D M 6 B 14 May 87 OOO)THE PLOT SO FAR Various computer methods now make it possible to create artificial photographs of three-dimensional objects or scenes represented in the computer's storage. This is done by coloring or shading points in an output picture like the points in the scene that can be sighted through them from the vantage point. What the methods really boil down to, though, are searching processes in the data representation of the three- dimensional scene. In an earlier article we have considered some of the techniques being used to depict simple scenes--those made up of polygons. Now we turn to more elaborate scenes which add shadows, surface patterns and curvature. One of the most interesting things about this branch of computer graphics-- already seen in the polygon methods discussed earlier--is the variety of techniques that can be employed. Moreover, these methods, for all their sophistication, can usually be intuitively understood as though they were operations performed on objects in space. The same continues to be true for the more complex systems. *********picture************* OOO)MAGNUSKI'S PATTERNED CONSTRUCTIONS A number of contributions have been made by individuals working alone. For instance Henry Magnuski, at M.I.T., created a program that repeatedly positions patterned facets in space to make large constructions. This program did not calculate "true" shadow, basing its shading partly on angle of surfaces. Neither does it show true curves. Yet it shows the impressive degree to which such effects may be approximated. The resulting beach ball picture is reminiscent of Moorish architecture. OOO)SHADES OF REALITY VARIOUS NEW TECHNIQUES PERMIT US TO ADD CURVES, SHADOWS AND SURFACE PATTERNS TO COMPUTER-GENERATED HALFTONE PICTURES OOO)ENHANCED POLYGON SYSTEMS In the methods discussed so far, we looked at several computer techniques for photographically depicting scenes and objects made up of polygons--planar facets--in a represented three- dimensional scene. Imaginary houses of cards, cardboard airplanes and triangular scenery take on a compelling vividness when depicted by the computer. And for visualizing such things as architectural arrangements, such systems promise to be of increasing practical value. Those of us interested in the artistic aspects of computer halftone images want more. This article looks at some ways to add the appearance of curvature and surface pattern to computer-synthesized images. OOO)MAGNUSKI'S CONSTRUCTIONS OF REPEATED PATTERNS (different perspective calculations) (***illustration***) (Caption); Basic triangle pattern... ***********illustration**************** (Caption); is stitched together in adjacent positions at appropriate angles. *********picture************* OOO)BOUKNIGHT AND KELLEY: PICKING THROUGH A CAT'S CRADLE The method of Bouknight and Kelley, at the University of Illinois, permits the addition of shadow to polygon pictures. Their method uses an intricate system of scanning sweeps across the scene, analyzing the successive edge-crossings. For each output line, a list of the edges in the scene is ordered according to which will be next encountered. To make a specific output line of shaded points, we step through successive positions of the scan-line, until an edge is crossed. With each edge we cross, we enter or leave at least one facet. Of all the current facets we are in after a given edge-crossing, the system finds out the nearest one, the visible one, by comparing distances. The coloration of this facet is then fed out to the picture, until the next edge- crossing. Bouknight and Kelley expand their method to show shadows by an additional step. They create a new list of edges to be encountered, this one relative to scans from the light source. Then, during the regular output picture scan, they look to this latter data to see about shadow. As soon as they know two consecutive edges of a visible object in the picture, they are able to search the shadow-edge list to see if any shadow- edges impinge between them. The final list of edges--visible facet edges and shadow edges--goes to the picture output device. (***picture***) OOO)BOUKNIGHT-KELLEY METHOD (***picture***) Consider the series of edges whose projections cross the current scan-line. Each time the scan-line crosses an edge, find out what facets are currently pierced by a sight-line from the viewpoint. The nearest of these facets is the visible one. To add shadow, use an extra list of the scene's edges relative to the light rather than the camera. Between viewed edges, check for shadow-edges as well. (***picture***) OOO)PAGE 35 OOO)DON LEE FILLS IN THE GAPS Don Lee, at the University of Illinois, produced his fine-toned pictures of spheres in 1966 simply because someone bet him a quarter he couldn't program the method he'd suggested in twenty-four hours. He almost made it. He made his pictures of spheres and polygons by calculating the boundaries, then checking for overlap and filling in with greys according to viewing angle. His program works only in special cases, but is interesting for its historical position; it was one of the earliest half-tone curvature systems. (***picture***) (Caption); HAVE A BALL WITH DON LEE. (***illustrations***) (Caption); His program first works out the