Chip 1996 October/CHIP 1996 oktober (CD08).zip/CHIP_CD08.BIN/cikkek.hez/t2-3d/t2-d3.txt

Text File | 1996-09-03 | 28KB | 427 lines

GROUNDBREAKING TECHNOLOGY POWERS
"TERMINATOR 2-3D"


DATELINE - A groundbreaking technological feat, "Terminator 2 -3D" pushes
the filmmaking envelope on many levels. On the production end, the desire
for state-of-the-art 3-D imagery necessitated the retooling of a technology
that remained largely unchanged since its popularization in the '50s though
it was significantly updated in the '80s.
    To optimize presentation impact in the film's custom-built venue at
Universal Studios Florida, it was decided that the look should be
larger-than-life. The resulting proscenium stretches across three
24x50-foot screens, for a total image span of 150-feet, making it the
widest 3-D installation ever. The very vastness of his palette prompted
director James Cameron to pay close attention to image quality, a concern
that led him to select the 65mm production format.
    The oversized 65mm format utilizes a negative that is more than
twice as big as a standard 35mm negative, doubling the sharpness and
clarity of the imagery. Unfortunately, the equipment is also roughly double
the size, posing it's own sets of challenges, particularly in the
sophisticated twin camera 3-D system used on "T2-3D." The fact that Cameron
demanded his cameras be completely mobile raised the bar even further.
Every creative principal involved with the project - many of them veterans
of numerous 3-D endeavors - concurs that "Terminator 2-3D" incorporates the
best 3-D footage ever shot.
    The project is also the first special venue attraction to make such
extensive use of computer-generated imagery and to blow it up to this large
a format. The digital solution was embraced as a means of realizing
fantastic sets, characters and images. To achieve visuals convincing enough
to hold up under such enlargement, digital studio Digital Domain dove deep
into its bag of tricks, utilizing custom computer software programs and
expanding  the limits of popular off-the-shelf packages by working with
Beta test versions that the companies were often willing to tailor to their
special needs. The advanced digital imaging presented a challenge not only
in its execution, but its integration with state-of-the-art model-building,
photography and stage work a feat  in form and function as well.

THE FUTURE IN THREE ACTS

    The 12-minute production was broken down into three acts, all of
them 3-D, and all of them digitally processed. During the 2.5 minute Act 1,
the audience is welcomed in from the pre-show and seated in the Cyberdyne
auditorium, where they are ostensibly assembled for a demonstration of the
 latest and greatest technologies. Things go awry, however, with the
unexpected visit of Sarah Connor and her son John. Members of the Human
Resistance, they are determined to save mankind from a future controlled by
mechanical despots. Inside the Cyberdyne auditorium, they are pursued by
two denizens of the future: the heroic T-800 Terminator, essayed by Arnold
Schwarzenegger and the evil T-1000 evoked by actor Robert Patrick. The act
is punctuated by stunning screen-to-stage transitions - one of which
involves a motorcycle flying out of the screen from a trap door - and some
edge-of-your seat computer-generated 3-D.
    Absconding from the theater, the T-800 and John Connor time travel
to the year 2029, where they are pursued by the T-1000 and the malicious
Terminators. A chase begins through the fantastic post-apocalyptic Los
Angeles landscape,  through which the duo search for the headquarters of
Cyberdyne's Skynet. Launched around the turn of the century, Skynet was
designed as a Star Wars-type national defense system, but things went
terribly awry when this "thinking machine" decided it didn't need man, and
turned its military powers on humankind. Skynet's powerful reach extends
around the globe through a network of linked satellites that are controlled
from a central core," the search for which comprises the 7-minute Act 2.
    Act 3 sees John and the T-800 infiltrate Skynet, where they attempt
to destroy the mainframe computer that is the evil empire's central nervous
system. A surprise awaits them in the form of the T-1 million, guardian of
the CPU. The
resulting confrontation sparks 1.5 minutes of awe inspiring 3-D CGI imagery.
    Act 1 was achieved through a combination of live-action and CGI,
while a more straightforward cinematic approach was taken for Act 2, which
is almost exclusively live action, shot on location in Desert Center
California and on a Hollywood soundstage. Aside from the live-action of the
actors, Act 3 - including the fantastic new T-1 million, or T-Meg, and the
backgrounds - was created entirely in the digital domain. Acts 1&3
incorporate live stage action while Act 2 does not. The filmed portions of
Acts 1&2 utilize only the center screen, while Act 3 opens up to include
all three screens, a move designed to totally envelope the audience in the
experience.
    "When we reveal those three screens it's just impossible not to
feel  you're a part of that environment," says Nanci Herbst, "T2-3D"
producer for MCA. Positioned at 60-degree angles, the 150-foot span of the
three 50-foot by 24-foot screens is so encompassing that audiences will be
immersed in it no matter which way they turn. The 3-D takes it a step
further. "It allows you to actually be there. You're in the experience.
It's not a passive thing you're viewing =F3 you get to be a part of it,"
Herbst explains.



