From Concept to Screen: Virtual Production in Film and Television
Hi everyone, I'm Kali Bateman
here for Mixing Light.
And today I've got Bryn Morrow here,
who's the virtual production supervisor
for Steel Bridge Studios.
It's really exciting to have you here
to have a chat today about volume work.
We met grading some car commercials together,
which actually I didn't know were shot on a volume
until about halfway through the session.
And I turned off a resize and went,
oh, look, there's the edge of the universe.
And it was really impressive stuff.
Once you started to look for it,
you could start to say, oh yeah,
maybe that's something to do with the volume,
but it was pretty seamless stuff.
And we got to talking about how it all works.
And I realized how little I knew
about how volume production works.
So very excited to have you here
to shed a bit of light on that today.
Thanks so much for coming.
- Oh, thank you.
I'm very happy to be here.
So thanks for the nice introduction, Kali.
- Such a pleasure.
So can you just tell me,
basically from the very beginning,
you need to have a image to go on these volumes.
So just to try to explain what a volume is,
and you'll probably do a far better job than me,
but a large LED wall or walls
that serve as a background for motion pictures.
So cameras in front of the volume
and usually actors or props,
some kind of integration of physical sets
in between the volume and the camera.
And then a whole bunch of data is recorded
and the backgrounds are happening in real time
through Unreal Engine.
Have I kind of got the basics down there?
- Yeah, actually, I think you did very well there.
I mean, there's obviously variations
to how virtual production is done,
but that is certainly one way of doing it.
And in the project that you were grading for us,
that was essentially how it was done.
Like the backgrounds were
projected onto an LED panel,
like a large sort of screen,
which is a combination of a
bunch of very small panels
as well that we combine together
and that particular one,
from if memory serves me right,
that was about, I think it was about 15 meters
by four and a half meters high for the main wall.
But yeah, so I mean, essentially a volume
is a combination of panels
that are sort of stacked together to make a wall.
And then whether you have a
ceiling LED panel in there
and a side, a couple of sides
or a reflection sort of panels as well
that can combine to create, as we call it a volume.
And you don't necessarily have to have
digital assets in there.
So that particular example, yes,
those were all virtually created
from our virtual art department.
And so those were full 3D environments,
but we've done many projects
where we have live action plates that we filmed
and then those are projected onto the panels.
But essentially that's what's happening.
We're projecting the digital environment
onto the back, onto the LED panels.
And we have sort of a 3D space recognition
that we understand where everything should be
in the volume in terms of its 3D positioning.
And the idea is for us to translate that
to the real world set that we've created as well.
And then that needs to exist
in our digital environment.
So we have specific distances that correlate
to distances to the panels.
And essentially you're able to then get parallax
purely because the camera that we have,
which is obviously a production camera
and can go into all the different sorts of cameras
that we would use for virtual production,
but that camera is being tracked in real time.
So we have a bunch of motion capture cameras
that we create our own little volume for.
And those are constantly tracking
the physical production camera in space.
And then that is translated
into our digital environment.
And we have a digital replica camera
in the digital environment.
And those two essentially are matched together.
So that's how when the practical camera moves,
then the background is moved in relation to that.
And it's field of view changes based on lensing.
Well, once we adjust those,
and so just the parallax
shifting in the background,
that's where that all sort of happens.
But that would happen more
for digital environments.
But if it's a live action
shoot that we've captured plates,
it doesn't exist the same way.
Of course, that then is not,
it's not a digital environment anymore.
So the parallax is still
sort of there in some senses
because we do a sort of a two
and a half D sort of workflow.
But in that sense, it's not quite as 3D
as the 3D virtual ones.
- Wow, okay.
I think for me, it kind of comes together
when you tell us how the camera itself
and the motion capture in the camera
is linked to a 3D camera.
I think that's the bit
that I didn't quite understand
before our discussion,
because it all seems quite magical to me.
So the process is that you
would get those environments
digitized prior to the shoot, right?
So if it's a fully 3D environment,
that's quite a bit of work that's gone in
prior to actually cameras rolling.
Can you tell me a little bit
about how you take an image?
And it can be a 3D one or it can be a,
something that you've sourced in the real world.
How do you then get that into the volume?
- Yeah, I mean, if we could start
from sort of the 3D side of it,
our virtual art department,
obviously we've been getting a brief
as to what that environment wants to be
and how realistic it also needs to be.
Because of course, it's subjective,
working in an artistic sort of realm,
there could be multiple sort of environments
that you can make,
but in certain, I mean, in most cases,
it is realistic as we can make it.
And then of course, we very
much create the environments,
sort of we like to Hollywood the environment.
So in that sense, we build a full 360 environment,
but we do Hollywood it out
to a specific field of view,
at least for a range of motion for the camera
that we're expecting,
so that we don't create too
huge an environment because--
- What does that mean, Hollywood it out?
I've never heard that.
- It's the term, you know, like back in the day,
in production, you know, you're building a set to,
you don't have to build in full environment,
you know, the whole world if you're building sets.
And so we consider it just Hollywooding it
to the environment, I mean, to the camera.
I don't know, it's a term
that I picked up way back.
So I don't know.
- I love it.
- So anyway, we ultimately build the environment
to the camera's, you know, like the range of motion
that we're expecting it to sort of go along
and we're able to obviously deduce that
by doing previs or techvis
and also obviously, you know, creating boards
and we have a, you know, we
get quite far into our previs
prior to, you know,
projecting it up onto the screen.
So we already sort of have a pretty clear idea
of how much of the environment's gonna be seen
during the production.
I mean, that doesn't mean that we're not giving
the director as much
opportunity to change the camera
as, you know, they generally would like
and also the DP for that matter, obviously.
So, but we certainly give
them a bit of a constraint
and we've already built it to sort of, you know,
we're expecting, for instance, you know,
we wanna see behind us,
we're gonna be shooting reverses,
we're gonna be shooting sort of POVs
and sort of wide vistas or whatever.
And so we'll build it to those constraints.
But again, there's many sort of factors
with regards to optimization and things like that,
which the reason why we don't build everything out
is purely because of optimization
and being able to run at the required frame rate
that we're gonna wanna roll at.
And so again - go down the rabbit hole.
We could sort of get into
that later if you want to, but--
- Yeah, sure, I'd love to.
- But ideally, you know, the idea would be
once we've pre-vised out those environments
and, you know, we can have
multiple environments in a day
that we wanna shoot,
we'll then have a
pre-light at least with the camera
and we'll project those onto the wall.
Now, when I say project those onto the wall,
we, within Unreal Engine, for instance,
we have something called
nDisplay, which is a plugin,
which essentially is a projection plugin
where you create planers within the scene
which represent the panels,
represent the position of the panels.
And that nDisplay node
essentially is what we call it,
that we then have the environment all around
and then with the camera,
it projects onto a very specific position
of where the panels should exist in space.
And that becomes, that's kind
of where the trick really lies
in understanding the real distance
that things relate to the real world,
from the digital to the real world,
which is why we tend to LIDAR scan our sets.
So we will go ahead and do a full LIDAR scan
of the set environment,
whether it's just in a studio,
where the panels in position,
where they are in position,
where our camera is expecting,
we create something called a home base
so we know where the camera wants to always be
at the beginning of a shoot.
And so we then do a LIDAR scan
and from that point, we filter
that back into Unreal Engine.
So we actually have real world scale positioning
of everything.
And once we've done that,
we'll then align our nDisplay node in Unreal
to where the panels are, and
then where that set exists.
And then we can work out distances
from practical positioning.
Like obviously, for instance, if we had Artabar
and bump in some foreground elements or whatever,
those elements need to exist
in the space digitally as well.
It's what'll end up happening
as things will overlap on the screen.
And then something on the background
is supposed to be in front, et cetera, et cetera.
