“Zack Snyder wanted
Man of Steel to appear very natural
because there’s some very fantastical things in there and he wanted
people to suspend their disbelief, and we the visual effects team had to
make it as easy as possible for them to do so.” So recounts overall
visual effects supervisor John ‘DJ’ Desjardin on the philosophy behind
Man of Steel’s visual style.
Superman takes to space. VFX by MPC.
Desjardin notes that the intent was to shoot a more handheld (the DOP
was Amir Mokri) and documentary-style film than previous outings in
this comic book character’s ‘verse. “We had to think about what that
would mean since we also had to photograph some crazy action,” says
Desjardin. “So for a lot of the previs we did, we’d start to think where
our cameras were and where our cameraman was. A lot of the rules are
the
Battlestar Galactica rules for the space cams that Garry
Hurtzel developed for that mini-series where we want to make sure if
we’re translating the camera at all it makes sense. Unless the action is
so over the top, like in the end where Superman is beating up Zod – we
had to break it a bit.”
fxguide talks to the major players responsible for bringing to life the visual effects of
Man of Steel:
overall supervisor John ‘DJ’ Desjardin, and Weta Digital, MPC and
Double Negative. With so much work in the film, we delve down into just
three of the many tech accomplishments:
1. The tech of Krypton
2. Live action and CG takeovers: the Smallville confrontation
3. Destroying a city: the invasion of Metropolis
And we also take a look at PLF’s previs work, Scanline’s tornado and oil rig effects and Look Effects’ work on the bus crash.
The tech of Krypton
A war brews on Krypton before its destruction.
Act I of the film takes place on Krypton, facing destruction from an
instable core. Weta Digital created alien planet environments, creatures
and also the key means of display – a technology the filmmakers came to
call ‘liquid geo’ meaning liquid geometry. “Basically,” explains Weta
Digital visual effects supervisor Dan Lemmon, “it’s a bunch of silver
beads that are suspended through a magnetic field, and the machine is
able to control that magnetic field so that the collection of beads
behave almost like three-dimensional pixels, and they can create a
surface that floats in the air and describes whatever the thing is
you’re supposed to be seeing.”
Listen to our fxpodcast
interview with Weta Digital visual effects supervisor Dan Lemmon.
The liquid geo devices appear in the planet Krypton scenes, as well
as later sequences on the Kryptonian ship the Black Zero. Similar
technology making up a panel display resembling a Greco-Roman bas-relief
– but achieved via a different method – is present in a scene in which a
hologram of Superman’s father, Jor-El, explains the history of Krypton
to his son.
In creating the liquid geo which took the form of anything from wide
planet views, x-rays, displays on floating robots and even to depict
Jor-El communicating with his wife Lara; Weta Digital took these steps:
1.
The look – The beads, which up close would appear
to be pyramids with a slight bevel, were designed to create a surface
of the object they were depicting inside some kind of console.
“Essentially we would have the normals of the objects that we were
targeting provide a simulation with an orientation that one of the most
dominant sides of the pyramid would align with,” explains Weta Digital
lead FX TD Brian Goodwin.
- See how Weta Digital created the liquid geo and History of Krypton sequences, thanks to our media partners at WIRED.
2.
Modeling and animation – The models used for
animation ranged from purpose-built (Lara’s face) to ones appearing in
grander scenes (such as approaching scout ships). Says Goodwin: “We had
to develop a pipeline to bring in assets, so instead of going through
the route of reducing the polygon count to something usable what we
would then do – you would take the model in whatever way it was made and
just scatter discrete points onto it, and extract the matrix onto the
animation and copy these points onto the matrix and have these sparse
points behaving in a way that the model would.”
“We had animation provide us with geometry that we would then track
beads on,” adds Goodwin. “Those beads would then be turned active in
front of the actual console and the console would decide what beads it
needed to provide to actually draw particles from the actual earth we
described or this invisible bowl.”
3.
Simulation – After animation, artists ‘copied’
little beads onto the animated geometry for a pre-sim’d lighting version
to get approval on how the object would read. Sims were then run “on
all the targets which would be discrete beads floating around on top of
the surface which would have its own set of parameters,” says Goodwin.
“The bead size or the turbulence that would crawl along the surface
constantly updating the orientation was based on the normal provided by
the surface. That was then saved to disk and we would use that sim as
the final target for the simulation.”
Liquid geo console displays can be seen in this pic from the Black Zero on earth, along with Metropolis backgrounds.