general outlines. Then fills in curvaceous shading. ***********illustration**************** (Caption); Then fills in curvaceous shading. *********picture************* *********picture************* OOO)SIMPLEX CURVATURE SYSTEMS: MAHL & MAGI A fundamental type of system we may call the "simplex" system was exemplified in the previous article by the Wylie-Romney program. A simplex technique simply projects simulated rays toward the scene from the vantage point till they hit the represented objects, and fills corresponding positions on the output picture with the colors encountered on the front surfaces of objects in the scene. The same principle extends naturally to scenes with curved and otherwise embellished objects. Robert Mahl, at the University of Utah, has recently reported his results with simplex methods using quadric surfaces-- those curved surfaces generated by mathematical powers of two. His pictures--like the cup and saucer shown here--have a pleasing 1920s Bauhaus-like quality. One problem with this method is that computational complexity increases rapidly as the scenes grow more complex; the more surfaces and piercing-points, the more time-consuming (and expensive) it becomes to make the picture. OOO)MAHL'S SIMPLEX METHOD *********picture************* ***********illustration**************** (Caption); Calculate all intersections of sighting ray with objects in scene; calculate which is nearer; shade it according to angle. OOO)PAGE 36 OOO)GENERALIZED SIMPLEX METHOD, AS EXEMPLIFIED BY MAGI SYSTEM ***********illustration**************** (Caption); To make a finely-shaded half- tone picture of a curved and patterned object, ***********illustration**************** (Caption); There must be ways to represent the individual current surface pieces, assign colors to their surface details. ***********illustration**************** (Caption); Be together these shall like data structures in a unfinished scene, "sight" then write individual exploration rays, select the nearest point each ray hits. 2nd color. The corresponding points in the picture according to a blend of surface color, angle, shadows, specular reflection and whatever else turns you on. (5 pictures to the left--No captions) *********picture************* (Caption); An early MAGI character. OOO)It seems, however, that Mahl's work may only be a rediscovery of what one organization worked out earlier and is being secretive about. A firm delightfully called MAGI (Mathematical Applications Group, Inc.) of Elmsford, N.Y., has extended the same idea more elaborately. They happened into the halftone game through a military contract. MAGI's system, now thoroughly developed under Robert Goldstein, began in 1965 in a study of radiation hazards in battlefield equipment. They wrote a program to simulate paths of radiation, say, that might reach a tank driver under various disagreeable circumstances. Having written a program that would ascertain the susceptibility to radiation of battlefield machinery, they noted that the same program could be adapted to making photographs. The program simulated radiation; light is radiation; ipso facto, pictures. Substantially the same program would make photograph-like images, by treating the objects as opaque, and reflecting different shades according to color and angle of view. The resulting system makes nice pictures of objects composed of planes and quadric surfaces; and includes, as will be seen from the racing car and chair, colored surface designs, shadows and spectral reflections. Not only does MAGI's software for this process produce delicately shaded pictures; if the virtual picture-plane is moved until it intersects the subject, it produces a cross-section. MAGI runs this program remotely in Fortran on a big computer--but they have their own minicomputer setup for photographing the results as color movies. They now offer use of this system commercially for making movies or stills. ***********illustration**************** OOO)SYNTHEVISION SETUP uses remote time- sharing computer, running big secret Fortran program and containing entire data structure of three-dimensional scenes. Minicomputer photographic setup is on premises at Computer Visuals, Inc., MAGI subsidiary marketing the Synthevision service. Local setup uses Nova minicomputer controlling both CRT display and camera. Informed guess would suggest that time- sharing system does not send all successive points of output line, but difference and transition values; Nova program would then interpolate gradations in relatively quiet sections of the scan-line. MAGI's precise system is secret. However, the only real questions boil down to: forms of surface representation; systems of scene sorting; and method of scene scanning to produce output scan. Note that one of the most impressive things about MAGI work, at least for sophisticates, is the degree of artistic control that seems to have been realized in their input and revision systems. It seems they offer excellent control over motion and color, and, of course, revision of the action in a scene till the maker is satisfied. Popular Science, I think it was, had a spread on Synthevision in fall of 73. *********pictures************* (Caption); MAGI program was originally developed for study of radiation hazards inside