a copy of the article that had appeared in Volume 6, Issue 6 of the CGI
Magazine, June 2001
MPC TIES UP WITH
BY CHAS JARRETT
Carlton Internationl’s acquisition of Gerry Anderson’s classic
1960’s and 1970’s TV shows is leading to a pletbora of re-releases.
But re-mastering picture and sound, although popular, isn’t the full
extent of their future poptential. Chas
Jarrett recalls the recent work he and others at the Moving Picture Company
carried out ot bring Captain Scarlet into the 21st Century.
SAID THAT FOR EVERYTHING HIS SHOWS WERE, AND EVERYTHING THEY’VE NOW BECOME, HE
WAS NEVER SATISFIED WITH THE CONSTRAINTS OF THE TECHNOLOGY OF THAT TIME.
THE SHOWS WEREN’T CONCEIVED WITH PUPPET ACTORS OR MODEL SETS IN MIND.
first met Gerry Anderson in December 1999 at MPC’s offices in Soho, London.
He was clutching a script and a folder of drawing and he laid them out
with enormous enthusiasm as he explained the job we were all about to begin.
It was a five-minute long, all-CGI pilot episode of his 1960’s series Captain
Scarlet, and we had only nine weeks to make it. With storyboards by Robin Shaw and conceptual designs by
Steve Begg, we broke the script down into its various components.
first thing that struck me was the amount of character work involved.
Almost 80 per cent of the shots contained one or more of the three
featured characters, and in keeping with Anderson’s other work this script was
heavy on dialogue to progress the story. This
meant lots of lip-sync and facial animation as well as running, jumping,
fighting and other interaction. Secondly,
there were three main environments which were either very large or very detailed
or both. Two were exteriors (one
daylight and one night) and the other a daytime interior.
There were also cars, aeroplanes, bridges, bombs, shattering trees,
lightning, explosions, rocket thrusters, et cetera. It all amounted to a busy nine weeks and the clock was
January 2000, with a limited crew and pounding New Year’s hangovers, we began
modeling. Anderson was adamant that the graphics retain the look and feel of the
original series, so although updated, the concept art was distinctly retro.
Particularly, Anderson wanted the characters’ faces to look like the
originals. He commissioned maquette heads of Captain Scarlet and Captain
blue (the good-guys) and Captain Black (the bad guy) which we sent to Viewpoint
in America to be cyber-scanned. The
result was three perfect head models with exact likeness of the priginal
puppets. We split the crew into
teams, one handling the character modeling, and animation set-up, another took
on the sets, vehicles and props and another, texturing and shading.
Generally, this is something we prefer not to do, instead allowing the
animators more flexibility in their workload.
But large jobs with tight deadlines require delegation and splitting the
job into parts, each with its own lead animator, is a generally proven method.
week into production Anderson and his producer John Needham arrived with the
dialogue track. After more than 30
years they’d managed to get tow or the original voice artists back for the
recording. Anderson told me that when he’d been casting for the original
series he’d asked Cary Grant to voice the role of Captain Scarlet.
It didn’t pan out, but listen carefully and you’ll hear Francis
Matthews is a dead-ringer for Grant’s voice.
the audio track in place we moved into Shepperton Studios to the offices of
Centroid Motion Capture Services. We’d
realized in pre-production that we would never be able to complete on time using
traditional keyframe animation alone. Instead,
we’d have actors mime to the dialogue track and use motion capture technology
to record their body movements. The concerns with using motion capture, however, were
two-fold. Firstly, we’d not yet
worked with motion capture data in Maya (our main 3D software) which at that
time was still in release 2.0. Secondly, the recorded data is notoriously hard
to edit and tweak once you’ve got it. So,
we decided to use a mixture of both keyframing and motion capture and got our
lead R&D programmer Jonathan Stroud to work on an animation blending system.
The result was a set of controls that allowed the animators to
blend any percentage of motion capture data with keyframe animation on any joint
at any point during a shot.
I should say that we’ve been using May at MPC since its first release, along
with Renderman for most of our rendering and although Maya has an extensive
toolset, we’ve found the need to write many additional plug-ins including our
own sub-division surface modeling tools. These
tools allow animation to build and animate very simple models and have them
sub-divided at render-time into smoother more defined shapes.
Also, since they’re based on polygonal geometry it allowed us to take
advantage of a very efficient texturing method which, I think, needs to be
explained in a little more detail.
1: shows the SPV TANK MODEL IN Maya. We knew from the storyboards that it featured in about 18
shots, often very close to camera, and that it would need very hi-res textures.
By using sub-division surfaces the model consists of only about 30
surfaces (unlike NURBS which would normally consist of many times that).
By combining this with the Texture View feature in Maya, we decided to
texture the entire model (except the engines) with only one colour map file,
rather than one map for each surface. Fig
2 shows the ‘unwrapped’ model in the Texture View window with each element
of the model placed in its own location in UV space.
In the colour map (Fig 3), you can see each object has been painted in
the same unique location. By
applying this one map to every object, each will receive only its portion of the
texture file (Fig 4). Of course the
texture has to be very large (this one is 4096x4096 pixels) to ensure each
object gets the pixel resolution it needs, but Renderman is very efficient at
handling hires images. The benefits
of this approach aren’t earth shattering, but it does simplify the texturing
process as well as the project’s file management.
When you’ve already got 400-plus textures files, every saving helps, so
we used this technique throughout the project.
With the motion capture session over, we returned to MPC and set about building an animatic from the recorded data. This was a hardware rendered version of the entire film with motion capture skeletons running around in place of the final characters and it allowed Anderson to direct the camera and editing and see the results instantly. Once he was happy, we began to replace the rough animatic models piece-by-piece with the final textured versions, added the characters heads which had been lip-sync’d separately and hit the big red render button. Of course, things are never that simple in 3D land and every step held its own challenges, but nothing was insurmountable.
the end of the project, as we saw it take shape, I found I had to question why
we were remaking a series which was perfectly good the first time around, and it
was Anderson who gave me the surprising answer. He said that for everything his shows were, and everything
they’ve now become, he personally was never satisfied with the constraints of
the technology of that time. The
shows weren’t conceived with puppet actors or model sets in mind.
They were simply the only technology available.
And as much as we love them – and they encapsulate the charm of a time
we reminisce over – he always wanted to rid himself of those constraints and
to work without limitations. “Compluter graphics is just the next best
technology,” he said.*
addition to interior and exterior environments, MPC’s animators created an
assortment of unusual vehicles and effects, the trademark of a Gerry Anderson
Chas Jarrett is a CG supervisor and senior animator at MPC in Soho, London. For more information on MPC go to www.moving-picture.com
BACK TO THE CGI NEWS PAGE
BACK TO SKYBASE CENTRAL HOME PAGE
BACK TO THE SPECTRUM HEADQUARTERS HOME PAGE