Something different
Weta Digital senior visual effects supervisor says the liquid geo look
“stemmed from an idea that Zack Snyder had. He wanted to do something
that was interesting and different in the way that you saw information
presented. He didn’t want to do the typical screen. So one of the ideas
Zack had was to make it a little bit more tactile – we looked at
different things – the idea of that pinboard that you put your hand in
and you see the shapes sort of form was what we had in our minds, but
the more we looked at that the more we realized you can’t do just
something like that it’s too simple and limiting – the shapes really
need to transform – you need something that has that look but be more
liquid.”
The sims were based on a fluid sim. “We used Houdini’s internal FLIP
solver that gave us the pressure, the sense of volume maintenance,”
notes Goodwin. “We’d have the console sim inside an invisible membrane,
and there would be currents that we would describe with what we would
find aesthetically pleasing within the shot.”
“The console was like a cup turned to the side,” adds Goodwin, “so
whereas gravity would be Y pulling ‘down’, in this case the Y is facing
into the back of the ‘cup’ console, ” which means essentially gravity is
pulling from the back of the cup towards the actors watching the
display. The beads then fall (after they pass some threshold) – towards
the inside of the geometry. It is as if one poured the beads into a
glass bowl, all filmed from below the bowl looking up, – except
everything is turned on its side, so the beads fall towards the viewer,
and the geometry of say a planet Earth is a hollow glass bowl between
the sea of original beads and the viewer. “We sort of reversed it by
having gravity faced towards the front of the membrane which would be
the meniscus and the surface of the actual water was the front of the
console, so we would have a constant force pushing towards the front
which would give us a sense of a flat surface with water traveling
around on it, but this surface was never rendered.”
4.
Noise – After simulation, Weta Digital ran every
bead through a temporal filter to remove jitter. ”Even with the highest
RenderMan settings we would still face a lot of noise, and that led us
to taking out all the noise within the simulation,” continues Goodwin.
“Even the most subtle twist of the bead half a degree, 2 degrees, would,
because of being mostly specular, would result in seeing a completely
different point within the IBL (Image Based Lighting) and that would
create a tremendous amount of variation between the slightest bit of
movement. By filtering it, it softened the whole piece out, to the point
of sometimes we needed to get a little bit of grittiness back in
because smoothing out the beads too far would look too boring.”
The team could control the flow from back to front and back again.
“We allow the simulation, to go through a series of noise fields to make
it a little bit more interesting and then join the target,” says
Goodwin. “Then once it’s joined the target it would essentially no
longer be registered in the simulation. Once the target appears we
release it and it finds its way back into the simulation, by using the
opposite force – we essentially use level sets to create some sort of
pressure and it would know when it was inside or outside this world and a
bunch of rules that would dictate whether it was allowed to be
outside.”
5.
Lighting and rendering – Lighting solutions were
from taken the set. For the consoles, Weta moved to the next level of
RenderMan to take advantage of improved raytracing and instancing
objects. Motion blur was also a particular challenge. “We had the
traditional motion blur – in that our particles do technically move,”
says Goodwin. “We did a test where we rendered the objects and we would
compare the motion blur represented from the object’s motion blur
literally straight from animation and we would line that up with the
render we would get out of the beads. In some cases we would have the
vectors that were provided to the renderer shortened ever so slightly,
so that we would have, as the beads form a target across two frames,
would result in spikes within the motion. You’d have a bead that travels
across the length of the frame and you’d have a long streaky specular
highlight – ultimately it’s shading at one end and smudging it and in
that case we’d shorten the motion blur so it wouldn’t create bright
little spikes.”
The History of Krypton sequence with VFX by Weta Digital. The studio also worked on the liquid geo communication shots.
The
history lesson bead shots were created slightly
differently and without an underlying sim. “In addition to working out
all the technical aspects, just figuring out aesthetically and
creatively what actual images we would use to tell the story of Krypton
was really important,” says Lemmon. “We did quite a lot of concept art
based on various sculptures. We looked at bas-relief from the
Rockefeller Center, we looked at Greco-Roman references and explored
those kind of aesthetic looks but applied to spaceships and alien
planets and alien technology – if you were depicting a sci-fi world
through the medium of stone sculpture, what that might look like.”