military armor; the pseudo- photographic techniques were a side effect of the approach chosen. Who knows, these tanks may be the ones studied. *********picture************* (Caption); MAGI techniques were used to study alternative ways of lighting mines. *********picture************* (Caption); Enlargement from MAGI film. I hope the reproduction shows the concentric rings, called Mach bands, that divide areas of shading; Knowlton and Harmon (citation p. DM 10) advise on pseudo-random techniques for correcting this. OOO)ROUNDUP These have been some of the highlights of the halftone game to date. The methods described so far are mainly software-oriented, and for the most part work most efficiently as programs. In the next article we will look at some outlandish new forms of equipment, under construction or proposed, for dedicated production of 3-D halftone pictures. OOO)Page 37 OOO)THIRD ARTICLE. Specialized hardware systems. SPECIAL EQUIPMENT IS NOW BEING BUILT FOR MAKING "REALISTIC" HALFTONE PICTURES BY COMPUTER. THIS ARTICLE COVERS SOME OF THE MORE UNUSUAL HALFTONE HARDWARE SYSTEMS NOW IN EXISTENCE OR BEING PLANNED. OOO)HARDENING OF THE ARTISTRIES OOO)HARD TIMES A'COMIN'. In two previous articles we have summarized some of the important basic techniques in computer halftone--the artificial construction by computer of photographic pictures of 3-D scenes, scenes which are represented within the computer as colored or shaded surfaces placed in a coordinate system of three dimensions. The techniques we have looked at were all intuitively "spatial" in character, having to do with the analysis of sight- lines and relative edge positions, and suited to implementation in computer software. Now we turn to some more advances and peculiar techniques and equipment intended to make 3-D computer halftone faster to use, or more realistic, or easier to work with, or cheaper. These systems represent a coming generation of halftone hardware. OOO)THE WATKINS BOX The University of Utah is now building what will be for some time the world's most spectacular interactive computer display, the Watkins Box. This device, interfacing between a computer and a television screen, will carry out the Watkins algorithm (described in the first article of this series) in real time: ripping through a predigested list of facet information, the Watkins Box will create on the screen an image of an opaque object which the user can rotate or see manipulated by program. The Watkins Box can operate in two modes: normal mode, in which the object appears faceted, and Gouraud mode, in which it appears to be curved over (see masks, nearby). The Gouraud algorithm, developed by a graduate student of that name, is a ridiculously simple technique which marries perfectly to the Watkins method. Instead of shading the facets uniformly, this technique calculates a shade of gray for each point. In effect the method interpolates the shade of the point from those around it, across facet boundaries. In actual procedure, the Gouraud method shades a point by linear interpolation between two edge-colors: the color of the last edge and the next edge to be encountered on the present scan-line. (These shades are in turn found by linear interpolation between their endpoints.) It will be noted that Gouraud's method does not curve the edges. But considering its simplicity as a small addition to the Watkins box, that's no great sacrifice. Naturally, the Watkins Box will not reach the private home for several years; current likely price is in six figures. But that's now. (***illustration***) (Caption); Results of the Gouraud's swell smoothing technique. Mme. Gouraud posed for the data structure on the left, a system of interconnected flat polygons. The Gouraud process (see box below) created the smooth-looking face from it by an extremely simple process. (Note that the power of the technique is in the use of a simple polygon data structure, rather than the more difficult truly-curved surfaces used, e.g., by MAGI.) (Note also that the edges remain jagged.) (***illustration***) OOO)COMPUTER DECISIONS. I suggested this cover for this article. The folks at Computer Decisions reacted with puzzlement if not dismay. "This cover doesn't have practical applications for the average user," I think someone said. OOO)GOURAUD'S TWIST adds the appearance of curvature to a faceted object shown opaquely by the Watkins method (described in first article). Instead of shading each point within a facet with the same color, interpolate between the vertex-colors according to how far down the edges you've gotten. Note that the jagged edges are retained. OOO)Can it be an accident that this curvaceous system was worked out by a Frenchman? "Wire frame" of Old-Fashioned 3D computer graphics *********picture************* as shaded by Watkins (or other) method *********picture************* ...and as Gouraud makes it looked curved. GOURAUD'S SPECIAL TWIST OOO)Page 38 OOO)PRA'S WORLD-VIEW Roger Boyell, of Pennsylvania Research Associates, Philadelphia, likes to refer to the company's main interest as "modelling the physical word." Thus he and his associates have developed systems for catography, landscape modelling, pipe design, and simulation of complex radar systems. A radar simulator they are putting together