And so you need to be very
clear about where the opening is
and how you align that.
And it's all about
calculation, real world calculation.
And then that needs to be translated
into the digital space and vice versa.
- Is that why it's called the volume?
Because it's all about the space,
the actual depth and distance between things?
Is that something that gets
its name or is that some other?
- For me, I think volume
because it's a volume of light.
So when you have multiple panels
and generally it sort of
encompasses you as the environment
and you're creating
essentially an exterior environment
inside an interior environment,
that becomes a volume of
light, its own environment.
And so that's how I...
That's where I'm getting it from.
I still like what you're saying though.
I don't mind that.
I can think of it that way.
- I've been trying to just,
as I've thought about speaking with you about this,
trying to imagine where the term came from.
So I'm looking for it everywhere.
It sounds so futuristic, but that is fascinating.
So you've kind of got these projected elements
as well as physical elements.
And that was certainly the case
in what we worked on together.
There was a foreground that was physically bumped
into the space.
And then on top of those two
things just existing together,
you've also done a LiDAR scan
of those elements in the space
and you fed them back into the digital simulation
of the environment.
- Yes, that's great.
- That must be the trickiest
thing is getting that scene
between the digital and the
real kind of working seamlessly.
- Yeah, that's definitely the trick.
And we tend to build a lot
of, in our digital world,
and depending on what sort of practical set
that we have available to us,
we tend to build into our
digital set little tricks.
Like, for instance, in that Sydney scene,
I know it's hard to talk
about it when you can't see it,
but we built in a
pavement, an edge, a pavement edge
where we're expecting the
edge of the panels to exist.
And so we will build in where
that edge of the panel wants,
sorry, the edge of the pavement,
which happened to be in the digital set.
And then we decided, okay,
we'll replace that pavement
with the real pavement
edge, so that becomes practical.
And then we align that up to our digital pavement
and we delete our digital pavement,
and then that becomes the new pavement position.
And so when the camera
moves, it should stick in a line.
I mean, I think there's just,
in those instances, you're then locked into
a distance to the panels.
And so I think one of the main advantages,
I think, in volume and in sort of quick shooting,
if you wanna do that,
is to be able to cheat the
distance backwards and forwards
and move things around digitally quite quickly,
and to be able to, so in other words,
if you build something into your set and say,
now that is now physically connected to the set,
and it's very heavy, for instance,
when you start spinning the environment around,
and I guess we haven't
really kind of gone into this,
but for instance, if we wanted to do a reverse,
there's no panels behind us,
so we spin the environment around digitally,
and then suddenly that's the reverse of the shot
and keep the camera in the same position,
so that you're kind of cheating it in that sense.
But if you've now built a practical floor
which wants to connect to your panels,
that's incorrect, right?
So then that would need to move too,
so you'd have to flip that around,
or you don't always wanna be doing wide shots
that need to obviously
align with each other in the cut
and things like that, so you just need to be quite,
you don't wanna build too much infrastructure
into your foreground,
which doesn't have the ability to move,
I guess is what I'm saying.
- Yeah, unless you have it on a turntable.
- That's right, yeah.
And a lot of that is the case,
and people do tend to do that
when it comes to much
larger virtual production shoots,
which is a massive advantage,
and if there's that sort of budget thrown at it,
then I will say yes and put my hand up
and say let's do it every single time,
because it just helps out,
and helps the production designer out too,
and of course, we also then,
every time there's new props or anything like that,
we try to scan those as quickly as we can,
but those are just sort of
photogrammetry style scans,
not full LiDAR scans,
and then we'll try and,
because we would, like I said,
when we spin the environment around,
there could be elements that were created
as the bump in practical set,
those then now suddenly exist in the digital world,
you'll actually see them through the screens,
and if you haven't created those exact props,
then they won't be able to be extended out
into the digital world,
so that's why we also scan and recreate all those,
and match them if we can.
- That is absolutely full on,
like the level of precision that you would need.
I know that in the color grading world,
we're typically not working
with that level of precision
in terms of getting
everything measured and accurate,
it's sort of the approach is
more about getting a feeling,
and instead of doing like really tight shapes,
we're typically doing softer shapes,
and sort of being a little
bit more gentle in our approach
to manipulating the image
and trying to do a bit less,
but this is sort of the opposite
where you're doing everything,
you're generating everything.
I imagine you get into the uncanny valley
pretty easily doing that sort of thing.
- Yeah, I mean, and look,
and having sort of broad power windows,
I mean, that stuff is great,
and it certainly lends itself to what it is
that you guys are doing,
when all you have is a 2D image
to manipulate,
but I think for us, it's just,
it can very easily go wrong,
and so we're just always
trying to stay physically accurate
as much as we can before the cheating,
and even though it is a full cheat anyway,
the only way for that to be a reality
is for us to be very strict and constrained
to the physical aspects of what's going on,
and so that needs to sort
of loop back the whole time.
And yeah, I mean, once you create a pipeline
for what it is that you're doing,
these sorts of things don't
have to always be thought about
as intricately.
I mean, ultimately, I will be the one
that needs to keep an eye on that sort of stuff,
and if it goes awry,
because everybody within our brain bar,
which I haven't really sort of discussed,
but they have individual jobs.
- Yeah, but it has now become a universal term
for some reason, so
everybody uses it as the brain bar.
They call it the brain bar.
- That's cool.
So I suppose I'll just do
a little bit of background
about the actual studios themselves.
So Steelbridge are quite new facility in Brisbane,
and it's been really popular,
and Australia's actually been quite a large adopter
of virtual production worldwide.
So we've got some
facilities in Melbourne as well at,
jeez, I can't remember the name, but--
- Nance Studios.
- That's the one. - Nance Studios
in the Docklands, yeah.
- That's right, yes.
And VCA have put in some
virtual production facilities there
in Melbourne as well, but Brisbane has Steelbridge,
and then there's some volumes available
on the Gold Coast in the studios there as well.
But Steelbridge are probably
the most active in the region
in terms of doing a lot of work.
They've got their pipeline pretty well down.
They've done quite a lot of commercials.
I'm not so sure about Longform,
but I'll have to ask about that,
what you guys have been involved in,
if anything, in that space.
But definitely in terms of commercials,
they would be the ones that
you would go to in our area.
And some of the advantages of
it in the work that I've seen
is we worked on a car commercial together,
and the environments were quite varied
because our landscape in Australia is quite varied.
So they were able to do sort of red desert dirt
and city driving and coastal driving,
all of these environments in one day.
So obviously the practicality of that is great.
You only have to take the cars into one space.
You don't have to freight them around,
and you can go to a lot of places quite quickly.
So, although you're doing
quite a lot more in pre-production,
I think in terms of production,
everything's a bit more sewn together.
But yeah, so in Brisbane,
they've got this great facility
and the pipeline's pretty well sorted out now
and some pretty seasoned staff on board,
including the friend who
we're talking to right now.
- Yes.
Yeah, look, I mean, Steelbridge has been
certainly pushing really hard,
and I think it's been,
I've been very fortunate to
be able to work with them.
I mean, initially, it was a bit of a brainchild
between myself and Colin,
who's the founder of AltVFX and owner.
And, you know--
- That's Colin Renshaw.
- That's Colin Renshaw, yeah.
And essentially,
obviously the new tech was coming in.
This was quite a few years ago now,
but it was just sort of something that we knew
it was gonna become a thing.
And at the same time for Alt Studios
AltVFX, sorry, they were pushing their animation,
which is obviously quite
integral to the work that they do,
a lot of creature work that they do.
And so motion capture was certainly a thing
that they used,
and they used a lot of XN suits.
- They were very good at motion.
Very, very good at that.
I remember doing some grades,
and occasionally I'd see people walking past
in those suits.
- Yeah.
- So those suits are sort of like a,
almost like a wetsuit with dots all over them.