Weta Digital had originally planned to do these shots with the liquid
geo simulation engine as well, but ultimately the look required was a
different one. “It’s more of a relief style,” says Goodwin. “It actually
looks like the space that each object exists within looks like it
exists within a world that’s flattened. The idea was that it went from
an idea of being a simulation forming things to being a relief. If you
look closely the background is flowing with the text and graphics – the
beads travel along it – but outside of that, things weren’t actually
moving to and fro.”
The action for the history lesson was animated based on greenscreen
performances by Russell Crowe and Henry Cavill in what became, according
to Goodwin, ‘humongous spaces’. “It was traveling hundreds of thousands
of meters in a Maya scene, then we sent that through a projection,”
says Goodwin. “It was all animated in world space and would then send
that through a transformation which we would then project onto a back
wall and relief it. We needed to represent all the information in a
confined space.”
The liquid geo shots on Krypton occur while the planet is both under
siege from Zod’s crew and as it becomes unstable and, ultimately,
implodes. Before that happens, wide views of Krypton depict an alien
atmosphere, which are mostly Weta Digital environments, spacecraft and
creatures.
Faora Zod on Krypton, note the women’s suits are real, the men’s are digital thanks to Weta Digital.
One shot of Jor-El riding a winged creature made use of a buck and
gimbal set up to replicate the move fashioned in previs. “We shot
elements of Russell Crowe in his flying costume on that buck,” explains
Lemmon, “and using previs as a guide tried to match both camera and the
movement of the gimbal to the previs. Then we put those elements onto a
digital creature and a digital world. But of course some of the stuff
that moved in such a way that it wasn’t possible to get those movements.
There were shots where we transitioned in and out – we go all digital –
then Russell Crowe for just five frames or so, and then back to a
digital character.”
For some dramatic shots of Zod’s ships approaching the house of El, Weta Digital referenced scenes from
Apocalypse Now.
“Zod flies in these attack ships and his descent on The House of El was
modeled on the Rise of the Valkyries sequence,” says Lemmon. There’s
actually a shot that everybody thinks is in
Apocalypse Now, but isn’t, it’s in the
Apocalypse Now poster that – the sun shot – the ships flying out of the sun.”
Suits of armour worn by characters on Krypton were mostly CG
additions to the actors wearing gray suits with tracking markers
(although female characters wore practical armor on set). The tracking
of these shots is therefore particularly complex to match all the
movements of actors sometimes engaged in hand to hand combat such as
Zod’s attack after Superman’s pod is launched from Krypton.
A phaser battle contained a specific look for blasts with plasma
residue. “Those were mostly Houdini simulations,” notes Lemmon. “We
wanted to avoid a straight laser beam and do something that had a little
bit more interest in it. The idea was the beam moves through the air
and charges and ionizes particles in the air. In Krypton there’s
particles that float in the air the same way that dust does here, but we
treated them as if they were heavier and got more excited by the beam.
As that the beam moves through the air it glows and starts to swim a
little bit and leave that residue, particularly when it hits somebody.”
Zod and his followers are captured and banished to the Phantom Zone (also completed by Weta Digital).
Aerial battle shots employed Krypton’s hazy environment and shafts of
light through rock pillars to add depth. “In busy sequences like that
it’s important to compose things so that you can actually see what’s
going on and see who’s good and bad and who’s winning,” states Lemmon.
“One thing that drives me nuts in big action sequences is when you can’t
actually – it’s just noise – and you can’t see what’s going on.”
Later, as the planet begins to destroy itself, the studio worked to
show various angles of the destruction including a ‘from space’ view.
Lemmon says he enjoyed “figuring out how it would look – playing it out
as a geo-thermal event that’s influenced by the planet’s magnetic field,
and maybe have it collapse along the equator rather than blow out
spherically and implode first then explode afterwards. Playing around
with those ideas was a lot of fun.”
Live action and CG takeovers: the Smallville confrontation
VFX by MPC.
A major challenge faced by the filmmakers and visual effects crew on
Man of Steel
was to realize elaborate close-combat fight scenes between Superman and
his Kryptonian foes. They wanted to take advantage of digital effects
to portray superhuman strength and powers, but without what had been
perceived previously as ‘cutting’ from live action to an obvious
digi-double and environment. Instead, the filmmakers wanted these shots
to be executed as seamless takeovers.