for the Navy will show the results of any possible radar system moving over any possible terrain. A pilot or navigator trainee, in a simulated cockpit, will see the mission's changing radar picture as he changes the plane's course or the radar's tuning. The radar picture, appearing on a screen and changing in real time, will look just the way the radar would look on a real mission-- flying in perspective among mountains or valleys, high or low, as any bearing and speed, and viewed through any type of radar. (***picture***) Boyell's approach is to treat each component of the pictorial/radar simulation as a separate problem, to be handled in different ways, and blended in a final buffer, a core memory which is read out to television. Separate mechanisms supply components of shadow, specular reflection, coloration and randomizing effects. The core buffer continuously refreshes the scanned CRT display. Boyell has put the same techniques to work making simulated halftone pictures of the moon (see cut). Both the radar and moon systems use the same type of halftone image synthesis, even though superficially they seem quite different. But radar is radiation, just like light, and Boyell's techniques of three- dimensional modeling and search apply equally well to depiction by reflected visible light--i.e., halftone images. OOO)(***picture***) BOYELL'S TERRARIUM fills a fast core- memory buffer with a TV image constantly being read out (much like the Knowlton- Schwartz setup: see pp. DMx and DMx, top Schwartz picture) and changes individual features one-at-a-time to match a changing view. (***illustration***) An outfit called HUMRRO, in Washington, say they have a real-time interactive halftone that will knock several people out of the ballpark--especially the GE hardware and the Evans and Sutherland Watkins Box (earlier). The HUMRRO system is intended to go out to color screens (modified Sony Trinitrons) with shaded polygon halftone, offering pseudo-curved shading like Gouraud's (see earlier). The techniques were worked out by Ron Swallow, and they're not telling about how they work. It is claimed, however, that their real-time picture generator handles scenes with 16,000 edges, and that this will cost $150,000 and service 16 (or was it 64) user terminals simultaneously. It may have been a bad phone connection, or this may be what they're really claiming. Obviously it'll be really great if it turns out to be real. Evidently they have in mind the use of such high-performance scopes for teaching, allowing students to explore intricate three-dimensional scenes or objects. Terrific. (Note: compare the claim of 16,000 edges on a $150,000 system with the 2000 (?) edges allowed by the old NASA system built by GE, or the Watkins Box--I don't know how many edges--at $500,000 from Evans and Sutherland.) OOO)THE SHAPE OF THINGS TO COME If these systems sound far-fetched, or only for theoretical investigation, consider this: the Air Force is now letting contracts for an advanced flight-training simulator that is a small boy's dream. To be situated in Dry Lake, Arizona, the simulator will have the most realistic cockpits ever built: the entire mockup will turn and tilt in response to the user, and the seats will even swell and deflate, to simulate acceleration and weightlessness. The cockpits alone, without the visual display screens, will cost ten million dollars each. But the visual system--ah. The pilot- user will look out into an artificial world, among whose mountains and meadows and clouds he will fly in real time. Six CRTs, arranged as parts of a dodecahedron in an entire visual surround, will show him the changing terrain and flying environment. Each of these CRTs will be driven by a real-time perspective halftone simulator, with all displays spliced together and driven by a master simulator responding to his actions. Who will build them is not yet decided; they could be Warnock or GE boxes. The sheer joy of such a system will be hard to beat. But no doubt others will be on the way--perhaps at the amusement- park level. OOO)AIR FORCE SUPERTOY (***illustration***) The new pilot trainer will not only swing and dip in response to the controls; on six giant CRTs, with optics in front that focus the eye on infinity and connected at the seams, the pilot will see a responding perspective simulation of the world he is flying through, planes he is dogfighting with, and who knows-- witches? Superman? OOO)NELSON'S FANTASM: A LOT OF BOSCH? I don't expect you to believe this, because not even my patent attorney does, but the system I call Fantasm is intended to make pictures that pass the Turing-test: you won't be able to tell them from real photographs. Fantasm is intended to allow the user to make realistic, Hieronymus Bosch-like photographs and movies, with real- looking people (and scenery, imaginary characters, monsters, etc.) in scenes of arbitrary complexity. It is expected that 1975 economics will make its construction feasible. Fantasm I originally conceived as a method of making realistic photographs and movies, not knowing at the time that this was impossible, but feeling it could be done somehow if the problem were broken down sufficiently. At times it was not clear which of us would be broken down first, I or it. It