Tuesday, August 18, 2015

A Microcomputer Behind the Video on 2010

The release of 2010, sequel to the 1968 film 2001: A Space Odyssey, represents the dawning of more than one age. Story line aside, the film is perhaps the first to give major roles to two emerging technologies of our own era: video and microcomputing.


Solid modeling system responsible for instrumentation sequences

"More than any other film, 2010 offers video and computer graphics their best exposure," says John Wash of Video Image, the production company responsible for nearly all of the onboard control panel displays seen in the movie.

Video Image produced more than an hour of computer animation for the film, and produced it, remarkably enough, on a system built around an IBM PC microcomputer.

The Polycad/10 solid modeling system, developed by Cubicomp Corporation (Berkeley, Calif.) produces three-dimensional images, either as wire-frame models or "covered" with shaded surfaces, that can be rotated, repositioned, and rescaled on a medium resolution color raster display. Only a few years ago this capability would have required a minicomputer or mainframe system - a prohibitive cost for a small production facility.

Also used on the project was Cubicomp's 2-D paint program, which runs in conjunction with the Polycad/10; and Hal, a package developed by Video Image that acted as the master control for animating the sequences.


Video on the set

Operating a stone's throw from MGM Studios where 2010 was shot, Video Image was established in 1983 by Greg McMurry, Rhonda Gunner, Richard Hollander, and Wash.

The four of them cut their teeth working with effects wizard Douglas Trumbull on Blade Runner, Brain Storm, Star Trek - the Motion Picture, and other projects.

Video Image specializes in providing video production services for 24-frame playback (versus the standard but incompatible 30 frames-per-second) on movie sets-creating the images inhouse, feeding them to CRTs during filming on the set, and ensuring that the images are uniform from take to take and fully synchronized with the camera.

According to Richard Hollander, Video Image won the bid for the film because of its ability to handle the entire project. It had the in-house artistic talent to develop the images and the technical expertise to deliver those images to the set.

The production company also offered the economy of a microcomputer based graphics system. "We were able to get away with less costly hardware because we weren't trying to depict a real object with computer graphics-the audience knows it is looking at a display screen," said Hollander. "The resolution on the Polycad/10 system is 512x 512-quite sufficient for our purposes. More computing power would have been overkill."

This is an image of what is seen by the space crew when sent to
monitor a mission probing the surface of Europa, one of Jupiter's
moons. In the film, a life form is detected on this moon and a
probe is sent out to investigate
Much of 2010 takes place on the Russian spacecraft Leonov, launched in a joint USA/USSR venture to dock with the marooned ship Discovery, now orbiting Jupiter. The interior of the ship is filled with color monitors used to provide instrument readout and other information. Video Image's mission was to provide the images for those screens as well as recreate some of the earlier screen visuals for the Discovery.

A comparison of 2001 with 2010 shows just how far video technology has come. According to Wash, the instrumentation in the first film incorporated only a couple of computer generated sequences, used to rotate three-dimensional objects. These were produced frame-by-frame on a penplotter and photographed on an animation stand. The great majority of images used traditional animation techniques with no computer involvement whatsoever. Instead of CRT displays in the set, rear-projection was used to bring the images to the set.

By comparison, the instrumentation in 2010 relies heavily on CRT screens, and the images, while not actually created in real time (they were shot frame-by-frame), are the stuff of silicon chips and software. Video Image was able to set up its post production playback facility on the Leonov set, enabling it to edit and modify animation sequences.


Two kinds of images

To produce the images, Gunner and Wash began by reading the script while referring to set design drawings supplied by Syd Mead.

The script made it apparent from the start that two types of images would be required: specific sequences vital to the plot, and others, more generic, that would appear only as background.

This is a wire-frame image of the
Discovery. This image is what the
Leonov crew sees in its monitor as
two members of their team attempt
to board the Discovery.
"We began researching to determine what exactly is the state of the art for this technology, talking to designers of aircraft-both commercial and military," said Wash, "and we tried to find out what these people were looking at for the future.