NEW ERA OF DIGITAL IMAGING

    Ultimately, computers were both the challenge and the solution on
"Terminator 2-3D." To create the type of striking images envisioned by
Cameron, the computer department at Digital Domain was pushed to its very
limits. "It forced us to mature our CG animation department very, very
rapidly, because we had to," says Cameron. "The mimetic poly-alloy we used
to create first the T-1000 and now the T-Meg is so well understood it's
been duplicated by everybody since T2, but it's never been done at this
resolution in three dimensions, so we had a lot of 3-D science to work
out."
    The computer work was broken down into two parts: computer graphics
imaging (or CGI) and compositing, the process by which separate elements
are sandwiched together in a single image. A total of 47 CGI artists and 8
compositors worked full-time on the project for more than six months
reports Digital Domain digital effects producer Amy Jupiter. That's as much
manpower as would typically be devoted to an effects-heavy feature film.
    "T2-3D" is the first major production of its type to be 100%
digitally processed. That means that every frame of film - including the
live action - was manipulated in the computer in some way. Live action
footage is transferred into the ones and zeroes of digital language by a
process known as "scanning," whereby the images are displayed and
photographed, frame-by-frame, by a digital camera. The material can then be
color corrected, composited with other images, or augmented with the
addition of CGI. (Since CGI is of digital origin, no scanning is
necessary.) A basic CGI tweak could consist of lighting elements -
reflections or light rays - while a complex CGI undertaking could involve
adding an artificial environment, or a new character. All of the
aforementioned was done to present "T2-3D."
    The digital material is then output back to film (for projection)
via a process known as "recording," by which the digital image is displayed
on a cathode ray tube (also known as a CRT)  and again, photographed
frame-by-frame, this time by a film camera. In terms of its complete
digital processing, "T2-3D" heralds the future. Cameron, who is considered
a visionary in computer imaging, feels eventually all films will be
digitally processed from start to finish, facilitating greater ease of
editing, effects and general negative fine-tuning. Though the current cost
of digitizing a two-hour feature film is considered prohibitive, because
"T2-3D's" 11-minute running time, coupled with the fact that it was
produced by a digital production facility, made it the perfect proving
ground for this technique.
    Largely reliant on its live stage elements, Act 1 of the film
incorporates some digital trompe l'oeil. The back wall of the theater is
actually a film element of a CGI image that includes the Cyberdyne logo,
which melts and morphs into the reflective head of the T-1000 then
transitions to the face of actor Robert Patrick. Telescoping into the
theater for the first jolt of 3-D, the T-1000 snaps back into the screen,
at which point a T-1000 actor is catapulted from a trap door in the stage,
appearing in a puff of smoke. Both the melting Cyberdyne logo and the
T-1000 head were modeled and animated using a beta test version of Alias
7.0 software. The CGI was rendered in Renderman using a proprietary Digital
Domain conversion program.
    Another digital image in Act 1, though one not likely to be
recognized as such, is the onscreen Harley motorcycle and its rider, the
T-800. "This is an example of something that would have traditionally been
done with a model, but we chose to go with CGI, because of all the things
we needed to do to the image," says Jupiter. "We needed to get the weight,
the size, the lighting perfect. If it had been a miniature, there would be
nothing much we could do to it once we'd shot it. Because it was CG, we
could endlessly tweak and tweak the hand-off image, making it match
seamlessly with the live action. The computer really adds enormous
flexibility and helps enhance the illusion that what you're seeing is
real." The effect was also achieved through a combination of Alias and
Renderman.