- That's right, yeah.
- And then as the character moves around,
those points are captured,
and then they can be transposed onto a 3D model.
Like you could see a
dancing polar bear or something,
but it's just,
I love seeing that behind the scenes thing
of the person doing the silly dance,
and then the polar bear does it behind them.
I think that's always a cool party trick.
- Exactly, yeah.
And so I think that Alt just wanted to push
their motion capture facility,
and actually buy and look into the Vicon cameras,
which are the very high end
motion capture cameras for,
to again, push their
pipeline, their animation pipeline,
but at the same time,
find a permanent place for that as well.
And I think that's where
this sort of started growing
from there, and of course,
with having the motion capture facility,
where you don't,
it's not an X-Sense one,
but it's more a camera driven one.
You can have multiple people wearing suits
at the same time interacting.
So, you know, there's certainly a lot of fun things
going on here, there,
and we do a lot of that here at the moment as well.
So it's fantastic to be
able to push both parts of it,
and it's quite integral in the
visual effects side of things.
Of course, I'm a visual effects supervisor too,
so it's a very important
aspect of visual effects too.
So that in itself drove the volume,
motion capture volume,
and in that sense, you know,
we're tracking a camera at the same time.
And so we decided that, you know,
with that technology working out and, you know,
moving into the real time
sort of software like Unreal,
those obviously go hand in hand.
And so we just decided to push
onto the virtual production side of things as well.
And while both those streams
continue the motion capture,
and then also the virtual production,
and then it's developed into
where Steelbridge, you know,
creates a lot of virtual production commercials.
And yes, there has been some long form inquiries,
and we talk about a lot of
long form projects with producers.
And I think, and again, look,
it's always gonna be something that producers
and or production people, you know,
it's a learning process for everybody.
And I think it's hard
sometimes for people to commit
to a project, a virtual production project,
especially if it's a long form project,
and it's just a really big
part of what they're doing.
And there's so many things that can go wrong
that they're not in control of anymore.
And I think it's certainly
hard to get people on board,
but I don't think that it's far off.
And we're talking to a lot
of people is all I can say.
- Do you think that's why, you know,
the kinds of things that have been shot
in the long form space in volumes,
like Mandalorian and the
Batman and things like that,
that have used them quite a lot,
they've had really big
visual effects studios behind them
that are actually responsible for the production.
So in the case of the Mandalorian,
it would be people who
intimately understand the technology,
who are kind of funding
and controlling the decision making process,
as opposed to more traditional productions,
going out on that limb and
having to have that leap of faith
and trust in someone else's,
like obviously amazing abilities,
but just, it can be
difficult when you don't understand
what's involved, right?
- That's very true.
And also, and this is certainly something
that Colin mentions a lot,
and it's very true is that
if you have a post-production
facility that's behind a
virtual production facility,
you have the ability to fall
back on sort of visual effects
to fix sort of things.
And that sort of happens a
lot in virtual production.
I mean, as much as we're all
trying to get 100% in camera
out of the virtual production techniques,
I mean, it's most probably more like, you know,
it could be up to 60 to 70.
If it's a long form project,
the problem is that so many changes can happen
further down the pipeline.
You know, it's very hard to make all the decisions
at the beginning of the project,
and that is essentially what is happening
in virtual production.
So I think you kind of then look at it as,
what percentage are you
gonna actually get in camera
that you're happy with,
and it's not gonna have changed
when you've suddenly decided
you're changing your character
or you're changing your monster, whatever it is,
later on in the process,
then of course everything that you shot
gets thrown out the window.
- But it's the same in
regular production, isn't it?
Like it's exactly the same, you know,
possibly more visual effects required
for things that are shot photographically
in the real world, right?
Like nearly every shot in a long form show
can have a hidden visual effect in it
that the audience would never know about.
- That's true, but the problem is,
is that when you don't wanna pay twice, you know?
And so the issue in that sense
is that if you're shooting it,
that's fine, but then you
might have a visual effects budget
on top of that and that works out.
But then if you're kind of, you know,
already doing a virtual production budget,
which in some cases can be higher
than just going out and shooting on location,
then you're sort of,
and then you then need to
do the visual effects process
back on top of that,
then that can become a second, like a double cost.
I think that there certainly is some sort of,
you know, you have to think about that
and make sure that everybody's on board
with what you're gonna get
and the fact that you probably have to build
a contingency plan into the budget
that there will be shots that need to be fixed
or shots that need to change or extended, you know,
and that tends to always be the case.
- So can you talk to me a little bit about
how you do that pre-visualization
and how you might digitally scout a location
and how do you get that world built digitally
before you go ahead and, you know, storyboard and,
yeah, how do you take the director
through these environments?
Because you're not physically location scouting
and looking for physical props here.
- Well, I mean, you say that,
but sometimes that is the case
and certainly in the proof
that we worked on together,
all those locations were
essentially sort of located
in the real world locations.
And then we went and, you know,
sometimes went and LIDAR scan those sets
and converted them to digital sort of sets.
Or we just took heaps and heaps of reference photos
from that and then just
digitally replicated those places
because those needed to be iconic locations
that people needed to recognize.
But again, in say, you know,
we've done a lot of interiors and things like that,
you know, in that sense,
we will actually initially
talk to the production designer
and or the client,
depending on the size of the project
and, you know, come up with
the locations with, you know,
treatments, directors boards, all the rest of it
that you generally come up with
and send that to our VAD team
who will start building it out
and will do sort of large renders
as we're moving along in the process
and showing those to the
client and director initially
for them to sort of kind of
enjoy what they're looking at.
And then once we know we're on board
with that environment we're building to the right,
or at least the correct creative on that,
then we will go in and we'll
do sort of digital scouting,
which is, you know, you can either do that
with the virtual headset,
like a VR headset at least,
or, you know, we'll just,
you know, get a director in,
you know, sit over one of the artists shoulders
or whoever it is, and then
we'll actually lens it up
for them, you know, with the
proper camera and sensor sizes
and then we can sort of look at finding
those right frames that they like.
And then those can become part of the storyboards
and then we'll go ahead and
either have digital storyboards
or we'll kind of draw back over top of them
and add people in, or we
could do that in Unreal as well.
I mean, that's also sort of
quite a few methods in that sense.
And then after we've sort
of built these environments
that I think the director's
also sort of enjoying and liking
and we know now what to build,
we'll then do something called techfers
where we create the panel layout,
which we've already sort of worked out.
Well, it's kind of like a
mutual sort of decision there
because here at Steelbridge,
we tend to build a volume out
based on the needs of the shoot.
It's not always a general LED
volume layout that you would have.
You know, we wouldn't always have a floating wall
that can be maneuvered if we don't have a car
or sort of some sort of object
that needs to be reflected on all the time.
So there's different ways to build the volume.
We won't always have a ceiling in.
We probably won't always build a curve to our wall.
It might be a flat wall.
I mean, there's multiple sort of constraints
and also reasons why we would
build them in a certain way.
So then in that sense, we would then work out
what size volume we need based on those
that tech scout essentially that we've done,
or sorry, that virtual scout that we've done.
And then we'll add in the
panels that we've built digitally
into those environments.
And you can actually then work
out the distances prior to actually,
you know, your your art
department comes in or anything like that
when you're starting to build out the sets.
And you can then work out
the scale inside the studio
and what your best focal length could be
or what the constraints are on
the the lensing that you need,
you know, and you can sort
of then inform the the DOP
as to sort of what, you know, we
then want to talk and engage the DOP
obviously early on as well.
And also with the director
and what sort of shots he wants,
but and what the constraints are within the studio
or whether we need to go to another
studio and build it out over there.
And then sort of all those
decisions are informed by this tech
that you end up doing after
sort of building the environments.
Yeah, that's cool.