Desjardin explains: “When we do these fights and these hyper-real
things, we don’t want to do the traditional, ‘OK I’m a cameraman, I’m
shooting a clean plate, I’m going to pan over here to follow the action
that’s not really there yet but we’ll put the action in later. Because
that’s us animating the characters to the camera. So we would do that
animation with the characters – grappling, punching or flying away – and
we would take the real guys up until the point until they were supposed
to do that and we’d cut. Then we’d put an environment camera there and
take the environment. And then a camera for reference of the actors and
get each moment. So then we had a set of hi-res stills for the
environment and the characters. Then in post we take the digi-doubles
and animate them according to the speeds we want them to move in our
digital environment.”
See parts of the Smallville encounter in this TV spot.
This approach was pioneered for the Smallville encounter in which
Superman confronts Zod and his crew after they have threatened Martha
Kent. They fight on the streets of the town and are further attacked by
the military via A-10s and ground assault troops. MPC handled visual
effects for this sequence (in addition to many other shots in the film
ranging from Arctic scenes to shots in the upper atmosphere when Louis
and Superman are taken to the Black Zero).
In order for the seamless takeovers to occur – and for the shot to
continue with a pan, tilt or other movie – a new capture and
post-production process was proposed by MPC visual effects supervisor
Guillaume Rocheron, in conjunction with Desjardin. Here’s how it broke
down for a typical Smallville shot:
1. The shot would be
previs’d and particular fight choreography for the fights established by stunt coordinator Damon Caro.
2. Knowing from the previs the shot that was required,
live action portions
of the scene would be filmed in little pieces. “If say Superman was
being punched and would land 50 meters away, we would shoot our start
position and end position, and then bridge that gap with the CG
takeovers,” says Rocheron.
3. A camera rig dubbed the ‘Shandy-cam’ (named after on-set VFX coordinator Shandy Lashley) obtained
keyframes of the actor.
“It’s a six still camera rig that’s built on a pipe rig so that you can
run it in at the end of a setup and get stills of keyframes of a
performance or an expression,” says Desjardin, “and then we could use
those hi-res stills to project onto the CG double and get really
accurate transition lighting and color – right from the set.”
MPC also handled the energy masks worn by the Kryptonites.
4. On set, another camera rig was also used to
capture the environment.
“We ended up calling it Enviro-cam,” notes Rocheron. “It was a rig
where we mount a Canon 5D and a motorized nodal head, and that allows us
to capture to full 360 environments at 55K resolution for every single
shot. The capture time is very quick – we were taking between 2 and 4
minutes for every shot, so it was really easy – the same way we capture
HDRIs.”
“The sets are there so why not capture them?,” says Rocheron. “It
basically allows you to film what is not filmable. Here there’s no cuts
with no interruption. We also did a lot of entirely digital shots which
had no live action. So we had our Enviro-cam, so we used that to capture
the environment rather than a plate and we could put our CG characters
in there.”
The set capture resulted in lighting and textures that could be
re-projected onto geometry (the sets were also LIDAR’d to aid in
reconstruction). “We wrote a little pipeline in Nuke that allowed us to
stitch all the photos together and then very simply calibrate them with
the Smallville geometry,” says Rocheron. “We would calibrate just one
angle, because for the full dome, all the photos would get automatically
calibrated on the geometry. For us it was a very good process – it
wasn’t just a sphere. Everything was re-projected in two and a half D on
the geometry to get parallax and the camera would travel technically in
all directions.”
Superman is also seen in several sequences, of course, flying though
clouds. “We used volumetric clouds,” says Rocheron, “using an internal
tool we have for clouds to mobilize geometry and transform it into
volumes and refine with layers of advection and noises for the fine
details.” In terms of environments Superman flies through, such as over
the Arctic circle, over canyons in Utah, over Africa, and over the Dover
cliffs, MPC developed these first in Terragen and then took them
through to matte paintings and geometry.
Smallville features in this Man of Steel trailer.
5. Full-screen
digi-doubles were of course a major
component. MPC led the digi-double Superman, Zod, Faora and other
Kryptonian creations which were shared with other vendors. Digital
armour was also added along with the energy-based Kryptonian helmets.
Cyberscan and FACS sessions were conducted with the actors, and
polarized and non-polarized reference photos were taken. Superman’s cape
and costume were scanned in high detail – the cape in particular became
a direct extensions of Superman’s actions. “Our main reference for the
cape were illustrations from Alex Ross,” states Rocheron. “We had the
cape here at MPC so we could really study its thickness and the
velvetness. The light is very soft on it. We did a cloth solve in nCloth
and we wrote a number of tools in animation to be able to animate the
cape and see it in real time. Once the animation was approved, we had a
basic representation of the cape and we would then use that to drive the
nCloth simulation.”