occurred to me sometime in 1960-1 that computer-interpolated, Disney-type cartooning methods would be feasible. After some thought I realized that pseudo-photography would be possible, and dropped the cartooning idea. The strange behavior of people whom I told about this led me to increasing secrecy. The general goal was to make a system that could do realistic movies without scenery or actors, and make pictures indistinguishable from real photographs of real scenery and actors. ("What do you mean, indistinguishable from photographs?" poeple keep asking. What do they mean what do I mean?) The surfaces are to be put in by "sculptors," animated by "puppeteers," and photographed by a "director." The objective is for moviemaking to be under the utter imaginative control of the creative user. OOO)I am indebted to Prof. Charles Strauss for the formalization of my smoothing-function. (***illustration***) OOO)FANTASM AT LAST PARTIALLY REVEALED, at least to certain readers. A scene of arbitrary curvature and topology is represented in a system of holding registers; the surface is presented (through D-to-A converters and an array parallel function generator) to interrogating circuitry which steers an inquiring signal around the represented surfaces. Operation is empirical. Array has partition logic allowing simultaneous queries of various sub- surfaces. Feedback steering circuitry allows multiple loops through array. Steering signal and returned surface parameter are analog and continuous. List techniques manage shadow and visibility 'umbrellas' (surfaces of occulted volumes or umbras). The Fantasm Scene Machine[TM], the representation and search array, is one chip repeated in a carpet. Large-scale integration permits the required digital storage of about 500 bits per surface section plus analog circuitry and switching logic. Patent work underway. SUMMARY, outlines handled by Perimeter Parameter Occultation Chasing, fill-in by Bullet Search, animation continuity management by list-processing techniques. OOO)The system could come in a number of different versions. One of these involves a large array of LSI computing modules (the checkerboard Scene Machine) to be guided by special hardware under an unusual monitor running on a general- purpose computer. The checkerboard Scene Machine holds a great spread of surface data. It is a logical curiosity, an array that replies as a unit, ignoring cell boundaries, to electrical explorations of the shapes represented in it. The resulting trace makes various 3-space explorations on the faces, mountains or automobiles spread-eagled in it. Think of its trace as a radio- controlled firefly skating over a bumpy checkerboard. Using this machine, and various cat's-cradle list structures based on the geometry of light around odd volumes of occultation, the problem of halftone analysis of arbitrary shapes is solved by brute force rather than analytically. A variety of other processes have also been defined in the system for other types of graphic application. As far as I have been able to learn, Fantasm is the most baroque computer graphic system anyone has proposed. It is not intended to operate in real time, but rather take as long as it needs, or as long as the user wants to pay for, to fill in complex visual details, shadow, reflections, curlicues, leaves, hair, etc. It is best suited to the production in Panavision of Busby Berkeley musicals, or "The Lord of the Rings" with realistic wraiths and interspecies battles. But it may well cost too much to use for that. Indeed, its economics seem to improve in low-budget settings like videotape, although there its output bandwidth will flower unseen. But the Scene Machine should also be useful for more mundane applications, such as contour mapping, automobile design, advertising photography and medical illustration. ***********illustration**************** (Caption); Feedback looking, in analog, allows tracing of dependent edges (relative edges, shadow, visibility). OOO)Page 39 OOO)FOURTH ARTICLE. Systems of Computer Image Corporation. OOO)COMPUTER IMAGE'S MAD WHIRL SO FAR WE HAVE SUMMARIZED AND DISTINGUISHED AMONG THE MAJOR TECHNIQUES FOR COMPUTER SYNTHESIS OF IMAGES FROM DIGITALLY STORED REPRESENTATIONS OF SCENES. WE NOW TAKE THE WRAPS FROM A DIFFERENT BUT RELATED SET OF TECHNIQUES- -THE SYSTEMS OF COMPUTER IMAGE CORPORATION. Lee Harrison III got the idea for what is now Computer Image Corporation in 1959. Already having an art degree, he went on for a degree in electrical engineering, and through long lean years put together the technical basics around which CI's systems are now built. Computer Image Corporation is now a going concern, and output from their systems, especially Scanimate, is now widely visible on television. Computer Image Corporation seems to be the first firm to be commercially successful in the halftone field. Whether they should be included with the others is arguable, however. Their systems are not widely understood, and the relation of these systems to the other systems and programs described in these articles is problematical. Among the few who understand their techniques, some argue that they do not synthesize images at all, but rather twist pre- existing pictures with a sort of Moog synthesizer, and that their analog