And then the DP, like, do
you find that there are certain
DPs who are becoming specialised
in shooting in these environments
or can sort of anybody with an
understanding of how to shoot
in the physical world
shoot in a digital environment?
Or is it a bit of a specialised thing?
No, I mean, I think that, you know,
for us, we love getting new DPs in
to certainly to train them up.
And also when I say train them up,
we don't obviously have to train them up on this
craft because, you know,
it's all very much translatable, like
everything's very much translatable.
It's more just about understanding the process
where he can relate it to what he's used to.
And then he just essentially
needs to know who he needs to talk to,
to be able to fix something that he needs to fix.
And at the same time, so it's still his eye.
It's still what he wants to do.
There certainly are some
constraints on lensing that that will be
purely based on the framing
and the boards and and then also,
you know, the aperture like, you
know, we really need to always be
on that fine line of
wide open if we can,
purely because of the scale
of our studio and also the
the distance that the panels are from from our
our main actors or whatever.
So the idea is is that we can't
allow those panels to come into focus
because, of course, we have
this moire issue that can happen.
And so we always need to have
quite a shallow depth of field.
Like we don't want to have too
much too much depth of field.
We need to limit that a little.
And so unfortunately, that's sort of something that
not all DPs want to be working with.
And it does give a certain look.
But we we always try to work
right on that edge of moire
because more focus in the
background, more realism creeps back into it.
You know, if it's everything feels a bit sort of
under watery, it's just not looking right.
You know, so it's always it's a very that's why it
becomes quite mathematical.
And the lenses need to be very we have to profile
the lenses and really test the lenses
to make sure that we're not
suddenly on the edge of moire.
And in that project that you and I
worked on, which was super, super crazy,
the lenses had a weird pincushion
scenario where the outer edges were
more in focus than the inner edges.
And so it was it was very strange.
And, you know, these were sort
of old school lenses.
And we really didn't do
our homework on those lenses.
And so it kind of, you know, essentially meant that
we had to push everything further
away from the screens even more,
which meant that we then were limited
by the framing that we wanted to
achieve. So, yeah, so it's not something
that I typically think about is like
the amount of focus across the lens.
I you know, usually you're thinking about depth of
field and, you know, your F stop
and all of that and how much
light you've got in the scene.
But, yeah, I mean, I suppose when
you're really on the edge there,
if it's a bit sharper in one area of
the lens and a bit softer in the other,
you don't want that sharp part of
the lens to start to see the actual LED
points, right? Because that's
what creates the moire pattern.
Yes, that you've got these, you know, grid
essentially of points that will strobe.
Exactly. Yeah. And that's yeah.
I mean, that's sort of the we're always working to
make sure we don't see that.
Yeah. Yeah. Wow. There's so
many considerations there.
And on top of that, you're actually
creating either a 3D or a two and a half
day environment, and I take it
that there's actually quite a bit of
like color pipeline that has to happen between the
start and the end of that,
because you're actually projecting an image that is
quite final at that point.
That's right. Yeah.
How do you how do you design the
pipe there to get the colors accurate?
Well, I mean, for us, you know, the most important
thing is to make sure that this
and anything on the screen just
looks realistic, you know, and natural.
So if you're looking at the screen with your eye,
it needs to sort of feel real
as well. But and I know it sounds sort of, you
know, you think that's what you would
expect, you know, but the problem is,
is that when you're creating digital
environments, you can very easily get illegal
colors, as we'd like to call them,
which are not really real life kind of colors.
And that all is dependent on the
renderer if you're doing digital content.
And that can be kind of quite tricky.
So we do employ an ACES pipeline.
And for us, we've got, you know, OCIO
library, which we output
from Unreal, an ACES color.
And, you know, whether we then
re-project that out onto the panels
as an ACES profile, and then that gets converted
with a look up on the processor
before it hits the panel.
That's sort of something that we sometimes shift
between whether we actually do
the conversion on in
Unreal and then onto the screen.
But the thing is, is that obviously
the panels are all slightly different
in some instances, like as in
different brands have got like different color
with different looks, you know.
And then the problem with the LEDs is that they,
you know, of course, it's an RGB
light, essentially, each one
of those little pixels on there.
And you can get something called
color shift very easily at certain angles
that you look at a panel and, you know, you get
these kind of weird magenta reds
or you get the greens if
you're looking at opposite angles.
And so that that's kind of
got multiple issues with that.
Is that how it translates onto skin tones as well?
That can be a real problem.
And if you start sort of shifting color, overall
color, for instance, on a panel
and what that actually does to skin tone is it's
quite remarkable, actually,
in some instances, you just
sort of hope that it just works.
Everybody's skin tones are slightly different and
the LED panels can really
just just warm it up or just make things look red.
Some skins just goes red under panels.
And it's very, very tricky.
And some skin tones just do not.
So I'm just look fine, you
know, and so you have to.
It's quite a tricky sort of
science to get that right.
So our main aim is to make sure everything's sort
of linear as it comes out of unreal
and into the processor.
So we want to try and make sure
that there's not too much done.
Like we don't want to do much color management
prior to it going to the Brompton
processor, which is the processors that we use to
drive a signal to the panels.
And on the Brompton processor, there is quite a bit
of color management that we
we tend to adjust and and use to help merge,
you know, merge the foreground like the practical
and the background together.
And then, of course, you really
only want the camera to be driving
the color after that fact, you know, with its color
temperature and then the lights
that then change accordingly.
So, you know, just like all lights,
panels can be warmed up and cool down.
And most LEDs run quite cool.
They're very sort of like as a default, they're
quite a cool sort of temperature.
So you can warm them up to
to try and balance it out.
But in doing so, you know,
it's very easy to break things.
And so that's why, like I said, we just try and
make sure that everything coming
out of unreal is very linear.
And so like, again, we transcode
everything that as in textures and or
footage that we're using
into the ACES color workflow.
And then we will then convert it out onto the
panels through the Brompton.
So it's kind of and then we could possibly look at
it in any other look up after that.
But we just want to have as much color gamut as we
can have, you know, and make sure
that those panels are working correctly.
And so, oh, sorry, you go.
Oh, I was just going to ask -
So about that linear signal, I mean, my
understanding of the usefulness of linear
for visual effects is that you can do more
realistic things because that's the way
that light kind of functions in the in the real
world is that it does work like linear.
Not like logarithmic curves.
You don't you know, you don't end up with those
soft roll offs that we then put on once
we're starting to work in the log space afterwards.
Is that why the linear pipe works so well for this
or is it just because it's such a
robust pipeline for visual
effects working in ACES and OCIO?
Well, I mean, I think just obviously keeping it
linear just allows us not to add any sort
of look ups on top so that you can kind of keep the
color information there for as long
as possible and then try and only, you know,
convert it later down the pipe.
And again, like you said, if we have some sort of
roll off that's happening prior to
it hitting the panels, then it's very hard to
invert that, you know, the other way around.
So keeping the signal as linear as possible all the
way through in terms of what you were
saying in digitally, it makes total sense that we
like to linearize everything so that
when our final look up happens in the renderer, we
actually then convert it so we can look
at it. Essentially, we'll look at it in, you know,
Rec 7- Rec 2020 or whatever it is
that we tend to look at it in.
But we will still be working
linearly the whole way through.
And it's also because of the renderers that we use.
So our renderer is the way that they actually
render, you know, whether you're using like
you know, it depends on if it's an unbiased
renderer, which means essentially it just shoots
out light rays just like you would have in reality,
you know, with photons, you know,
essentially, your materials are like balanced and
real world balanced as well.
So they just work far better in terms of a linear
workflow, because essentially computers
work linearly initially, and you don't want to bake
into the the any of the color or anything
like that within your digital, the digital
component of your renderer.
If you start introducing some sort of look up, it
just breaks and then you have to invert it
somehow down the line. So we stay linear as much as
possible all the way up until the end.