Superman’s cape
“Zack was very amenable to shooting a lot of stuff with Henry Cavill
without using a real cape at all,” says Desjardin, noting that as the
cape was envisaged almost as its own character, it would need
significant visual effects art direction.
But the cape’s VFX also had to remain within the illusion of the
filmmaking. “There were a couple of shots where someone might say, ‘I
don’t quite like the way the cape moves there because it looks like it’s
a real cape with a wire attached to the end of it to pull it,’ and Zack
would say, ‘That’s fine! Don’t change that animation – I want people to
think that maybe we did that, even though it’s a CG cape.’”
MPC used the latest version of RenderMan and its raytracing
capabilities to help with the chainmaille look of Superman’s suit and
the Kryptonian suits and armour. “They were all painted as displacements
but we did hi-res displacement,” says Rocheron. “Raytracing allows you
to capture that very subtle detail between the reflective pattern of the
chainmaille and the light absorption of the blue part of the suit,
since the underlying layer is bleeding through. We have infinite area
lights which are the dome and the finite direct area lights that are the
direct light sources you want to position in space. And that physically
based setup gave us a terrific look for the reflections and the
fall-off of the light – really key to get the details of the suit and
armour, which in reality was mostly black.”
In one shot, Superman fist fights with an eight foot Kryptonian. “We
shot a live action piece and just replaced the performance capture stunt
guy and added the cape onto Superman,” says Rocheron. Then we just
thought about it and said it would be much cooler – since the Kryptonian
is very tall then Superman should fight against him while he’s
hovering. We did those shots as entirely digital shots. It has that very
cool feeling of flying around and punching him from all different
directions.”
6. For each shot, it then became a matter of choosing the right
transition point.
“There’s a little transition zone that’s maybe only one or two frames,”
says Desjardin. “We knew that we wanted to keep Superman real in
certain places because it was say super-sharp and we want to use that to
anchor the shot, even if just for a couple of frames, and then we’re
going to go into digital because it’s crazy right after this.”
MPC also worked on shots of Superman ‘learning’ to fly.
“We layered a couple of other things on top of that,” adds Desjardin.
“One, if there’s a punch being thrown, you can lose the arm real fast
if it’s too fast. So a lot of times the arm of a CG character may be
going just slightly faster than a human’s – we put a sonic boom-type
signature around the forearm and we might put a little heat luminance on
the leading edge surfaces of the fist. It puts an idea in your brain
that it’s moving really, really fast even if it isn’t.”
7. Not only were fights being depicted with digi-doubles and
environments, they also traversed cornfields, through buildings, glass
walls, on roads and against flying A-10s. That necessitated incredible
destruction
and for this MPC looked to its finite element analysis tool Kali, which
had first been developed for the wooden pagoda destruction in Synder’s
Sucker Punch.
This time around, following a few years of development, Kali was able
to handle so many more kinds of materials. “So we could take a tarmac
and break it differently,” explains Rocheron. “It’s more resistant so it
has a crater but cracks at certain places near the surface. And
Superman crashes into a bank vault and crashes through a glass door with
a metal frame and finally into the vault which is made of super strong
steel so we made that bend and wraps around him.” Particle sims and
Flowline were also incorporated into the destruction pipeline for
Smallville.
Destroying a city: the invasion of Metropolis
Superman takes on Zod at Metropolis.
Determined to conquer Earth and transform it into a new Krypton with
‘world engine’ machines, Zod launches northern and southern hemisphere
strikes in Metropolis and the Indian Ocean, respectively. The result,
until Superman battles and then defeats the Indian Ocean world engine,
is that significant parts of Metropolis and its soaring skyscrapers are
destroyed – a task given to Double Negative. The studio also realized
further full-scale destruction as Superman and Zod wreak havoc on
remaining buildings and each other.
“Down in Metropolis there was a very clear design edict that came
from Zack about how the evolution of the battle was going to be,” says
Desjardin of the lighting design for the film’s third Act. “The sun had
to be not quite setting when the Black Zero comes down and then very
quickly it’s in its setting position and by the time Superman and Zod go
to fight it’s down below the horizon and there’s a Hawaiian cloud and
colorful clouds and it’s getting dark with a twilight sky – more an
ambient look. Then once Supe and Zod jump up into that sky then you have
some other lighting options with the sky and certain lit billboards.
