They’re out there. They’re not
going away. They have their own audience, you know, and the high grade
independence and premium films also have their, their, their place.
And what we’re seeing a lot is, is this technologies are allowing
people to punch above their, their weight, you know,
something that could have cost them a lot more, they’re now able to
execute within their budget. And that’s what’s kind of exciting to
many people out there in the field at this moment.
Orlando (24:26)
Yeah.
Yeah. And then, I mean, the other thing I wanted to kind of talk about
is that the the studios, obviously, and, you know, all of them are,
you know, relate to you. And I think that what’s very interesting is
studios have always obviously been very reticent to output, you know,
A.I. generative A.I. content on a consumer level.

Yeah. And then, I mean, the other thing I wanted to kind of talk about
is that the the studios, obviously, and, you know, all of them are,
you know, relate to you. And I think that what’s very interesting is
studios have always obviously been very reticent to output, you know,
A.I. generative A.I. content on a consumer level.
Obviously, because of all of the copyright concerns, they are the IP
owners of all of the most valuable copyright entertainment material in
the world. So they don’t want to get it wrong. They also don’t want to
step on each other’s feet as both a professional courtesy, but also
because of the legal implications. But the fact that they’re engaging
with you, you know, is I think it really encouraging. And, you know,
it’s very interesting that they’re not blind to it just because
they’re not out.
Erik Weaver (25:01)
Yeah.
Orlando (25:12)
things and you know, obviously Warner Brothers has the AI content lab
that is experimenting and it’s run by Kurt Zechmeister and he’s doing
some very cool stuff and it’s really exciting and they’re excited too,
but they’re just they’re not willing to output it. And so people kind
of say, the studios are being too slow to uptake this stuff. I’m like,
well, they’re not they’re being too they’re responsibly slow to
putting it out, but not
Erik Weaver (25:13)
Yeah.
Orlando (25:39)
blind and certainly not ignorant. In fact, I think between the ETC and
the things that they’re doing internally, I think that they’re kind of
on the front lines of it. They’re just not showing their hand yet.
Erik Weaver (25:51)
Well, so again, like like you said, there’s two primary legal issues
you have a problem with and the first one is your models your models
Can make or break a studio? So when I look at the industry, there’s
three types of models There’s clean models where they ethically
sourced or paid for all the content to train it There’s really only
three maybe four companies that I’m aware of that have done that that
would be Moon Valley Marey model with the stereo
That’d be Google because they’ve trained on YouTube content. And that
would be Adobe. So that’s your top tier. I also believe Nvidia has
something going on with ⁓ the Getty Foundation for original content.
Beyond that, you’re going to have a level of models that say they’re
going to offer writs and warrants. That means they’ll go to court for
you and they’ll fight, right?
Orlando (26:40)

Orlando (26:40)
Adobe is one of those.
Erik Weaver (26:42)
And then at the very bottom, you’ve got the wild, wild west, right?
And so that’s your tier one problem. Your tier one problem is, am I
going to spend the rest of my life in litigation? None of these
studios want to do that. That’s not worth their time. So they’ve got
to solve this first. The second problem is exactly what you said. It’s
the copyright issue. However, we are resolving how to solve that. And
there’s two use cases here that highlight that.
Orlando (26:52)
Yeah.
Erik Weaver (27:07)
slice of American cheese, which is the first AI piece of material that
got copyright and it wasn’t just simply based on prompts it was about
proving that they had manipulated the AI image later and they got
their copyright and the second one Yes, and then the second one is
critters critters is the first actual animation piece in which they
were able to document that as well and get copyright
Orlando (27:13)
Mm-hmm.
Significant human intervention, yeah.
Erik Weaver (27:34)
So these two are the pillars that you have to have solved at a studio
level before you can release these content. And so like we’re writing
custom comfy UI nodes to track provenance, creative intent, all of
that stuff, right? So that we can file copyright any of these things.
And then your second two major issues for the studios are gonna be
control and consistency and quality. So we’re already dealing with
quality as I talked with about.
Orlando (27:49)
Mm-hmm.
Erik Weaver (28:01)
And I have a clip that I can share with you about control and
consistency. It is getting there. It’s not a hundred percent. Like I
said, it’s a little bit of a wild horse still a little, but it is
changing on a daily basis.
Orlando (28:13)
But I mean, as the copyright owners, mean, the argument becomes like,
aren’t they, you know, training their own model on their own IP? And
that way they’re, you know, they’re sort of theoretically legally
black box to begin with.