And then, of course, as you know, like, you know,
the camera itself will be adding to that
as well. And so, you know, you just have to
mitigate too many things that can change all
the way up until that point.
Yes, yeah, because the camera itself is going to
have, you know, its own ISO natively and it's
going to have its own range of stops that it can
render onto the actual image and capture.
Wow, so much can go wrong,
so much needs looking at.
So do you essentially color
the material in that process?
I know that you're saying that you
are trying not to do very much to it.
But like when I looked at the raw ungraded footage
for the car commercial that we looked at
together, there was nothing like there weren't any
parts of the image that were wildly different
from others. It did feel as though somebody with an
eye had integrated and composited a world
together that was kind of
harmonious and worked color wise.
Choices had been made, I
felt, on the way to the volume.
Is there somebody like a compositor or a colorist
or a visual effects artist who is kind of
responsible for that or is it
something that just happens?
No, I mean, so I mean, to answer your question,
that's probably me that's doing that.
But I think, you know, again, like if we've got a
realistic environment in the background, the aim
is to make sure that that looks like a realistic
environment with your eye looking at it and so that
it needs to actually, you know, it needs to react
correctly to light in digitally, digital
light, at least. And then that light needs to sort
of feel realistic, you know, as you're looking at
it. So that's the first part of it. And then
secondly, when you start
integrating your foreground
elements into that, you know, the there's only so
much light wrapped, for instance, and this depends
on the size of your volume that is encompassing
your person or objects that you've got that are
being lit as well. And so essentially, a lot of
that can be matched, you know, from the lighting
that's coming from the volume. But again, like, for
instance, if you consider that project, we have
the floor, you know, that the, you know, if we look
at the the outback scene where, you know, it was
like a reddish, the reddish color never sort of we
never quite got that one working quite right. So we
can essentially, we're trying to match to the
foreground environment, texture and color to our
digital environment needs to match to that. So
there is a level of that happening in real time.
And in our brain bar of people that we have,
they'll be making those adjustments, looking,
looking at the screen, looking through the split to
make sure that that sort of feels more in line,
especially if it's like a wide global, a global
adjustment to a particular material, for instance,
not a global color adjustment that you do to a 2D
image, but to a specific piece, you know, like
the sand will change the color of the sand to match
to the sand that we have in the foreground.
Those are things that we can sort of do on the day,
and we'll try to get those as close as
possible. And that generally would happen during
the pre-light day, which is you'd hope to be in
that position. And then afterwards, there's sort of
subtle adjustments in color, overall color of
the panels that want to then match with the, I
would be working directly
with the DOP to understand
when he exposes in a certain way what we want to do
to our highlights in our scene, for instance,
to make those work and sort of make them feel
correct as well. And then we have something else,
which is called, well, ICV effects, which is
in-camera visual effects,
but essentially it's a new
projection, kind of like a, you know, power windows
that you can add to the background panel,
and you can actually isolate areas and draw mats
around certain areas and grade those independently
to try and fix that scene where you might see the
integration of the foreground elements hitting the
panel at the background, and so that those can be
graded independently. But we try to stay away from
that for as long as possible until there's just
like a scene that we need to fix because those,
you know, because they actually exist there all the
time, like they're stuck to that piece of the
panel. And so when you start moving your camera
around, you can imagine
that that doesn't necessarily
relate in volume, per se. And so like, as in like
volume tracking, so it
might then stay in a position
and just not be in the right position after that.
So we try to avoid that if we can, but sometimes
you have to actually use that. And also, you know,
this ICV effects option also allows us to add like,
you know, hotspot lights that we want to use for
reflections and added reflections and animate
those that you have, kind of like what you would do
with sort of like a certain LED lighting that
the gaffers use to be able to actually have some
running lights over cars and things like that,
which is quite cool. That's really cool. Yeah,
that's fun. That's really cool. I mean,
it does seem like there's an opportunity there for
a colourist on the set to be part of that
process. And I know that visual effects artists and
compositors are extremely good colourists,
because that's sort of essentially what a
compositor does is matching elements and making
them integrate seamlessly. Can you tell me about
the brain bar and who's on that and what's
happening there? Yeah, I mean, before I do that, I
mean, you're 100% right. And you know, having like
a colourist eye is probably, you know, would be
amazing to have. And I think that, you know,
you know, the DITs, after it's coming out of the
camera, I think is a better place for that
currently the way I see it. And that's purely
because, you know, we want the foreground to
stay married to it. I think what ends up and can
end up happening is that when,
if the background panel is being colour adjusted to
match to what initially is being seen,
the problem with that is that the balancing of
reality can kind of really shift, you know,
in that sense. So when our background becomes sort
of separated, I mean, yeah, there could be very
easy adjustments that we can make. But when you
start making adjustments before it hits the panels,
you're then in this world of changing the
environment to match prior to doing it after
the fact when you have your integrated environment
in there. So it's just been hard to then work out
why certain things aren't working. You know, if we
have a grade that's happening prior to it,
then going, we've tried this and then it going into
the Brompton and then getting
projected up on. And so when something is not
working, it can't all be fixed in grade before
it goes through, because some things, like I said,
for instance, we need to adjust the ground colour,
for instance, specifically on a material, like a 3D
material for that to work, or we need to adjust
the position, the sun position and what that then
ends up changing in the environment by just
changing the lights on everything and how then the
grade fights against that opposite, you know,
because we need to be quite fluid, essentially. But
I think just maybe before that would be okay,
but like, there's so many shots that we're doing
that sort of have to happen. I mean, I think
there's certainly, you know, heel and toe and
levels that need to, you know, mids that need to
be moved around every sort of shot. And that would
be amazing, you know, almost like a one light feel
that would be great, you know, for that to sort of
happen. And that is sort of something that is
happening every time just before we roll. But a DIT
approach after the fact where we have the
foreground in would just be amazing, you know, and
if that was sort of set with the director and the
the DOP to have a look at what's more the DOP, I
suppose, at that point, I think that would be,
you know, amazing, you know, because what that does
is gives the client even more
you know, happiness when they see that. Yeah, it's
perfect. It's great. This is exactly what we want.
And then we sort of designing a look and
automatically on the day, and then suddenly,
you know, just like you'd normally have in a long
form project, it just becomes perfect, you know,
in that I say perfect, but you know what I mean?
And then the clients just even more into it,
you know, and then they feel like and they, you
know, you just they just feel like a way to lift
it off them, because sometimes it might not feel
exactly like or they're not used to that process.
And it just, you know, sets them at ease a little
bit more. That's really interesting. So in terms
of the skills that you think that that colorist who
who could be part of the virtual production
pipeline, they would be more of a DIT to you than
somebody who is skilled in Unreal.
Yeah, I mean, I think so again, skilled in Unreal,
if a colorist is skilled in Unreal,
I think that that's an amazing thing to have. I
think that that would be and integrating those
that person would be would be sort of a priority.
I'd love to have somebody like that in the team
to be able to do that, because I think and I know
there's a lot of tools that are being developed
now where there is more control in grade that
happens in Unreal and as a, you know, as a wrapper
that goes to the next stage. And I think that that
that's sort of quite important. I think one
thing to keep in mind, especially in Unreal Engine,
if we're talking about digital sort of creation
here, the renderer is not a real, it's not a real
world balanced renderer like you would get in say
some new visual effects style renderers or not,
sorry, not new old renderers that have been around
for a long time. So there's a lot of cheats that
happen in Unreal Engine to be able to get that
renderer to, you know, work correctly for
optimization for shadows for bounces and all these
sorts of things. And the problem that ends up
happening is is that you
need to actually understand
what the limitations are of the renderer in order
to look at an image as well and go, this is what's
missing. Because in reality, when you're looking at
something, you just expect things to look real
and when you see it on the wall, you're expecting
things to look real. But the reality is is that
there's a lot of cheats that are happening. And you
need to be aware of those. And if you make
color adjustments to fix things to make it feel
visually appealing, the problem is is that these
cheats that are actually happening are just
exacerbated, they keep getting worse, you know.