"The trend does indeed point toward the replacement of the myriad dials and gauges found on traditional craft with a few CRT screens that give multifunctional readouts. Computers on board the craft will determine what information the pilot needs to see, or alternatively, the pilot will be able to request specific read-outs manually."

To produce the images, Video Image purchased the Polycad/10 system in October 1983. The system augmented the production company's existing graphics system, built around a Cromemco microcomputer.

The Polycad/10 system produces three-dimensional, shaded-surface, fullcolor solid models using an IBM PC or compatible-a capability once confined to larger systems priced at $100,000 or more. The Cubicomp system enables filmmakers to create and manipulate images interactively, using a graphics tablet and keyboard, and macro commands listed in an external file, residing on disk.


Artistic license

Video Image took responsibility for its own art direction, running Polaroid screen shots and completed tape sequences past director Peter Hyams on a create-as-you-go basis. The production company began work several months before shooting commenced, and stayed two weeks to two months ahead of filming, depending on the volume of images required.

"It was obvious from the beginning that we would have to invent a lot of general material," says Wash. "For example, we created screens that monitored the functions of the ship, that showed communications, that listed daily activities. We provided static diagrams of Jupiter, navigation readouts showing star positions, and radar sweeps. Plausibility was the key; beyond that we were afforded a great deal of artistic license."

But other scenes required very specific images integral to the story line. For example, the company was asked to create a graphic sequence that matched a shot of the pod flying over the surface of the Jovian moon Europa - the graphic look-alike to represent a computer analysis of the terrain. During production, the model shot and its graphics counterpart would be played side-by-side on adjacent screens.

Hollander started by patching the photographs of the model together, like a mosaic, to produce a single, cohesive view. He then digitized the results, using a graphics tablet, and manually added some of the more prominent geographical features. A screen showing the model fly-by was placed above the screen of the Polycad/10, and Hollander began duplicating the camera pans from the original. "I must have viewed that sequence 5,000 times," he recalls.

Some of the company's earliest work involved the creation of "stock characters": the Leonov spaceship, for example, as well as the Discovery, the Europa pod, and even the ubiquitous black monolith.

Unlike traditional two-dimensional animation, however, where a character is first drawn blueprint-fashion from several angles, a three-dimensional system requires the creation of a computer database, consisting of an x, y, z coordinates listing. The images created from that database are similar to a snapshot of an actual object in that the viewing direction and distance can be selected. Moreover, some images are left as wireframe models while others appear with completed surfaces.

This is one section of the Russian ship Leonov.
It was used as a background detail in the film.
The Leonov database, for example, was created from an orthogonal projection drawing of a 10-foot model built by Mark Stetson for MGM. Over a two-week period, Hollander digitized these drawings, selecting areas of the blueprint to be entered as two-dimensional polygons. These polygons were then extruded, multiplied, "swept," and otherwise transformed into 3-D images.

The Polycad/10 system uses the boundary representation method, in which a 2-D polygon is the foundation for constructing a 3-D image. For example, a pentagon may be extruded to form a pentagonal prism, or multiplied to create a dodecahedron. Likewise, a half circle can be "swept" to create a sphere, and a full circle can be rotated around an axis to a donut shape.  Several such objects can then be combined to create a single database.

"Given this arsenal of techniques within the Polycad system, you can generate very interesting, well-defined solid models," says Hollander.

In the Leonov's case, two databases were required to show the rotation of the control center. Once they were complete, the ship could be incorporated in scenes as needed.

The image shows the spacecraft entering an orbit around
Jupiter with the ballute making a protective path.

One of the more elaborate examples depicts the Leonov as it skirts Jupiter, using the planet's atmosphere to decelerate. To minimize the heat, a "ballute"-a combination balloon and parachute-is inflated and ejected out the rear of the ship.