But I mean, as the copyright owners, mean, the argument becomes like,
aren’t they, you know, training their own model on their own IP? And
that way they’re, you know, they’re sort of theoretically legally
black box to begin with.
Erik Weaver (28:25)
Yeah, so that’s what you would do. You would take a clean model and
then you would begin to Use a laura to create a custom model based on
your own content This is kind of where they were going with the
legendary ⁓ deal with runway So that’s your first example of some a
studio heading in that particular direction One other thing you got to
realize is that a lot of the studios do still really want to make sure
that they’re lockstep
With the guilds and unions and other things you might not know but in
the last ⁓ basically Agreement for sag they have to report all the
tools on a quarterly basis that they are using in any of these films
right? so, again, they’re not trying they don’t want to Throw cold
water on and I mean, it’s not for fair for me to speak directly for
them but I would say they really don’t want to throw cold water in the
face of
Orlando (28:53)
Mm-hmm.
Right.
Erik Weaver (29:17)
their beloved, know, the people that create those. So it’s a mixed bag
here. They’re trying to move forward in a cautious, genuine step as
far as I can see and tell.
Orlando (29:18)
Yeah. Yeah, absolutely. Yeah.
Yeah, I mean, so, you know, it feels like you’re very much at this
intersection of of what feels like a sort of clash of civilizations
in, know, the the technology sector that seems to want to, you know, I
feel like the the Silicon Valley and Hollywood have always had been at
loggerheads to a degree. and the thing that always kind of gets me
about the relationship is that very often Hollywood says.
Well, we do this this way and Silicon Valley goes, yeah, but if you
just did it this way now, it would solve all your problems, you know?
And they’re constantly, you know, speaking, they’re both speaking
English, but they’re speaking a different language. ⁓ What do you see
in that kind of clash of civilizations as it were? What do you see as
the places where they’re incredibly aligned and the places where
there’s unbelievable friction?
Erik Weaver (30:03)
Yeah, yeah.
Well, I think that everyone understands they have to evolve. No, no, I
totally get it. everyone realizes they have to evolve. you know,
Hollywood has traditionally been a really technically forward thinking
or institution where we’re, you know, constantly have been growing
over the years. If you look at the history there, it was interesting.
I was recently at one of the major

Well, I think that everyone understands they have to evolve. No, no, I
totally get it. everyone realizes they have to evolve. you know,
Hollywood has traditionally been a really technically forward thinking
or institution where we’re, you know, constantly have been growing
over the years. If you look at the history there, it was interesting.
I was recently at one of the major
Orlando (30:17)
It’s a big question. I’m sorry to just land that on you.
Erik Weaver (30:39)
Silicon Valley groups and they’re like, look, man, we’re a bunch of
computer scientists. We’re not filmmakers. We don’t speak that
language. We don’t really understand what you need. And so that’s why
I see partnerships like Astaria Moon Valley being great, you know,
cause Astaria is filmmakers Moon Valley or machine, you know, groups
that are working and saying, I need this feedback. I need these
aspects, you know? So, um,
I can’t give you the answer to all of it, but I think that it’s a
little bit of, you know, sword sharpening swords here. You know, as
they fight to the top, I think that they will both help each other
along the way.
Orlando (31:16)
Right.
⁓ That’s a great way to look at it. Is at the end of the day, each
group will be better. I think of the fear, obviously, from people I
speak to in entertainment, obviously this is no news to you, but with
Amazon buying MGM, Paramount being taken over by an Oracle proxy, that
something will be lost about the kind of tradition of
you know, filmmaking. mean, obviously, you know, USC and the ETC being
born out of, you know, the mind or the the the impetus of somebody
like George Lucas, who is undeniably been a clarion voice across
physical production, VFX, sound, practical storytelling. You know, I
mean, it’s.
Erik Weaver (32:04)
Mm-hmm.
And he lives up in the valley.
Orlando (32:09)
Yeah, exactly.
He’s not far away. Yeah, it’s it’s I think that that’s the fear is
that that something about the tradition will be lost. Do you do you
share any of that fear? Do you think there are things that we should
be doing to ensure that that. You know, the tech Ford entertainment
world that we might be going into remains.