MOVIE WITHIN AN ATTRACTION

    Act 2 is primarily live action, and features puppet robotics and
animatronics by legendary character creator and effects maestro Stan
Winston, one of three founding partners in Digital Domain. For "T2-3D,"
Winston recreated the unforgettable Terminator endoskeleton he brought to
life in the two Terminator films, updating the technology slightly. "He's a
bit improved over the version in 'T2.' In this case we used real chrome
instead of artificial chrome. It still looks like it did in the original
film, but it has a little more life, a little more reality." Winston - one
of two effects wizards to receive co-director credit along with Cameron,
the other being John Bruno - helmed the Act 2 shots that featured his
life-sized, chrome endoskeleton. "Jim shot the stuff with Arnold and Eddie,
and I shot most of the endoskeleton shots leading up to that. Usually the
footage I directed was of 'my' actor,  the endoskeleton. I did that and I
did some shots of the Saucehead."
    "Saucehead" was the friendly term for a modified version of actor
Robert Patrick as the T-1000. "It was a mechanical effect: we rig it and
create a little animatronic version of Robert that blows open and creates
the appearance of being the liquid metal head." Fans of Terminator 2 will
thrill to the return of Saucehead, noting that this time there's a twist:
his head liquefies beneath goggles and a helmet. "You always want to do
something subtly different so the audience doesn't say, 'Oh, I saw that
before.' You want to let them know they're seeing a new movie," says
Winston, who says the most exciting aspect of reprising some of his most
memorable characters was having them "live in a 3D world." As far as
challenges go, Winston notes that for him it was in the actual shooting.
"The breakthroughs here were in how we used 3D technique and worked with it
in the digital world, at Digital Domain."
    Act 2 also includes Digital Domain-produced CGI elements in the
form of the Mini Hunter-Killers, or Mini H-Ks - unmanned flying probes
approximately one-foot in diameter deployed to search out the heroic duo
once they have taken shelter in a destroyed parking structure where they
can no longer be followed by the helicopter-size Flying H-Ks. The Mini H-Ks
were created as models that were digitally scanned into the computer, where
they were enhanced: given texture and shading using the Softimage modeling
package and Renderman.    In one shot where Schwarzenegger reaches out and
actually grabs a Mini H-K, the model was used in filming to facilitate more
believable interaction between actor and gadget. However, digital science
was even able to improve on that scene - it was sped up in the computer.
The slower versions looked like Schwarzenegger was shaking the saucer,
while the faster version had the intended effect of it shaking him.
    Model Mini H-Ks were also shot (sans actors) on a miniature set of
the destroyed parking structure, constructed by Alterian Studios and
photographed by 4-Ward Productions' Robert and Dennis Skotak, frequent
Cameron collaborators who also worked on "Terminator 2: Judgment Day,
"Aliens" and "The Abyss."  The miniature photography was accomplished using
a 35-mm camera and crane set-up, allowing for much more fluid tracking and
greater freedom of motion than the 65mm camera rig allowed, explains
director of effects photography Russell Lyster. The miniature shots will be
used as "inserts," edited in with the 65-mm live action work lensed on
location at Kaiser Eagle Mountain and on a Hollywood soundstage by live
action director of photography Russell Carpenter. The exterior of Skynet
was also an elaborate miniature, 24-feet wide at its base and proportioned
to appear to 800-feet tall in the finished film.
    Digital magicians also lent their skills to everything from
repositioning and reshaping explosions to extending doors and other set
pieces so they'd look larger than life, according to compositing supervisor
Mark Forker. The compositing itself could involve anything from layering
three or four elements in scenes involving the insertion of the CGI Mini
H-Ks to layers 40-thick in creating the detailed computer-generated Skynet
interior for Act 3. Flame was the primary compositing tool in Act 2 while
Acts 1 and 3 were composited using a proprietary Digital Domain program
dubbed Nuke.