And so I think you're not starting, you have to
keep in mind that you're not starting from,
you know, something that looks real. You need to be
looking at something and understand
the fact that what you're cheating already to get
it to look real is happening, you know. And
I think that that's a broader thing to think about.
You know, it's like, oh, we've got one
light that's on and it's casting some beautiful
light on something, but we've turned off shadows
for some objects here and there because it's
costing too much money. Sorry, too much in
processor and it can't render correctly on the wall
because it's starting to get stepy. And so
then we go, well, there's a shadow missing over
there. We've done that on purpose so that we can
actually render it out in real time, you know, and
that needs to be hidden and translated a certain
way. And we're doing a lot of these sorts of things
all the time. And when you integrate sort of
somebody who, well, people need to be aware of all
of those sorts of aspects of what's actually
happening. And then on top of that, add two rather
than just change in a 2D because of course,
as a grader, you most of the time, everything
that you're grading is realistic already is a real
environment is real, you know, and so you just need
to keep that in mind. I think.
Oh, look, that's so true. I mean, in terms of my
experience grading, I mentioned at the start that
I didn't realise that it was a 3D background. I
didn't realise that it was virtual production
until halfway through the grade because you're 100%
right. You expect to see reality and you're
very forgiving of what you're seeing because why
wouldn't it be real? But then once you start
looking, you start to see things. And I'm not
saying that there was anything glaring or major,
but I just started, you know, OK, so this is
constructed. I would like to see some evidence
of that. So I start looking around and then I find
little things. And then, you know, you start to
get in there surgically and shape and reduce
anything that might not look real. And the goal
is always realism as opposed to some kind of
stylised treatment or just like you say,
just changing sometimes as a colourist, you go,
well, we don't want it to look like the offline
did, so we better do something. But, you know, that
might not be the right approach to this.
Yeah, because I could break it. I could just break
it like immediately. And you're like,
oh, actually, now I can see why it's digital now.
Yeah. So like a light touch and not doing something
unless you have to would be probably the best way
forward as a colourist in that scenario
is not to do anything unless you have to. But
again, because like I do, you guys add
this massive add to so I'm not saying sure, but,
you know, we we couldn't we can always
do the grade afterwards. I'm saying what what do
you do during the production? You know,
how do you add to that process? Because you can
always grade it afterwards. And I think there's
like, you know, every time I've ever graded
something that has been virtually produced,
whether it's a virtual production or
a 3D animation or something, you know,
it always adds to it because it gives it something
cinematic and it gives it something filmic that
it just needs to give it coherence and to make it
feel correct to to be part of our screen language.
But it's slightly different to the job of creating
the thing to begin with. It's 2D. It's about
it's just a different approach, right? Vignettes
and exposure and, you know, tones in different
parts of the curve and, you know, generating that
graceful curve. You're just thinking about a whole
different bunch of considerations as a colourist.
Exactly. Yeah. So what happens in the brain bar?
Well, we've got we've got so we've got a operations
controller, I guess, is what we call him head of
operations, Johnny. And he sort of so I guess to
start from the beginning, I think Unreal has a
multi-user approach where you have one scene that
is the environment, for instance, and then you
have a server that's sort of hosting that
environment essentially. And then you have these
multiple machines that are all connected to that
environment or that scene. And those multi-users
then can be identified as separate machines that
exist working on the same scene at the same time.
So it's kind of slightly different to how we would
do visual effects projects in that sense,
because things, you know, when something changes on
a scene that changes and then, you know, you
have to publish that change down the line. So in
this instance, you kind of have a bunch of
collaborators were working on one scene at the same
time. And that's part of Unreal Engine's
architecture. And it's something that's sort of a
game driven architecture, but it works really,
really well in this instance. So we have someone
who's head of operations, and he would be he or
she would be integral in hosting this scene. And as
the server, it would also be where all of our
other sort of peripherals attach to. So like, for
instance, the virtual, sorry, the motion capture
Vicon system that then is plugged into its own
suite, and that has another computer running that.
And then that is then streamed into Unreal. So that
gives us the real world position of the
camera. So then again, in the scene, the operations
managed would then have this virtual camera that
we sort of liken to the real world camera. And that
has these the crown that I don't know if you've
seen it, but obviously a little, you know, crown
that exists on the top of the camera, and has
little dots on top. And when that moves, the Vicon
system is sort of plugged into Unreal and
streamed into Unreal, and it tells it where that
position is in space. And then Unreal,
in that Unreal scene, we can move that camera
around in the scene to reposition where we want
it to be existing. So then those are sort of two
roles there at that point. So you've got the
the Vicon operator, and who's also like looking at
the cameras because they overheat,
over time, you know, we've got things in the scene
that are occluding that are in front
of those cameras that can cause the tracking to
mismatch, there's quite a few things to keep in
mind over there. And so the tracking of the cameras
is super, super important. And so that
Vicon system, we're constantly, you know, it's
harsh, because, you know, certainly for the gaffer
and the grips, you know, where they want to put up
flags, and they want to put up, you know, like
big bounce slides, or they want to put a bunch of
stuff up to also help with the scene. When those
get in the way of the Vicon cameras, as well, then
that so we have to work together to make sure that
that doesn't become an issue. So anyway, so we have
the Vicon controller as well, and they operate
independently. And then that plugs into the main
scene with the operations manager running that.
And then we have two other, generally, it depends
on the size of the project, but then we have two
other positions there where we have the head of the
virtual art department, so the VAD supervisor.
And then anytime there's an update to the scene,
which were at least two objects and color and
things like that, that we just talked about by
updating the sand or updating the plants and
repositioning plants here and there. And we want to
maybe say, let's move an object in the scene
quite quickly. That will be taken care of by the
VAD supervisor, who then repositions all those.
And then we'll have an extra person that would be
essentially working independently in another
environment, and then that might be publishing new
objects into the scene, like brand new assets.
And so that would then be published into the VAD
supervisor's scene, and then he would make a
decision of when that gets published before we
change the setup. But of course, you can imagine
that when you've started shooting something and
then you add something new to it, and it can affect
everything. In fact, the optimization, the speed,
the lighting, absolutely everything. So we have to
be very careful about that sort of stuff. But
generally, that's sort of the brain bar and those
guys. And then I would be in direct contact with
the operations manager in general and also probably
the VAD supervisor. And then especially if we do
lens changes or we make any sort of changes like
that, or if the motion capture volume is not
capturing correctly, all these sorts of things,
then I would then translate all that information
back to the first AD and/or the DOP and director
so that we can be sure that before we do a take,
we've got that take. That take is sort of
working. And so there's a few steps that we need to
make sure that we're locked and ready to go
before we can actually roll on something as well.
Oh, cool. Wow. That's a pretty big team without
even thinking about the productions team. Yeah.
Well, I mean, digitally, yes, you're right. It is a
biggish team. I mean, also we record every single
camera as well. So every single camera that we're
sort of filming, we do an internal recording of the
3D positioning of that in case we need to
re-render things later on down the line. And that
needs to be time coded. We need to have things
genlocked so that the frame rate is running
optimally with our motion capture cameras,
with our actual, sorry, production camera, and also
that the screens are all locked together
at the same time. So we have quite a few things
that can go wrong and people that are in place
to keep an eye on those particulars. Oh, wow. I've
got a couple more questions for you. I know
I've kind of taken up already an hour of your time,
which I'm very grateful for. 11.08. Okay.