He’s not far away. Yeah, it’s it’s I think that that’s the fear is
that that something about the tradition will be lost. Do you do you
share any of that fear? Do you think there are things that we should
be doing to ensure that that. You know, the tech Ford entertainment
world that we might be going into remains.
Erik Weaver (32:12)
Yeah, so.
Orlando (32:33)
as charming as the one that we’ve just come out of, you know, or is
the era of filmmaking we’ve come out of.
Erik Weaver (32:39)
So I kind of, on my own personal beliefs, think that what we actually
need to be a little more afraid of is becoming a modern opera. You
know, when you look at the new generation, they want agency, they want
interactivity, they want a bit of control. They’re not obsessed with
the quality and the features that we are, most of them are watching on
iPhones. So…
Orlando (32:53)
Mm. Mm.
Yeah.
Erik Weaver (33:05)
I just hope we don’t become too obscure. I think that that’s what we
have to worry about the most at this particular juncture. And again,
that’s a personal opinion.
Orlando (33:11)
Right.
Yeah.
Well, I mean, that brings it up an interesting point. mean, user
generated content, as you said, like, you know, people want to be
engaged. They want to be, you know, part of that storytelling. What
are you exploring any technology technologies or, or storytelling, you
know, tech use cases that enable UGC on a, on a, you know, broader
interactive level. mean, ⁓
apart from just people prompting scenes and editing them together.
Erik Weaver (33:47)
So I do have a, I don’t know what the best word for this, but I
basically have a trans media program in which I empower five teams to
look at story from a younger perspective or that this generation
perspective. And so, we were kicking that off right now for this year.
It’s our second year doing that. So yes, I am still exploring that. I
think that there is a sleeping giant in some of these things.
That obviously we see a large format which I would include, know
sphere IMAX Cosms doing an amazing job. I think that large format
drops Yeah, I’d even categorize Apple vision Pro in that I’d love to
see if that changes or what but I think that there’s a huge sleeping
giant around the corner ⁓ that’s not tapped and it may not be tapped
for another year or two, but

That obviously we see a large format which I would include, know
sphere IMAX Cosms doing an amazing job. I think that large format
drops Yeah, I’d even categorize Apple vision Pro in that I’d love to
see if that changes or what but I think that there’s a huge sleeping
giant around the corner ⁓ that’s not tapped and it may not be tapped
for another year or two, but
Orlando (34:22)
Yeah, I went to see that Matrix thing, it was great.
Mm-hmm.
Erik Weaver (34:40)
I definitely think the whole concept of stories and engagement with
stories has a very strong future. And please don’t forget that the
whole thing is about telling stories. So tech is wonderful and
beautiful, but it’s here to serve story in our opinion.
Orlando (34:55)
Mm-hmm.
Yeah. You you said it, you you talked about kind of, you know, is it
an accelerator or just a cohort of, you know, young people, the
initiative you just talked about? Could you, yeah.
Erik Weaver (35:07)
So we basically go
to USC students and we put up an application process where five teams
are allowed and we give them budget and everything and they come up
with their own ideas.
Orlando (35:18)
That’s awesome. Yeah, because I think that like what’s what’s very
cool about that is that, know, you obviously there’s the engaging with
the studios, but then you are sitting at USC. You know, you are
engaging with the next era and the next generation of storytellers.
And if you if you focus too much on what the the the studios think
about and care about all the time, you’re going to get the skew
morphic lens on, you know, on the next generation of entertainment
and.
It’ll look like a more technologically advanced example of the
previous generation of entertainment. Whereas the opportunity, I
presume, of sitting at USC is you go, well, how about people who, you
you know, don’t have the same viewpoint on the industry and think that
it changing and it putting down its guard rails is the greatest thing
that could ever happen.
Erik Weaver (36:08)
So I actually got fascinating story here is that the ETC at the board
meetings, every other board meeting, we used to have a situation where
we would bring in six or seven current students and the board would
then ask them questions. And it would be really interesting to see the
completely difference in opinion. So I don’t know if you remember when
there was something around called MoviePass, right?