"At that point in production, nobody was certain how much of this action could be shown through model photography," Hollander recalls. "So it was determined that, at the very least, the video graphics should record the event."

The sequence shows the red ballute emerging from a wireframe image of the ship. The ship is then rotated to the side, and the airflow is traced.

To animate the sequence, Wash used the Easel system to hand-draw the ballute as it filled the screen, with each frame consigned a separate disk file on the computer. The Hal control program then read in the files in sequence, overlaying each with the Leonov database in the framebuffer.

Color animation was also used. The ballute image was shown pulsating in intensity between a dark burgundy and crimson. To accomplish this, the Hal software cycled through a predefined 24-bit color lookup table that changed the ballute's colors dynamically from one frame to the next. (The picture elements within the Polycad/10 system are governed by a "lookup table" of the primary additive colors-red, green, and blue-each of which can be displayed in one of 256 intensities. Up to 4096 such combinations can thus be cycled through to display a palette of 16.7 million colors.)

In the film, a computer creation called a ballute (combination balloon and
parachute) is launched to allow the Leonov to enter Jupiter's atmosphere
to slow down the speed of the spacecraft. The image shows
the ballute being inflated.

To simplify rotation, the ballute was changed from a two-dimensional overlay into a 3-D wireframe. Colormapped animation was used, this time to animate the patches representing air flow-a process analogous to light "moving" on a theater marquis.

Other overlays used for the sequence included Russian lettering (created by a typesetting program that added a second character set to a computer terminal), a digital readout, a blue border, as well as the rotating center of the Leonov, which had to be redrawn every frame.

According to Wash, one of the major benefits of producing graphics with a computer is speed. "For example, on one days' notice, Peter (Hyams) decided he wanted to show the ballute being jettisoned before the ship is rotated. We managed to accommodate him overnight.

"The process was almost like traditional animation in the sense that John made individual pictures of how the ballute was going to animate off," Hollander explains. "When we did the shoot, we would read in the 3-D rendering of the ship and overlay John's pictures sequentially in the order that he planned the animation."

"I've worked with a lot of films that generate graphics using traditional animation techniques to mimic a computer graphic readout - and I admit I was at first skeptical that a computer could, in fact, do it faster," Wash recalls. "But in fact, we could never have generated that shot in time without the microcomputer to manipulate the images."

In the sequence, the Leonov ship was depicted only as a wire-frame model. But some images used solid fill as well. One sequence, for example, was created to show a medical analysis of a woman cosmonaut. The production team began with a plastic foam model that artist Peggy Weil covered with interlocking polygons. Hollander then used a milling machine to record on paper the coordinates of the vertices, which in turn were merged into a Cubicomp database.

For the first part of the Leonov mission, the American Team is placed
in deep sleep. When Dr. Heywood Floyd (Roy Scheider) is brought out
of this sleep, a monitor charts his progress. The monitor is an
animated computer graphic.

The completed image shows the wire-frame head rotating, a deep blue surface added, and then sections of the head changing colors.

Once a sequence was approved, Hollander and Wash filmed it off a 13-inch raster tube. A negative of the tape was then given to an outside lab for conversion to 24 frames/per second, and given to McMurry or Gunner for use on the set.


Play it back

At the MGM soundstage, Video Image installed a playback system consisting of custom switching devices, a patch panel for cabling, assorted monitors, an editor, and a custom video synthesizer-all modified for 24-frame video. Ten machines could be run in the system simultaneously by one operator, and this permitted the operator to cue the sequences to precisely match a retake.

"We actually had 13 machines and 13 feeds, and sometimes produced duplicate images," Hollander explains.

"But when this occurred, we spaced the redundant ones around the set in a logical arrangement."

In all, the firm produced more than an hour of video animation. "You never know how long the director is going to shoot for," says Wash. "It may look like 30 seconds in the script, but the sequence may go on for three minutes or longer-so we generated all the material way over length.