I would love to know a little bit about your
background. Okay. What brought you to the
position that you're in to be somebody who can
understand how this emerging technology can be
implemented in this way? What's your history? My
history? Okay. Well, I was a painter, a traditional
animation artist. Then I went to digital sculpting
or traditional sculpting first, then digital
sculpting, became a character artist way back,
studied at the Vancouver Film School quite a long
time ago now, and then was in sort of pre-rendered
game cinematics back then, doing characters and
sort of lighting and texture work, and then moved
into the film side of things, and then became more
of a lighter, so a 3D lighter and lighting scenes
for visual effects integration into live action
plates, and then continued doing that for a little
while and became a CG supervisor. So I ran
the floor of the 3D artists and implemented new
techniques. So back in the day, I was into fur
and feathers and hair and all these sorts of
things, and there weren't pipelines in place in
those days to be able to roll those out that were
like plugins to make those work, so we had to use
multiple bits of software to make them work, and
then I transitioned from that role as a CG
supervisor, and then I just wanted to, I was just
more into practical sort of filming as well. So I
went to AFTRS and studied cinematography, and I was
already sort of a visual effects supervisor at
that stage as well. Then I went to AFTRS and
studied cinematography in Sydney, and I was working
on film, still work with film back then, which was
great, and then transitioned to digital cameras,
like the red one, and you know sort of early days
with sort of digital cameras back then,
and then sort of pushed cinematography for a little
bit, went to a few different studios, it's
like a combination cinematographer and heading up
their 3D departments as well, and got into
sort of virtual cinematography, said doing sort of
full 3D environments, but you know working with
the camera because I was quite skilled at that
point in practical cinematography, and then sort
of started directing a little bit more, and then
sort of started my own production sort of company
as a visual effects production supervisor, where I
sort of worked mainly on long-form projects,
doing second unit directing, and just pretty much
being on set all the time working in visual effects
in long-form, and pushing motion capture as well,
and then sort of got back
into working in Steelbridge
when we sort of pushed this tech, and I had a LiDAR
scanning company as well, because I've always been
into the LiDAR, photogrammetry, drone photography
as well, and all those sorts of things, so I'm
creating digital assets, just all sort of work
together, and it all just sort of put me in the
right space to be able to sort of do this kind of
job, which is pretty much all the things you need
to know sometimes. Yeah, so you've done a very very
good mix of kind of physical work and digital
equivalents along the way with the physical
painting and sculpting to the digital sculpting,
and physical lighting to digital lighting, physical
camera to digital camera, so also working in
sort of emergent technologies as you go. It does
seem like a pretty pretty special combination
that can allow you to understand so many things
that need to come together.
I mean when you were doing the onset work in visual
effects, when you're talking about second
unit VFX shooting, with that you know you've got a
shark movie and they've shot the principal
photography, and then they've got a really
important scene with a shark that needs to
come up underwater or something, and you're there
with the sort of gray shark head and capturing
everything you need to create the 3D shark. Is that
the sort of thing we're talking about? Yeah,
exactly, yeah, and I still do that, and that's sort
of kind of a big part of what I still do,
so but you know I'd have wranglers that I would get
to go out and you know do that sort of stuff.
I wouldn't be the one holding the shark head, but I
would certainly be the one that decides on how
the methodology to be able to accomplish the shot,
and that's why doing sort of long form,
you know I might sort of you know break down
scripts and we'll look at what needs to be
visual effects, traditional visual effects, you
know and integration in that sense, and then also
what may require virtual production, and so I'm
sort of in a good position to also sort of break
down scripts and look at them because there's you
know it's yeah there's no as virtual production
is kind of there's only certain circumstances where
it really is required, you know in certain
cases where it's just it just works, you know, but
in other senses you just need locations,
you need movement, you need much larger expenses to
be able to do that, and sometimes you know
shooting on green screen is the best approach for
something, and sometimes shooting in a location
is the best approach for something, and so you know
there's quite a few reasons why you would
and wouldn't want to do that, and so that's sort of
what I'm still doing, so I'd be you know I work
with big studios as well, and we'll talk about
whether VPP is an option, but you know inevitably
I think it's not as mainstream, it is sort of in
some studios overseas for some instances, but you
know there's certainly type cases for for why you
would want to use virtual production, and I think
for commercials it actually lends itself really
well because you can just you know bump out heaps
and heaps of environments within a day, and there's
just you know the cost works out far better, you
know, and so you know the quantity, the scale
economy works just sort of well in that sense too,
so in that respect it works, but I think you know
long form again like I said I'm a production
visual effects supervisor, so I look at those sorts
of options as well and make those decisions.
And so you would, if you're on set you would be
looking at okay you've shot these lenses,
we need to profile them, and you've shot under
these lights, so we need to get reflection passes
and have the balls go out and all of that, and yeah
do the LiDAR scanning, and I can really see
how that works right in with doing virtual
production, but also like it's really great
that you know that it's not right for everything as
well, I think that would be, I don't know if
you were the studio trying to sign off on something
you don't want the hard sell, you just want to know
what's going to work the best for the scene that
you're trying to shoot. Exactly yeah and no one
really wants to you know put their hand up to
decide to do virtual production and then it just
breaks and doesn't work you know either, so I think
just you know if you're trying to sell a
technology you know you better be in it in the
middle of it to really know whether it can be
done as well and what the obvious
problems are but also the things that
the gotchas that can happen and they can happen you
know very quickly and technology is you know
is reliable but it's not always reliable, that's
all I can say, so you have to have contingency
plans and everything yeah exactly. And so thank you
for that, the final thing that I wanted to
ask you about was as a colourist what do you see as
being the most useful things a colourist can
bring to a virtual production, whether that's if a
colourist wanted to get involved during the
production side or colourists who are working on
something that's already been shot and edited.