So I actually got fascinating story here is that the ETC at the board
meetings, every other board meeting, we used to have a situation where
we would bring in six or seven current students and the board would
then ask them questions. And it would be really interesting to see the
completely difference in opinion. So I don’t know if you remember when
there was something around called MoviePass, right?
Orlando (36:23)
Wow, great.
yeah.
Erik Weaver (36:33)
Every single one of the students went to the movies at least once if
not twice a week when they had when movie pass existed They said it’s
a social activity all of our friends have it. We all go right and
guess what? As soon as that was lobbed off they stopped going to the
movies and it’s like oh I Mean and I don’t know how to make something
like that work, but but it’s just
Orlando (36:39)
Yeah. Yeah.
Yeah.
Yeah.
So close. You were right there. Yeah.
Erik Weaver (37:01)
we can empower a whole new generation to love cinema in that physical
form. It would be awesome.
Orlando (37:08)
Yeah, there used to be a thing in the UK, I remember, called Orange
Wednesdays, and it was orange, I don’t know if it even still exists or
it’s been consumed by a larger telecoms company, but if you had an
orange mobile phone, you got a free movie pass every Wednesday. And it
was, yeah, I did some of the advertising for it, and it was incredibly
popular. When it was first launched, our budget was next to nothing.
And then I think six months later, they did a renewal campaign.
Erik Weaver (37:24)
Wow.
Orlando (37:36)
And our budget had tripled because they were like, it’s one of the
most popular initiatives we’ve ever done. And it’s like, yeah.
Erik Weaver (37:41)

I mean, it almost feels like
this whole idea of Kodak saying, I don’t want to promote digital
cinema because it’s going to eat my film business. You I don’t want to
eat into my revenues of traditional theater. Therefore, I’m not going
to do this. And I’m like, no, please.
Orlando (37:57)
What do you think? I mean, getting your sort of prognostication hat
on, mean, are, are we even able to keep pace with this rate of change?
I mean, are you, you know, do you feel like the stuff that you’re
doing now, for example, the bends,
Is there a sense at all that by the time you’re finished with the
production, the workflow is almost out of date? mean, it feels like
things are changing so fast that how do you even keep pace? ⁓
Erik Weaver (38:25)
That is absolutely a concern. And as a matter
of fact, you have to keep putting lines in the sand and saying, this
is how it is now. And then you have to eventually change that line. It
is very, very hard. so like we traditionally release a white paper,
like maybe six months later, that doesn’t work anymore. You just can’t
do that. So we’re beginning to do a lot more social media.
Orlando (38:44)
Mm-hmm. Right. Yeah.
Erik Weaver (38:50)
In a real time nature, so going to ET centric on IG or LinkedIn and
catching what we’re doing because knowledge is now. And it’s fleeting
and you know you’re trying to stay a little more high level when we’re
saying things or suggesting things. Look at this from this point of
view or how this would impact your workflow.
Orlando (39:00)
Yeah. Yeah.
Mm-hmm.
But I think, listen, I think that’s the future. I mean, I think that
we all know that building out loud is a huge part of the modern
consumption experience of craft. I I spent just the other day, I spent
15 minutes watching somebody clean a very dirty rug. It was very
satisfying. And people kind of say we’re losing craft. I don’t think
so. That rug cleaner was a craftsman and I watched it for 15 minutes.
Erik Weaver (39:39)
Hahahaha

Orlando (39:41)
⁓ I think that
actually entertainment on some level has a craft problem only because
it’s not exporting its own craft in real time. There would be nothing
better than watching somebody who’s producing a gory horror film, an
A24 film or a Blumhouse film, showing me in real time what potentially
somebody getting their head crushed in with a…
Erik Weaver (39:47)
yeah.
Orlando (40:06)
you know, I should speak to my therapist that I keep using this
example, you know, some VF, you know, some practical effect of
somebody getting hit and blood splattering everywhere. I think that
the entertainment certainly could be in a situation where if you
followed a production designer or you followed a, you know, practical
effects person while they’ve been making a film,
Erik Weaver (40:07)
Ha ha!
Orlando (40:28)
I think there’s a beautiful future in which you might go to the film
for the VFX person, for the production designer, for the makeup and
hair person who you have been invested in their story. You saw them
go, I can’t believe I just got booked on Superman. I’ve been doing
Indies forever and now I’m working on Superman. And like, let’s be
real. I stopped watching trailers about…
Erik Weaver (40:36)
Yeah.
Yes.
Yeah.
Mm-hmm. Mm-hmm.
Orlando (40:52)
you know, about two years ago, and I kind of put my headphones in,
even at the movies, if I show up too early, I put my head down, I
listen to music, because they’re giving away too much anyway. I tell
you the whole story. And you know, after Endgame, anytime there’s a
surprise new character, it’s in the trailer. I’m like, well, what’s
the point? You know, and I do think that like, you’re giving away so
much anyway, I’m seeing the costumes, why can’t I see the costume
designer, you know,