Since completing 2010, Video Image has put the solid modeling system through its paces on three other pictures: the forthcoming Michael Crichton film, Runaway; Hal Barwood's Biohazard (working title); and John Carpenter's Starman.


(Originally published in part in C&GA, February 1985)

Monday, August 3, 2015

An Odyssey for Arthur C. Clarke

by Steven Jongeward & Gerard Raymond


Last May, Arthur C. Clarke visited the MGM soundstages in southern California where principal photography was underway on 2010. He was quite pleased with what he saw. Stopping in New York on his way home to Sri Lanka, he chatted about the eagerly awaited film.

Although obviously enthusiastic about the project, Arthur C. Clarke respects the studio's wishes not to reveal too much too soon, and is a bit circumspect in his comments. Having just watched the film's trailer (playing with 70mm versions of Indiana Jones and the Temple of Doom), he is still bubbling from what he describes as "a rewarding experience. It was quite a thrill seeing the trailer," he admits, "and I was very happy with the audience's reaction."

Clarke is also extremely pleased with having filmmaker Peter Hyams as 2010's guiding light. Hyams, the film's writer, producer, director and cinematographer, was happy to have Clarke's help in restructuring the story to translate it from printed page to silver screen.

"All the changes that Peter has made were quite justified," Clarke states. "If I had thought of them myself, I might have used them in the novel."

One of the problems in shaping the story into screenplay is that Clarke was not thinking cinematically when he wrote it. "I honestly didn't think about a film even for a minute," he confesses. The challenge was large enough, he says, that Peter Hyams complained and suggested that the next book Clarke writes should be entirely about two men sitting in a room.

Clarke admits that while writing 2010: Odyssey Two, he felt that he had a story "that even Stanley [Kubrick] couldn't film!" But Hyams accepted the challenge. Because of the distances involved - California to Sri Lanka - the two men collaborated via satellite through their personal computers. The "electronic mail" that passed between the pair will be collected and published later this year, under the title The Odyssey File.

The author is confident in Hyams' ability to follow Stanley Kubrick's landmark motion picture. "First of all," he says, "Stanley approved the choice." In addition, Clarke explains that, "I had seen Capricorn One and Outland and I enjoyed them thoroughly ... even if I did dislike some things in both movies."

Clarke feels that the Hyams film is not really a sequel to the Kubrick classic, but that "comparisons are inevitable. But they are not fair because this movie is being made in a totally different age." He also notes that 2010 will benefit from current technology. The look of the film will be more accurately reflective of real space hardware.

Peter Hyams has said that 2001 was a cerebral film, while 2010 will be more of an emotional adventure. Clarke agrees. "There was no point in developing any emotional background to the characters' personalities in the first movie."

Roy Scheider, director Peter Hyams,
and Arthur C. Clarke chat about 2010.

He is also quite comfortable with Hyams' nuts-and-bolts approach. "I have been accused of over-explaining things, but being a serious writer, I rather prefer to over-do. Stanley, on the other hand, has been accused of the other direction."

On his way back from Los Angeles to New York, Clarke made an important stop-over in Washington, D.C. He had an appointment at the White House.

However, he didn’t meet with the President inside, but with a film crew outside, shooting a scene from 2010 with Dr. Heywood Floyd (Roy Scheider). Clarke was there to do a cameo in the scene – as a disheveled drunk lounging on a park bench.

But he couldn’t report on the dialogue from the scene, because he has been concentrating immensely on “just feeding breadcrumbs to the pigeons and occasionally swigging from a bottle in a paper bag.” He explains that he was given no actual direction. “Peter Hyams left it entirely up to me,” he says with a grin. “It was not,” Arthur C. Clarke confesses, “an Oscar nomination bid. But, if you can't be a successful bum, being a writer is the next best thing.”


(Originally published in Starlog Magazine)