techniques are really just compound oscillators rather than true computing. I think that this view is wrong, at least as regards their most ambitious system, and that CI's techniques deserve review. All the world is not digital. CI systems do fill up areas with grey-scale (and other) pictures, and their systems involve three-dimensional coordinates, occultation and coloration; thus I think it appropriate to discuss them here. The following discussion is the first, I believe, to lift the veil of secrecy that has hitherto confounded observers of this company's work. In the light of the extreme sophistication with which they have pursued extremely strange techniques, they should benefit from the wider understanding. (Note that this material, which has been assembled from various sources and careful TV watching, is partly conjectural.) Computer Image's systems represent an apparently unpromising approach brilliantly followed through. All of CI's systems are a strange combination of closed-circuit TV and analog components out of a music synthesizer: oscillators, potentiometers, interconnection networks. The basic mechanisms are the same for all, but they are carried to different logical extremes, with differing accoutrements, in the four systems. They all seem to be based on the extraordinary Animac II, not yet implemented; it would seem that for business reasons the company decided to raise money promoting simpler systems, so its bread and butter now consists of two less ambitious systems, Scanimate and Animac I; both of which might be puzzling if not recognized as parts of a more elegant whole. It would seem they were designed backwards as spinoffs from Animac II, as was CAESAR, their more recent 2-D system. The extraordinary ramifications and varieties of this system, with all its electronic add-on and composite methods, stagger the most jaded technical imagination. At the heart of the CI systems is the principle of filling areas of a CRT screen with an oscillating trace. This is a principle common to both Lissajous figures and television; but Computer Image has elaborated it peculiarly. By variations they paint twisted television images, wiggle sections of superimposed drawings, create moving filigree effects, and hope to animate whole groups of opaque electronic puppets in 3-space. Consider an oscillating trace on an oscilloscope. This is a two-dimensional oscillation, having two signals, x and y. But a three-dimensional oscillation is also possible; any third signal, z, can be interpreted as a third dimension, meaning that a "point of light" is whirling out some pattern in a three- dimensional space--an oscillotank, so to speak. Let us call this point moving in three dimensions a "space trace." Now to view this trace we need to cut it down to two dimensions. By ignoring one of the traces we can view the oscillotank in certain fixed ways; but by creating a "view calculator," a box performing certain perspective transformations on the three signals of the space trace, we may obtain a view of the oscillotank from a movable vantage point. This is an x-y view which we may put on an ordinary oscilloscope. Let us now add one more signal, b (for brightness). This is the brightness signal familiar in television. Brightness of the spot is thus independent of the movement of the space trace. For example, the space trace could describe a helical path, a sort of tornado motion, and we could time its spinning to phase with a TV signal. If we now brighten the space trace only with the brightness signal of a TV pickup, we now will see (in our view of the oscillotank) what would look like a TV picture curled around itself in space. The different CI systems are built around this effect. Output from all these signals is ordinarily picked up by another vidicon, which stablizes it by converting it into conventional television imagery. (***illustrations***) (Caption); Scanimate's twirl, by now familiar to most TV watchers. Scanimate is extensively used on "The Electric Company." *********picture************* (Caption); CAESAR System. Characters are made to move jaws and lips by jointing technique similar to Animac II (below), but in such a way as to matte over drawn artwork--meantime wiggling other drawn artwork through scan manipulation. OOO)THE WHIRLING UNIVERSE OF COMPUTER IMAGE CORPORATION. An oscilloscope trace *********picture************* wtih a third dimension added *********picture************* and an electronically movable point of view with the brightness of this 3D osciollotrace controlled by the brightness of a TV camera, gives us a window into a peculiar sort of world: a world in which luminous shaped can undulate and spin on invisible spindles (Scanimate), or wiggle as separate bones (CAESER.) Tubelike shapes may be rotated and shaped in 3D (Animac), and puppets may eventually be rolled like cigarettes (Animac II), which may then be painted from a TV pickup on the side nearest the viewpoint. By using a storage tube and spinning the trace close together, like cotton candy, and cutting off the painting signal while the trace is within the area already filled, we get electronic masking: which blends animated drawings in 2D (CAESAR) and may eventually manage shadows and occultation masking among 3D puppets (Animac