CREATURE FEATURED

    The transition to Act 3 is facilitated by an elevator descent that
reintroduces the live stage actors. As the elevator reaches its
subterranean destination, the presentation opens up to reveal all three
screens, featuring a background conjured entirely in silicon. The
mechanized Skynet environment of gleaming steel features on its center
screen the Skynet "Central Core," with an immense chrome pyramid in the
foreground of a cold corridor that seems to go on forever. Catwalks and
conveyor belts criss-cross the scene at weird angles. Liquid nitrogen moves
through translucent walls and streams down from vents and piping that line
the back walls an ceiling - an effect heightened by the real liquid
nitrogen that pours off the walls in the auditorium. On the left and right
screens are branching tunnels. To the left is a gigantic video wall from
which Skynet monitors its empire. (It also allows the filmmakers to magnify
the stage action, using close-up footage of actors Schwarzenegger and
=46urlong).
    Most of the Skynet modeling was handled by an outside vendor, 3
Name 3D, which used Side Effects' Prisms software. "We wanted to be able to
concentrate on the lighting and the textures," says digital effects
supervisor Judith Crow. Indeed, getting the lighting just right was the
most challenging aspect of creating Skynet. "As far as the background goes,
nothing much moves. There are some mechanical arms that swing, and there's
an elevator unit that matches up with a real elevator on the stage, but
most of the animation is in the lighting effects, which produce changes to
the look of the environment," Crow explains. For its part, Digital Domain
used Prisms, as well as some Alias applications, to set up the lighting and
shading, as well as the odd modeling job. The end result is a world as vast
as it is detailed. "We basically built everything to human scale," Crow
says.
    It is in the sleek, futuristic environment of Skynet that the
ultimate screen monster makes its  appearance. The most awesome feat the
byte brigadiers pulled off was creating the deadly T-1 million. A
six-legged poly-alloy arachnid, it morphs into being from melting chrome,
springing to spiderlike life. Standing 30-feet high, with razor-sharp legs
that clang metallically whenever a bladed foot hits the ground, the
creature scrambles from screen to screen, extending its lethal limbs into
the audience through the 3-D pursuit that marks the attraction's combative
climax.
    "To build it and animate it with six legs was enormously
complicated," says Jupiter. The stakes were raised even further by the fact
that the creature goes through several cycles: morphing to life, scrambling
across the three screens, freezing into a rigid form that shatters into
tiny pieces - only to congeal and reform. To implement all the phases of
the creature a variety of software packages were utilized. The T-Meg was
modeled using Softimage. "We built the T-Meg as a NURB file," says digital
effects supervisor Neville Spiteri, explaining that NURB stands for
Non-Uniform Rational B-Spline. The cutting edge in creating organic
creatures, "It's a type of surfacing that allows a lot of deformation. We
needed that because the creature stretches and has a jaw that does all
kinds of crazy things."
    The creature was imbued with digital life courtesy of Softimage's
Inverse Kinematics animation program. Lighting, color and texture was added
using Softimage's Mental Ray, which offered the ability to produce
"self-reflections." "Because of the shape of the T-Meg, with all those legs
moving against each other, self-reflection is critical," Spiteri notes.
Chosen for its "particle animation" properties - which control things like
velocity, mass and weight -  Prisms triggered the explosive shatter effect.
"It's a very intense sequence, because we generated this entire explosion
inside the computer," the digijock marvels. Targeting a special effect for
the six spidery legs, which don't explode so much as "collapse against each
other," the digital crew used the Alias Dynamics animation package. A
custom developed in-house program called Shatter SOP (Shatter Surface
Operation), was created by Digital Domain specifically to "break" the image
into interestingly shaped pieces and also to interface with Prisms (which
animated the broken up shards). The reformation was accomplished using
Softimage Meta-Clay.
    In order to integrate all three software packages -Alias, Prisms
and Softimage - Digital Domain crafted yet another home-grown software
program that transcribed the different code into a sort of universal DD
tongue that could then be ported over to Renderman for speedy rendering.

