Looking Back: On Shooting Miniatures for Science Fiction Movies

Alex Funke, ASC, Academy Award-winning visual-effects cinematographer |

King Kong Minatures Set Unit

Alex working on King Kong

I’ve been shooting movies for almost fifty years, and shooting miniatures for over 30 years. I thought it would be worthwhile to take stock of what we’ve learned, and the principles we’ve followed. Incidentally, that’s not the “editorial we.” One of the most potent things about shooting miniatures is that it is absolutely the creation of a group. In effect, the hearts and the skills of many people – perhaps twenty or more – are focused upon a patch of real estate only a couple of metres square.

It is this collaborative aspect of making and shooting miniatures which lends to the result a power and vitality which is extremely difficult to create from within the computer, where only a few souls are driving the creation of a shot.

Today, Moviemaking is at a crossroads. There is a strong drift toward abandoning existing knowledge and proven methods in favour of the new and the sparkly. The notion that a digital synthesis can replace the ‘real’ under camera is attractively seductive.

Historical side-trip: a connection with the theme of perfection and detail. The Romans, and the illuminators of manuscripts in the Middle Ages, had a favourite pigment; it was the lovely orange-red of lead trisulfide, and it was called minium.This colour played so much a part in the illuminators’ craft that “to miniate” became a catchword for artists working on the tiny detailed paintings at the pageheads of those manuscripts. And by linguistic osmosis, over time, the “miniature” came to be anything that was created in a very small space with great craft and superb detail. The classic miniature of the 18th century was the tiny, almost photographic image of one’s beloved at a scale to fit inside a locket.

Thus, tiny perfection. That might be a fine definition for miniatures in their movie application. We have three possible goals when a scene is created via miniature photography:

• To create something that doesn’t exist, and has never existed;
• To extend what does exist, and modify it for the story’s sake;
• Or to make possible the difficult, the expensive, the time-consuming.

We can see how the spacecraft Rodger Young (Starship Troopers), the major part of wicked King Miraz’ castle (Prince Caspian), and the sunken submarine USS Montana (The Abyss) fit within these three categories. Humankind has never built a combat space corvette – probably never will. Building only the lower six metres of a full size castle set in Prague and extending and creating its towering pinnacles and spires with a miniature matched to the live plate is an efficient (and cost-effective) way to create a grand castle. And to set up to shoot a full sized nuclear submarine, underwater, would be outside even a James Cameron budget.

Let’s consider the central issue in using miniatures. It is that the miniature scene must never draw attention to itself. It has to be absolutely of a piece with the photography of the real. And to do this, it must resonate with the audiences’ perception of what ‘the real’ looks like.

The worst remark that a miniature builder or cameraman can hear is:  “… wow, that was a nice miniature shot …” Oh, the horror! That is failure! That means that the viewer was pulled out of the story by the miniature effect, instead of being drawn into it.

Utterly damning! Because as soon as the shot is perceived for what it is rather than what it represents, then the audience has been jerked right out of the story and is lost forever.

What we must do: even though the scene has nothing to compare it to, nothing the viewer has ever seen before, he has to believe that it is consistent with his perception of the world he knows. In fact, our watchword is “do your work, and then erase your tracks.”

Miniatures; small replicas of large objects. Some are really small; a six-metre version of a space ship that is supposed to be 550 metres long. And some are pretty gigantic; even though the sets of the Mines of Moria  (The Fellowship of the Rings) were 1/14th of full size, their footprint was over 25 metres square and they touched the stage ceiling at 8 metres! It’s not about how big the miniature is, but how well it represents the real.

Incidentally, one of the driving concerns in choosing a miniature scale is: how big can we make it and still get it through the stage door?

I remember shooting my very first miniature shot, in 1979. It was a flyby of the ‘big model’ of the Galactica, and we were kicking off the first season of that ageless show, Battlestar Galactica. We had just moved into the spanking new Universal Hartland VFX facility (Universal for the Studios, Hartland for the street we were on) with network air dates staring us in the face. We were moving equipment and miniatures in at one end of the giant warehouse, and starting to shoot (with motion control rigs that were still being built) at the other end.

When I set up for this first shot (my first encounter with movie miniatures, remember), I looked closely at the surface of that large miniature ship. And I saw that the hull was literally packed with almost microscopic detail. The closer I came, the more detail I saw. How was this attained?

We got the show on the air. Now, I mention this bit of ancient history to focus on this: the miniature of the Galactica was about four metres long. How do you make a four-metre ship look as if it’s 610 metres long?

The key point is this: the audience has never seen a 550-metre-long spaceship. They have no idea what one looks like. But, the viewer does have an ingrained sense of what the world looks like. It is the task of the modeller and the cameraman to find the cues that speak to this perception, and help the viewer to willingly accept the reality.