Erik Weaver (40:59)
Hate it. They tell you the whole story ⁓
Orlando (41:19)
in real time, outputting that stuff. Once the trailer is released, let
the costume designer release, know, dailies on what it was like on
set. I think that’s a huge opportunity for physical craft.
Erik Weaver (41:32)
Yeah, I mean, there’s some aspect of a flatty or a film is really
going to be just a introduction to a world, right? And then narrative
or breaking off pieces or interactive pieces are going to be the more
engaging components. ⁓ And I wonder how that will evolve over the
years. One other thing that I thought was a really interesting story
is USC does run an AI film program and
Orlando (41:41)
Mmm, yeah.
Mm-hmm.
Erik Weaver (41:59)
someone had gone to a senior official and said, you know, why have I
even bothering spending the money getting this degree? And they took
the course and at the end of the course, they said, I absolutely
understand why I have to have a degree now, because I have to
understand how to tell a story, you know, and yes, some of the
technological components will change. But if you don’t get that story
aspect, that way in which to craft a narrative, to take somebody on a
journey,
Orlando (42:13)
Mm-hmm.
Yeah.
Mm.
Mmm.
Erik Weaver (42:29)
then
you’re still gonna look bad no matter what technology you’re using.
Orlando (42:35)
Yeah, and the rules are different for each environment. I mean, you
can show the ordinary world in a film, and then you can take somebody
through the inciting incident and move on from there. On social media,
you have to hook somebody immediately. You start with the inciting
incident on social media. You know, that is moment one. That’s
fascinating. mean, you know, so…

Yeah, and the rules are different for each environment. I mean, you
can show the ordinary world in a film, and then you can take somebody
through the inciting incident and move on from there. On social media,
you have to hook somebody immediately. You start with the inciting
incident on social media. You know, that is moment one. That’s
fascinating. mean, you know, so…
And so just if I could kind of tell people where they can find, where
you’re kind of building out loud and you’re sharing all that stuff,
where is that being shared?
Erik Weaver (43:04)
So our white papers are published on etcentric.org. I do a lot of
stuff on LinkedIn. We have ETC at USC on IG. So we’re a little bit
around and just reach out, connect with us, follow some stuff. Again,
we’re not a for-profit organization. not trying to
Orlando (43:22)
Yeah.
Erik Weaver (43:29)
push a brand or a product or anything like that. We’re just trying to
look at it at the future and execute within context of story. And you
know, sometimes we’re wrong, sometimes we’re right, sometimes we fail.
We always deliver product though. But you know, again, we’re just
trying to look around the corner and be helpful.
Orlando (43:31)
Mm-hmm. Yeah.
Yeah. And so, you know, let’s say somebody is already in the creative
industries or is technologically advanced, very committed to this.
What is the what is the path or the place where they should be going
for the most amount of information about, you know, what they should
be doing to learn about training? Laura’s and understand how to get
towards closer to character permanence across shots. And, you know,
what are.
you know, where does one go for that information?
Erik Weaver (44:16)
So mean, beyond what we’re releasing, I think Curious Refuge has some
really good courses. Caleb over there is amazing. I love his weekly or
bi-weekly videos. They’re really great to show you what’s updating.
Beyond that, it’s kind of tricky. You’ve just got to hunt and peck out
there. I say that apologetically, but I have access to a lot of very
bright minds. And so we work through problems with those minds.
everyone might not have that access.
Orlando (44:43)
Yeah, those are the creatives I always gravitated to. I was a producer
at ad agencies and various other places and the creatives I always
gravitated towards most were the ones who kind of would come in with
bags under their eyes going like, I couldn’t sleep because I found
this new tech and I just was playing with it all night and I kind of
did this and I kind of did that or, or, you know, I got myself a
speed. Yeah.