A NEW DIMENSION OF THRILLS

    The digital feat was further heightened by the fact that everything
had to be processed for 3-D. "Three-D compositing is particularly tricky,
because in 2-D you can get away with murder, compared to 3-D where things
really have to lay in at the correct plane of depth," Cameron says, noting
that when marrying separate elements together, one must be conscious of not
only the X axis (horizontal) and Y axis (vertical), but of the Z axis, or
depth, as well. "If you're putting an object in that wasn't really there,
it not only has to go in at the right place in the frame, but it has to go
in at the right depth as well," Cameron continues. " Like for example, when
we're adding in the big Skynet bunker half-a-mile away across the
landscape, it has to go in at the right distance across the landscape or
it'll appear to be floating in front of things that are really floating in
front of it."
    Another even more complex example in Act 3 sees the T-1 million
shatter into a hail of shards, and the array of their trajectory - through
wafting clouds of debris - was critical. To simply lay one element (the
shards) over the other (the clouds) would render flat coverage that
wouldn't work in the 3-D world. To layer the images in three-dimensional
space Digital Domain utilized an in-house software program called
Z-Composite. "The clouds were soft and the shards were hard, and soft and
hard is really hard to composite in a computer," says Jupiter.
    As far as composites go, the T-1 million alone was made up of four
different elements, while the Skynet environment into which it was inserted
was itself made up of as many as 40 elements. Live action footage of Arnold
Schwarzenegger and Edward Furlong were the only non-digital elements
composited into the scene.
    Overall, the imaging chores were multiplied many times over by the
fact that the images were produced in stereo - which requires separate
images for the left and right eye -  as well as, in the case of Act 3, for
exhibition on three screens. Further complicating matters was the fact that
every frame contained at least one composite =F3 most of them much more than
that. This painstaking assembly was required, to a greater or lesser
degree, on frame of film. Once a satisfactory composite had been achieved,
the film was run out in duplicate, to simulate stereo. Outputting of a
final composite could take anywhere from one-hour-and-20-minutes per-frame
for complex images requiring motion blur to a several minutes per-frame for
the simplest light beam.
    To present Act 3, three sets of stereo imagery, or six separate
images, were required. Each pair of stereo elements had to match precisely.
All stereo elements were output simultaneously on dual film recorders
steady-tested for precise calibration and synched via computer to SMPTE
timecode. "Even though the actual presentation is only 11 minutes, we'll be
outputting a total of 32 minutes of linear film," says effects producer
Jupiter. The digital information required to present the filmed portion of
T2/3D occupied 800 gigabytes of disc storage.