The model builders of that era were a new and young generation; they were not going to build models the way they were built in 1950. Behold: the Nurney! The name, and the technique, both date back to the late Sixties, and to the miniatures in the ground-breaking film 2001.

Nurneys are the tiny plastic parts that come, in their hundreds, in every high-end plastic model kit, whether of a battleship or a tank or an aeroplane. They are the tiny, non-descript but very highly detailed little shapes that are intended to become the anchor winches of the battleship Yamato (this kit is still a favourite as a Nurney source!) or the tracks of the Sherman tank, or the tail turret of the Lancaster bomber. The discovery of this new generation of modellers was that by encrusting the miniature with layers of Nurneys, in rhythmical patterns, an effect was created that was almost fractal in its appearance.

It is very hard to make a surface that has the essence of presence, especially if it represents something huge in the real world. This is because we see objects in layers of detail. If you look at a rubbish truck at 50 metres, the surface you see will be totally different than at 5 metres, which is again different from the close view at 1 metre. We naturally understand that the real world has almost infinite layers of detail and texture. Our goal as miniature builders and shooters is to deliver the feeling that there is endless detail, and that if you got close enough, you would see more and yet more layers of texture.

The busy and tiny detail of the Nurneys provides part of that complexity, artificially. So does the simple dodge of spray painting the model with a very dry spray, so the paint instead of lying down and creating a smooth film, actually forms infinitely tiny pebbles on the surface; another layer of texture!

In addition to the generation of texture through the Nurneys and the pebbly spray, the issue of surface deposits (read: rust, moss, gravel, soot, grease, dung, slime, peeling paint, and so on) is critical. There are very few surfaces anywhere in the real world that are free of some overlaid surface deposits. And it may not be very obvious, but even a pristine marble hall in a grand State House miniature needs to have just the least trace of wax smears from the floor polisher, soot stains near the ceiling, and greasy rubs near the door knobs. This is part of locking the image into the perceived real world.

Looking at the 1/72 scale miniature of Minas Tirith in The Return of the King, speaking to a model painter:

“… What are those yellow and brown streaks down that stone wall?”

“… Oh, the royal stables are right above there, and that’s where the horse pee washes down the wall.”

What are some other cues to realism and scale?

• Cues of shadow and the specular nature of the light;
• Cues of movement;
• Cues of focus;
• And cues of camera position.

Consider our Sun. By the time sunlight arrives here on Earth, his light rays are very nearly parallel, and this means that shadows will be cast with sharp edges. When are they soft-edged? When the light has passed through natural diffusion, the haze and mist of the air.

It is essential that the character of the light on a miniature have qualities that are correct for the setting, and the set itself. Consider that a ship in deep space is illuminated by a small point source, the Sun. But if the cameraman uses a big light, a Ten-K, with a lens that is a metre in diameter, the quality of the light is no longer a point source, but rather a large and rather diffuse one!

Earthly light is a combination of the hard sun and the dispersed sky light. For many big landscape miniatures, I evolved a method called Big Sun Country. Instead of casting sunlight with one large source, a large number of smaller lights, spread over the area of the ‘sky’ and all pointing in parallel, builds up a much more realistic synthesis of real sunlight. But it doesn’t work on spaceships: it is far too broad for that!

Movement of the camera or the model should be locked into real physics. Once the camera is moved at a speed which, scaled up, would be impossible on any kind of camera mount, the shot has been compromised. This is often called ‘the video game effect,’ and appears all too often in effects done with computer animation; behold here this useful adage: “Just because you can, doesn’t mean that you should.” Don’t kick the viewer out of the story!

The same is true of model motion. The human visual system is very good at interpreting mass and momentum; it is a survival trait that kept our ancestors from falling into crevasses or being squashed by falling trees. So when, sitting in the dark, the viewer sees an object of huge mass moving with a velocity that feels wrong, then the mood of the scene is lost. And it only takes a feeling; the viewer doesn’t need to quantify what’s wrong, but the damage is done.

Shrinking Starship Troopers

Shrinking Starship Troopers

You wouldn’t think that a small space craft lifting off would be hard to program. The Starship Troopers dropship  (Carmen in the pilot’s seat), which rescues the infantry from Planet P, simply had to blast off from the metal deck of the fort and fly away. But this shot was the very devil to program. This is because the ship was very small, and proportionally should be light -yet before this lift off shot would work we had to treat the ship as if it weighed tons. This is all related to creating a believable acceleration curve as she lifts; it can’t behave like a kite or Gypsy Moth biplane.   