Yeah, those are the creatives I always gravitated to. I was a producer
at ad agencies and various other places and the creatives I always
gravitated towards most were the ones who kind of would come in with
bags under their eyes going like, I couldn’t sleep because I found
this new tech and I just was playing with it all night and I kind of
did this and I kind of did that or, or, you know, I got myself a
speed. Yeah.
Erik Weaver (45:06)
That’s exactly what we, yeah,
you want those people who just are fascinating, wanting to learn,
wanting to change, wanting to get their hands on it. And I guarantee
you that that’s not just a this generation thing. I mean, again, I’m
working with a lot of wonderful folks from unions and guilds that
just, that’s how they’ve always been. ⁓ I want the new lens. I need
the new sensor. I need to…
Orlando (45:13)
Mm.
Mmm.
Yeah.
Erik Weaver (45:30)
play with this and let’s do that, you know? And they’re embracing this
just as much as anybody, so.
Orlando (45:32)
Yeah.
Yeah.
Yeah. mean, it’s it’s like we said earlier on, it’s like, you know,
the these are the things that should be exciting. I mean, I remember
when Dreamworks came out with the, you know, they cracked mud. That
was that was such a huge, you know, moment. then, you know, that that
package to be able to do mud reliably went around to every, you know,
post house in the world. And, you know, people kind of realized what
they did and.
and got so excited about it. I think that those are, there’s a world
in which, if you’re in the creative industries, I think that the most
important tool in your toolbox is an open mind and curiosity. And it
sounds like you just get to live in that every day. I’m incredibly
jealous. It sounds like a fantastic position to be in.
Erik Weaver (46:15)
You
Mm-hmm.

Yeah, pretty great. And I think the biggest reward for me is when I
see the people that just need a chance rise up. And we’ve seen some
absolutely amazing people across come out of some of these films that
we’re making that just flourish. So it’s awesome.
Orlando (46:29)
Mm. Yeah.
Mm-hmm.
Yeah, if I can ask one more question, I mean, is there anything you’re
working on now that you’re particularly excited about or even anything
you’re working on now that you’re particularly nervous that you might
not get right? I mean, what, you know, what’s, what’s just beyond your
horizon even?
Erik Weaver (46:50)
So
I’m working on two key pieces right now, the bends, which is a purely
generative, like we mentioned about the blobfish. And then I’m working
one on pathways where we shot with iPhones ⁓ and we shot practically
and we are now changing everything out using AI. That one is becoming
a little more complex or tricky, nuance wise than we expected. Tools
like Beeple.ai are amazing or Fota, which is remit.
Orlando (46:59)
Mm-hmm. Yeah.
Mm-hmm.
Erik Weaver (47:17)
Lensing the scenes are fantastic, ⁓ but we’re not quite there yet. The
next level or the next thing that we will be touching is performance.
And I think that performance is the current edge of tomorrow.
Orlando (47:19)
Mm-hmm.
Wow.
Yeah. Do you think that’ll be the white whale? I mean, is it it is it
something that you feel is is achievable within a period of time or do
you think it will? I mean, uncanny value. Will it be just uncanny
Valley sort of forever?
Erik Weaver (47:44)
It will definitely be achievable here in the next nine to nine to 24
months. Its application is going to be what’s curious. Like I would
really only see a lot of this for things like dangerous stunts,
challenging environments, things like that. But who knows where it
will be taken.

It will definitely be achievable here in the next nine to nine to 24
months. Its application is going to be what’s curious. Like I would
really only see a lot of this for things like dangerous stunts,
challenging environments, things like that. But who knows where it
will be taken.
Orlando (47:53)
Mm-hmm.
Right. Well, Erik Weaver, thank you so much for being on. It’s been a
real thrill for me. You were you were slightly my white whale. I just,
you know, I I just have been so jealous that your job exists and you
get to do it for so long that, you know, talking to you has been a
real thrill and I’ve really enjoyed it. Thank you so much.
Erik Weaver (48:21)
You
⁓
thank you.
Orlando (48:32)
Absolutely.