Oh I mean that's a, I know that's a tricky
question. Yeah I mean I think for me
I think like I was I kind of alluded to earlier on
I think just what LED panels do to skin tone
is just is really really tricky and is certainly
something that needs to be you know given its
you know just the time in fixing and looking at and
if you as a colourist are and this is now
posts now at the end of and you know you're sort of
grading a final spot or whatever
those are the things you need to sort of kind of
look at and understand I think so understanding
maybe the limitations or the things that can cause
issues within in a virtual production environment
are things that would be so if you have this sort
of knowledge and the things to look at in terms
of trying to fix those are really those would be
super advantageous for for for you as a colourist
so that would be you know in my opinion is you know
that there's colour shift that happens too
so if you look around certain areas and you'll
notice that some things are looking greener and
redder in some areas again those are things that
aren't real they shouldn't be there and those are
things you want to sort of grade out so I think
just understanding what can happen in a virtual
virtual sort of environment and then looking at the
the pictures and trying to be critical you know
not overly critical but I mean critical in that you
know those are things that if you just
automatically fix those and hit those sort of bits
you're in a really good place to then do your
your creative grading on top of that you know so
it's almost like the one line which is almost
like to fix to fix the led issues that can happen
on the day you know and and for me that would be
great you know like for a colourist to actually
understand what could be problematic you know
and then I think you know moving ahead and and sort
of being sort of an earlier part of the pipeline
I think and I've worked with a few colourists that
have been really interested and just want to be
there you know and go can we can I just be there
while you guys are doing some testing and sort of
maybe we can get some panels you know like some you
know balls out and see if we can like plug
into that and just see what we can do with the
image and and try and understand it I think all
of those are invaluable for us as well and and
would be the kind of experience that would really
help you even afterwards so like post you know in
the post process just being there on those
shoots and understanding what's happening is
actually a really good place to be and then help
you down the line fixing sort of stuff yeah yeah of
course I mean when it comes to the skin tones
we're looking at the light that's reflected from
the panels that then hits characters here in the
space right so instead of being hit with like the
full spectrum of colours in the light there would
be certain colours that would be enhanced and
others other parts of the tonal range that might
not even be there right so you might get like a a
flatter skin tone instead of like all of the
variation that you would get under natural light is
that the kind of issues yeah that and also
there's certainly a bit of colour shift that can
happen and there's sort of you know you mentioned
red skin tones yeah for some people and some will
just go greenish or not have any change and then
they've got two people standing side by side like
for instance in the one job that we were on like
you had those two people on the beach and the guy
he kept going too pink and then she had a different
tone like more melanin in her skin and hers sort of
almost went green and there was like a shift
that you needed almost do between the two of them
and it's when they walk through underneath the top
panel you if you just move your hand and go along
there and along there the colour is slightly
different over top of the person all the way
through which is problematic very problematic
you know and we're always chasing that sort of
stuff as well so that's certainly something to
so keep an eye on um i would say and and also as a
as a as a colourist realize that
you know what's happening is not just an a natural
artifact it is a mistake well it is something that
shouldn't be happening if you know what i mean and
not just accept things that are happening and
just try and fix them because ultimately you know
when i'm in the grade or if uh you know
Colin or the director is in the grade if it's Colin
it's different but if it's maybe a director
that's not used virtual production before they
might miss those sorts of things you know and
so then it needs to have say the virtual production
supervisor or somebody that already has an idea
of what they should be expecting in those instances
and try and correct for those you know i think
right so it's just being being aware that things
aren't natural to begin with i suppose you know
i'm thinking back to a um a story that a friend of
mine told about checking renders for quite
a large 3d film that was very popular and saying
that the hardest thing about checking those
renders was knowing that if a character had been
removed for a scene their matte might still be in
there so they might be like a little ghost of a
character like lingering and and anything could
happen anywhere um in the shot so i suppose you've
got to be in that mindset a little bit when you're
grading virtual production if somebody moves
through light normally as a as a colorist i
would say well they're moving from you know
daylight into tungsten we want to see some
we want to see those because obviously that's been
done on purpose if it's in front of the camera
it's meant to be there but you're saying no we need
to be a bit more critical here we need to say just
because it's there doesn't mean that it was meant
to be we need to really interrogate if somebody's
skin tone changes as they walk from one side of the
screen to the other that's right is that because
they're walking past a window or is it because the
panel was outputting a slightly different shade of
white there yes exactly and that happens on the
floor too you know if the floor is in the
practical floors in shot and then the edges might
have slightly varying color adjustments on them
and it's not something we can fix on the day um but
again look you know panel technology is getting a
lot better they're introducing white pixels into
the leds now which really really helps a lot
especially for overall exposure but also the the
the angle of the so the
the field of view essentially
of these lights is broader and that actually helps
with a lot of the the color shift as well
and the tighter pixel pitch actually helps with
color shift too but then you run into sort of
issues with brightness levels so i mean the the
panel technology is getting a lot better
and so and so that does help with that but i would
always be on the on the side of trying to keep an
an eye on that and also you know the panels
themselves i mean there's a lot of light pollution
that can that can exist and hit the panels you know
and so you have a you know you have this
coating on the the the lights on the leds and the
better the coating and they've really
improved their technology in the coating the better
the coating the less light pollution
from practical sources or even from other panels
that you've got in the scene are not lighting the
the the panels themselves that is then losing
contrast and when you start losing the contrast
in your panels the blacks are getting lifted up
even more and obviously the idea is to retain the
blacks as black as possible and then when you get
that light pollution on there we can get shadows
you know like there's we've added a practical light
in in the shot and it actually then starts
showing off the edges of the panelists you know
because of the the reflected quality of the
panels themselves the physical kind of panel so you
know those are the things that you can actually
end up seeing in the screen but not really realize
it so you just also need to be aware of those
sorts of things and look at them and go oh actually
the blacks are just looking a little
too milky here for some reason they weren't like
that in this particular shot just before
um and it could have been that we added a practical
light in that instance and that
is also something to keep a real keep an eye out
for yeah i suppose um the colorist then can ask
as well because you do have so many elements that
can be separated potentially you've got some mattes
for the foreground and the background and if the
backgrounds are starting to look a little
milkier than the black point in the foreground you
know it'd be much more simple to request a
matte and pop that in to adjust the background rather
than try to spend your whole session
you know doing some complicated animated shape that
is a nightmare for everyone to sit through
yeah well do you often have a lot of mattes well
actually and that's funny you say there's this
new tech and unfortunately it's only the red camera
that can actually do subframe and actually
maybe you know in maybe there's more cameras that
can do it but the subframe technology is something
that we're looking into quite heavily at the moment
which means that your camera can pick up
a subframe and the Brompton processor can add
inject a subframe into the panel so you wouldn't
you can't perceive it with your eyes but
essentially you're adding an extra frame
and they have green screen in the background and
your camera is picking up subframes so
so you say this for instance is at 25 frames per
second that you're rolling at
it's essentially picking out 50 frames a second but
still interpreting it as a one over 50th sort
of shutter angle for a 25 frames per second sort of
film so it's not ultra sharp no it still retains
its motion blur across the the frames but then you
can have separate outputs coming into your split
where you can see straight green screen on the
panels and that action is happening and then this
on the other output that's coming out is your
beauty which is what the panels are looking like
at the same time which is amazing you know and then
you can then use that to create your mattes
from and inject that into your beauty after so you
always then have a mat running at the same time
yeah it's amazing i can't see any reason not to do
that that just sounds so useful i know it does
it sounds good the problem is is signal and so you
know we wouldn't have to do it from unreal
so we're spitting out 25 frames per second which is
fine or we'd probably go to 50 and then we just
roll at 25 but then when you inject an extra frame
you're actually adding so we've got say a 4k
image that's being projected onto the screen you're
then running it at 50 hertz or 50 frames
a second on there so you need to have that much
more signal going through the panels and so you
have these bronton processes can only output a
signal of 4k across these panels and then it's
all about math to work out you know what the signal
ratio is through and you want backup and
you want you know like so it's the power ratio as
well so we have the ability to have 50 frames per
second on 4k which is good but if you were trying
to shoot at a different frame rate and then it goes
up over a certain frame rate then the panels aren't
able to run that way so you then need another
processor which is fine it just implicates the cost
goes up every time we have more of these
processes which are very expensive pieces of kit so
nothing comes for free in terms of options
no nothing comes for free exactly yeah yeah so yeah
that's super interesting well as as useful as
it would be perhaps it can still be somebody's job
to clip around those elements in roto yeah i mean
look i think it's actually super super helpful
and i'm sure there's going to be projects
where we just go we just need it and and that's
we'll just you know pay for that extra
processor and just you know make sure we've
got the power for it and i think we'll be good
you know don't take down brisbane's grid yeah well
we've got at steel bridge here we've just had
like another we've got like three 32 phase
outs or ins or whatever so we've
really like you know bumped it up a little bit more
so for a small studio so also a good place
yes also a good place to uh to go and charge a
tesla if you need to well yeah let's not tell
too many people none of us are driving tesla's out
here we should do that it's a great idea
we'll take that on be a very quick charge yeah look
thank you so much bryn for your time i
really appreciate it i think it's a very
complicated bit of subject matter and
i'm sure we've only just scratched the surface on
understanding how it all works but definitely
know a lot more now having spoken to you about it
than i did beforehand because it's a totally
foreign different universe to me so i can't
thank you enough for explaining some of these
concepts and i really look forward to seeing how
people can integrate the the world of color
grading a little bit more into virtual production
as as the technology begins to become more
mainstream and you know more widely adopted yeah
well it was a pleasure chatting to you Kali so
you know it was good and i'm glad i mean you know
like i said you know the more people that
are on board especially you know colorist as well i
think it'll just make the the final
results so much better and i think that's what we
should all be striving for so happy with that for
sure absolutely well thank you so much for Mixing
Light this is Kali Bateman see you next time yeah