It’s a way to make the city come alive to make it even more dramatic to
keep characters backlit.”
Supe and Zod battle it out.
To create a convincing Metropolis, Dneg looked to Esri’s CityEngine
to help procedurally deliver the city, a tool it had first employed for
the sprawling future world of
Total Recall. “That was a much
more sci-fi based role,” notes Dneg visual effects supervisor Ged
Wright, “so we took what they had done and extended it a great deal. The
work we were doing was based around the Downtowns for New York, LA and
Chicago and that gave us the building volumes for heights. We’d skin
those volumes with kit parts but most of it then had to fall down! So we
had to rig it for destruction and use it for other aspects of the work
as well.”
The previs effort
Director Zack Snyder worked closely with previs and postvis artists
at Pixel Liberation Front to orchestrate some of Man of Steel’s most
dynamic scenes.
The studio worked on more than 15 sequences, with PLF supervisor Kyle
Robinson suggesting that some of the most challenging previs creations
were the oil rig rescue and the Metropolis invasion. “They shot that oil
rig rescue in a parking lot in Vancouver and they didn’t know how big
the greenscreen needed to be, so we modeled it for them to the specify
what was necessary.”
Superman’s cape was also a major challenge. “We didn’t run a cloth
sim for the cape,” says Robinson. “If we had run a cloth sim it would
only appear to react to the forces applied to it and not be a character
in its own right. We rigged the cape to have a deformation in it and it
had a bone in it so we could move and pose it, for the right position
for comic book poses.”
Using Maya for main animation and its proprietary character and
camera rig, with extra work done in After Effects, PLF followed Synder’s
boards and art department concepts to flesh out scenes, and later help
editorial with postvis.
For building destruction, in particular, the studio re-wrote its own
asset system to be geared towards dynamic events. An implementation of
the Bullet engine inside Houdini – dubbed Bang – became Dneg’s main
destruction solver, with a core philosophy of allowing for quick
iterations with heavy control. “We wanted to be able to run an RBD event
and trigger all these secondary events, whether it was glass or dust
simulations – all of those things needed to be chained up and handled in
a procedural way,” says Wright. “One of the advantages of this was
that, because it was all based around a limited number of input
components, you can make sure they’re modeled in a way they’re useable
in effects – you can model something but they’ll be another stage to rig
it for destruction.”
In addition, fire, smoke and water simulation tools were further
developed at Dneg. The studio moved from its existing proprietary volume
renderer DNB to working in Houdini and rendering in Mantra for elements
such as fireball sims. Dneg’s in-house fluids tool Squirt also
benefited from new development to handle larger scale sims and
interaction for more tightly coupled volumes and particles. Overall, the
studio’s rendering pipeline has moved to a more physically based
approach in RenderMan.
Within the Metropolis sequence there were other numerous requirements
including attacking and destroying aircraft and, of course, digital
representations of Superman and Zod when they fight. One particular
element Dneg contributed and also shared with other facilities was Zod’s
armour. “There was no practical armour for Zod,” states Wright, “he
only wore a mocap suit. We took concept art, came up with a ZBrush
sculpt of the armour and could show them turntables of what it would
look like during filming.”
Dneg took MPC’s Superman and Zod models and adjusted them for their
own pipeline in order to rig, groom hair and adjust shaders. “We also
have more of a photogrammetry approach to facial,” says Wright, “so we
made the actors sit there again with an eight camera rig – similar to
Light Stage but portable and gives us polarized photography to
reconstruct the facial expressions.”
Zod’s powers come to the fore in his battle with Superman.
Zod and Superman battle in amongst buildings and when they hit each
other tend to generate enormous shockwaves that rip skyscrapers in half.
Although much of this was completely digital (some live action was shot
in Chicago and then on Vancouver greenscreen soundstages), Wright says
Dneg implemented real photography onto its digital doubles wherever
possible. “Because you have their performances you engage with it – and
your eyes go straight to their faces. If they’re big enough in frame and
doing something, you want to use a photograph of them. As soon as you
buy that and get what’s going on, you’re more willing to take on board
what’s going on with the rest of the frame.”
Adds Wright: “There’s one shot where Zod hits Superman up the side of
the building. Superman is hovering above. Zod starts running up the
side of the building. This is just before he rips his armour off and is
taking in more of the sun’s energy. Superman flies down to hit him and
the two of them collide causing that shockwave. DJ and Zack were both
really keen to make it feel like two Gods were fighting, and they were
at the height of their powers right then.”