NEW 3-D STANDARD

    While the 3-D images for the digital scenes were created by
outputting twin strips of film from the computer, the live action portions
had to be captured in 3-D by using a special 65mm camera system. Key
members of the T2/3D crew were 3-D veterans. Three-D director of
photography Peter Anderson had worked on numerous early 3-D theme park
attractions ("Captain EO," "Muppetvision"), while show producer Chuck
Comisky and technical consultant Dr. Ken Jones had previously teamed
Universal Pictures' "Jaws 3-D" and later on the NASA film "Mars in 3D."
    For the live action portions of T2-3D, the production had a total
of three 3-D rigs at its disposal, "which was unprecedented," says director
of photography Russell Carpenter. The technical team put together a stereo
camera package that utilized the best existing technology, pushing it even
further. The heart of the main system was a rig originally designed in
1984, which was then "beefed up considerably," according to Anderson.
Created by engineer Steve Hines, the rig is also known as an "active 3-D
system" (since it allows the adjustment of 3-D convergence and "interocular
on the fly"). Among its advantages is the fact that it is universal and
works with numerous film and electronic camera systems. The T2-3D live
action crew relied primarily on Showscan 65mm cameras. Complete with
cameras, film, lighting and other accessories, the rig weighed roughly 450
pounds and was the size of a washing machine.
    In addition,  two other 65mm 3-D rigs were deployed. One of them,
custom-designed and manufactured by Digital Domain, was used heavily.
Additionally, the Paradise rig was used when three operating systems were
needed, particularly for coverage of spectacular pyrotechnic scenes. All
three were twin camera with beam-splitter systems.
    These systems utilize two cameras to capture stereo images. Light
coming in through the lenses hits a beamsplitter mirror that sends 50% of
the visual information to a camera filming the right eye's image, the other
half to a camera filming the left eye's image. In general, 3-D is simulated
by placing the cameras so that the spacing between the two lenses
approximates human vision (the average human interocular =F3 the distance
between two eyes  =F3 is two-and-a-half inches). The beam-splitter is
necessary because the large 65mm cameras and lenses are too bulky to be
positioned that close together.
    On T2-3D, this problem was compounded by the super-large screen
format. "Because of the incredibly large screens and the image
magnification produced to fill these screens, the proper 3-D computations
involved placing the lenses even closer together to enhance the 3-D
experience without ripping the viewers eyes out," Anderson says. The
result, however, is that T2-3D audiences will be treated to some of the
most spectacular spatial imaging ever presented, since the larger format
results in a 3-D presentation with more depth.
    The Hines rig's advantage of interocular and convergence that could
be "adjusted on the fly" allows precise control over the placement of the
3-D images. The Digital Domain system also had this feature, which was
computer-controlled. "Picking up on the style of T2, this was such a
kinetic show. Everything was always moving - the actors were always moving,
the cameras were always moving. Intense motion. The ability to adjust and
fly the interocular and the convergence is what made this show work."
    In addition to the trio of 65mm 3-D rigs, there was a Stereovision
lens system on an Arriflex 35mm camera that was used for Steadicam and
other specialty 3-D photography. For the miniature photography a
lightweight 35mm rig and Digital Domain's motion control systems were used.
Programmed with custom written programs, the motion control systems were
able to create 3-D by shooting separate passes for the left and right eye.
Primary films used were Kodak 5298 for production, and 5294 for models and
miniatures.
    Live action imagery was scanned into the digital realm at 10-bit
logs, registering 1.6 million colors per picture. They were output at 2,048
x 896 pixels of resolution. "The CG is resolution independent and can be
rendered at any resolution. It's the original live footage that you have to
be careful of," says digital effects supervisor Karen Goulekas.
    T2-3D is presented in-theater using polarized 3D, a process that
utilizes  two 70mm film projectors. In front of each projector are
polarized filters, offset from each other at 90-degree angles. The same
type of polarization is in the glasses worn by each guest in the theater.
By putting the polarization and on the audience's glasses, the left and
right images are kept discrete, allowing them to see the projected images
in 3-D. The projector screen is aluminized with a metallic coating on it to
hold the polarization. The film was shot and will be projected at 30
frames-per-second instead of the usual 24 fps feature film presentation
format. As a result, sharpness and image quality is dramatically improved,
particularly in fast action sequences, due to the rapid image refresh rate.

    "It wasn't  until we started doing the 3-D testing at Digital
Domain that I began to realize just how fantastic this thing could really
be," says Cameron, adding, "It's a completely different kind of experience
than a feature film. It's a heightened experience =F3 shorter but more
intense." The director lauds MCA for constantly raising the bar creatively.
"They want showmanship. They consistently pushed the Digital Domain team,
and the Landmark design team to go further, to create better showmanship,
to make it more fun. The collaboration pushed it beyond what any of us had
initially thought possible."