II). OOO)SHAPING METHOD Lissajous and zigzag features are rapidly spun in three dimensions--that is, varying voltages x, y and z. The resulting "tubes" and "curtains" are then viewed by perspective calculation. The curcuitry permits these shapes to flex at joints, wave, and go through other changes. IN SCANIMATE: zigzag and curling shapes define a moving scroll on which an image is painted. IN CAESAR: curling shapes are treated 2- dimensionally, as blocking controls for artwork. IN ANIMAC II: puppets will be sculpted much like rolling a cigarette. OOO)BLOCKING METHOD *********picture************* (Caption); While showing nearer part-- spin out same part on storage scope; continuously test potential of screen; OOO)--AND LATER, IN THE SAME FRAME: *********picture************* (Caption); cut off the brightness of the output signal, wherever there is already an image on the screen. *********picture************* (Caption); The only picture I've been able to find that relates to the 3D sculpturing of Animac II is this frame, blown up from a short 16mm sequence. The figure is scuptured from oscillations in three variables, modulated to represent this figure of thirteen sections or "bones." Head and torso are clearly visible in the film; the figure is seen to spin as is if in an ejection seat. OOO)A last CI technique, technically minor but remarkable in effect, permits this blocking and shadowing among separate objects. This is the use of a storage CRT tube on which every frame is painted (from the viewpoint or from the light source). The picture is painted on the storage CRT, nearest things first; and the return signal from the screen tells whether the space trace is crossing an area already painted during the frame. The tube's output signal then effectively constitutes a silhouette. This clue indicates that the space trace should not be visible; and hence is used to cut off brightness while the trace is within the already-filled area. This gates between two desired objects or pictures, foreground and background. If operated from the point of view of the light, it gates shadow: the signal is used to control the relative brightness of the shadowed and unshadowed features of a puppet in 3-space. A fascinating variety of embellishments has been put into these systems by CI's ingenious engineers. Coloration of the final video signal is added by gating color levels under control of the brightness signal, permitting pictures with several grey-levels to be transformed to up to four rainbow hues. Separate shapes described by the space trace may be independently moved and jointed at the same time: Harrison pointedly calls such separate shapes "bones." Darkening at the backside of a spun shape, or brightening at edges of a painted portion, and brightening in proportion to curl, are all strange capabilities of this machine. Lip- synchronized mouthlike motion can be imparted to any part of the shape spun by the space trace (whether or not a mouth is painted on it), by an audio detector feeding directly to the circuitry from a live mike. And the limbs of CI's ghostly figures can be made to swing by connection of sensors to the animators themselves--in a living pantograph. SCANIMATE is a popular device now widely used (at CI's studios) for the making of TV commmericals and station-break emblems. This is their simplest system, used for the conversion and discombobulation of flat artwork. In Scanimate, the space trace is controlled by hand-operated potentiometers. Two separate oscillator settings are available, so that the space trace can have two separate oscillation patterns, spinning out two entirely different virtual shapes in 3-space. A hand- throttle eases from one oscillator setting to the other. This permits an image to be moved, shrunk or enlarged, or flipped; to go from whirling around to a sort of hula; and many more effects. The picture painted on it may be seen to roll on invisible spindles, bloom into fountains, or undulate as pennants--all by modulating the brightness of the flying spot as it traces its unseen shape. This shape, in turn, can move between its two forms under control of the throttle. Animac I (usually called Animac) provides greater flexibility in controlling the space trace. The system's oscillations are controlled by an input vidicon, which artists may quickly modify with pastel check at the pickup. Ghostly tubular lettering, swarming pendulum-patterns and jiggling filigrees are among the possible doodles. CAESAR, their newest system, is oriented toward Yogi Bear-type animation. The artist's cartoons are automatically superimposed on a background or each other. They may be moved, and made to wiggle under real-time control by the user. But is is to Animac II that these curiosities lead. What Harrison calls the "Snow White Capability" of Animac II will permit the sculpture of full humanoid puppets, with perhaps thirty articulated "bones," opaque to one another and casting shadows, colored, moving and talking. OOO)Two young fellas in a Manhattan loft, Messrs. Rutt and Etra, are offering a machine similar to Scanimate but much cheaper. *********pictures************* (Caption); It's not as finely detailed-- the inner screen runs at 525 lines rather than 700--but it costs some $15,000 instead of $150,000.