Rounding out Man of Steel’s effects
Helping to round out the effects work on the film were companies like
Scanline and Look Effects. Here’s how they added crucial shots to the
film.
Scanline – tornado and oil rig
Scanline delivered shots of the tornado sequence in which Smallville
residents shelter in an underpass from an approaching twister, while
Clark Kent’s father Jonathan returns to a vehicle to rescue a pet dog.
“For the tornado itself,” explains Scanline visual effects supervisor
Chad Wiebe, “we actually came up with a unique methodology by combining a
number of individual fluid sims which would be wrapped around the
funnel. This allowed us to create a bigger and denser funnel without
some of the overhead that would have been generated by trying to do a
single sim for the entire funnel. This also allowed us to pick and
choose from a library of different sims which gave us greater control
over the look and speed of the funnel, the variation of different parts
of the funnel, as well as the technical aspects such as density and
resolution for some of the more close up shots.”
See part of the tornado work in this TV spot.
Along with the tornado, artists added ground dust and debris, farm
buildings and uprooted trees. “We also had to create digital versions of
all vehicles that were shot on set as well as a number of additional
vehicles to suggest a longer line up of traffic stopped on the freeway,”
says Wiebe. “As the sequence progresses most of the vehicles end up
getting damaged or destroyed to some degree so in addition to a typical
vehicle rig for the basic motions and wind buffeting, we also created a
system where we could dynamically damage the vehicles based on
collisions with one another or based on forces as was the case with the
Kent truck which gets destroyed at the end of the sequence.”
An earlier sequene of Clark saving workers from a burning oil rig
made use of reference of the BP Deepwater Horizon explosion and the
Toronto Sunrise Propane factory explosion. “We tried to make sure we
were as accurate as possible regarding the look of the fire and smoke
plumes that are generated by oil fires, which have a very unique and
identifiable quality,” notes Wiebe. “The exterior plates where shot with
the actors on a set built helipad, with a real helicopter, and green
screen surrounding 3/4 of the set. From there we created and entirely
digital oil rig, and we would composite the actors and helipad onto our
digital rig, or at times we would replace the helipad as well. Many of
the hero helicopter shots also utilized a digital version of the
helicopter in order to get the interactive lighting and reflections
matching.”
“The oil rig collapse was a series of rigid body simulations created
using Thinking Particles,” says Wiebe. “From there we would also add
fire, smoke and dust trailing off the rig using Flowline, which was also
used for the fluid sim when the rig came crashing down into the ocean
below. There were also a series of explosions happening throughout the
sequence also using Thinking Particles for the RBD’s and Flowline for
the fire and smoke.”
Look Effects – the bus crash
In
the film, Clark remembers key moments from his youth, including those
they gave hints to himself – and others – of his tremendous powers. One
is the crash of a schoolchildren-filled school bus. After it blows a
tire and launches off a bridge into a river, Clark dives out of the bus
and pushes it to the bank, and then rescues another child from under the
water. Much of the crash was filmed practically on a bridge and quarry
location, and then on a tank stage. Look Effects helped piece the scene
together.
Some of the work included rig and camera removal and also clean-up of
the bridge railing. “There was a POV camera angle from the bridge
looking down at the bus as it was sinking into the water,” notes Look
visual effects supervisor Max Ivins. “The bridge part was shot
separately from the sinking shot and the exterior sinking was shot in a
rock quarry so there was no moving water like the river. We did some CG
replacements of the vents on top of the bus and we had to make the
remnants of the splash when someone runs up to look at the bus over the
edge. We added the foam ring and the bubbles coming up. Used stock
footage and CG elements to make a post-splash surface of the water.”
For interior shots of the bus with the children, Look altered water
levels to make the danger appear more prominent. “They had a surface
outside of the bus that was basically the same as the inside of the bus –
they couldn’t really sink it because there are kids involved,” explains
Ivins. “So they had to make it look like the outside water was 2 feet
taller than the inside water that is rushing in to give it that sinking
feeling. So we did whole simulations and cleaned up some of the
lighting. We made bubbles coming up and making it turbulent.”
Look’s other contributions to the film included several monitor
comps, including ones at NORAD, and some artefact clean up for a
flashback signature shot that had been time-ramped of Clark wearing a
cape and with his hands on his hips in front of some blowing dandelion
heads.
All images and clips copyright 2013 Warner Bros. Pictures.