Shooting full-sized objects in the full-size world, the chances are that everything will be in sharp focus. That’s because even a very close object is still perhaps 2 metres away. But when you scale that two metres down to 1/24th scale, it’s only 80-plus millimetres away- and that is very hard to keep in focus.

Bad focus is the bane of miniature scenes. Everything must be sharp; softness is an absolute giveaway. Once again this is a negative cue that will disrupt the viewer’s acceptance of what he sees.

Now, the creation of perspective is based upon only one thing; the distance of the camera’s eye (and hence the viewer’s). One will notice that camera angles are almost invariably low, at least when shooting miniatures representing things that sit on the ground. The camera is the eye of the audience, and if the set is 1/24th scale then the eye should be eye height / 24 – which is about 70 millimetres. But the effect is more than just getting the viewer’s height fixed in space. The low angle helps give a ‘looming’ property to the object. This is part of our perception, and can’t be explained mathematically, but big objects take on grander scale if some part is close to the eye and the rest towers above – though the ‘towering’ might be a deep chasm instead of a tall cliff!

Let’s look at a couple of examples of miniature setups. In the film Waterworld (1995), a key plot point is that the Mariner character owns a handful of soil, something that has not been seen on the ocean’s surface for generations. Using some rather preposterous equipment, he takes his girlfriend to the bottom of the sea, to reveal the ruins of the drowned city of Denver.

Convincing Light in Waterworld

Convincing Light in Waterworld

Now, what are the requirements of this scene? The city has to be convincing as a real city that just happens to be submerged. It has to have the visual cues that make the viewer feel that, yes, this is just what I imagine a sunken city might look like. The light has to be convincing, and the presence of the water has to be displayed perfectly. Remember, we are showing the audience an environment that they could never have experienced, but they have to accept it implicitly without worrying about why it is the way it is.   

First requirement, a set large enough to work around with the camera. Making a scaled down model of even a small part of high-rise downtown Denver means that the scale has to be small; in fact, the scale was about 1/100th of full size; 1 centimetre is one metre.

A kind of non-descript mossy weedy stuff grows on and festoons sunken cities. At least, that is our visual brief; it has to look right, not necessarily be right. We made the set sticky, and then puffed fine clothes dryer lint onto it. Clever modellers! And then the set got endless coats of extremely thin watery paints which are misted on and then washed off with alcohol, misted on and washed off – that’s how the complexity of colour texture is created.

When Denver was drowned, the populace fled the land and moved onto the rooftops. Cities of shacks grew atop every building. All modellers love to build shacks, because they are the ultimate exercise in texture – broken boards, flattened petrol tins and corrugated iron, rusty pipes, flattened oil drums. So naturally a lot of urban shack landscapes were built to dress the Denver skyline.

In reality the great depths are utterly black. But an absence of light is very hard to photograph! But the audience will accept that the light seems filtered from a great distance.  This means that it must be extremely soft, and from a very broad source; as indeed it was.  The sole illumination was from an enormous piece of calico stretched over the whole set and then lit with very soft broad blue sources, bounced into this overhead from below. No hard shadows in undersea Denver: the water is defined by how it softens the light.

Finally, the entire set is filmed in dense oil-based smoke. This is essential to create aerial perspective (whether it’s in the air or the murky water).  Today a new digital post-production technique called volumetric diffusion allows the effect of greater diffusion at greater distance to be created in a scene that has originally been shot with no diffusion at all.

How might we use miniatures to extend a live action plate? Think about Starship Troopers. In the depths of Hell’s Half Acre in Wyoming (a big dusty ten-mile-square hole in the ground), standing in for the planet Klendathu, the Mobile Infantry from the starship Rodger Young are landing in their dropships. To set the scene for this climactic landing, just the doors of the dropship and the landing ramp were built full size, so the troopers could pour out screaming weird cries. But the sets were just built static on the dusty ground.

The director (Paul Verhoeven) wanted a continuous shot of the ships blasting in and landing and the doors opening; no cuts!

We had a very carefully constructed dropship in about 1/24th scale, which had very close geometry to the full size partial build in the plate. In 1996 our tools for lining things up to exactly match were still pretty primitive (the young digital artists of today don’t realise just how lucky they are!).

Our primitive (but successful, though tedious) method was to cut the film of the miniature with scissors and lay it over the plate of the live set under a microscope; and to do this again and again, adjusting camera position until the size, angle, and geometry matched. That was the ‘landed position.’ Then we could make a move to fly the entire miniature up into the sky.  Reverse the shot, lay some dust over the composite at the moment of landing (that’s when the transition is made from complete miniature to miniature-plus-full-size set), and the shot is complete.

Cabs in The Abyss

Cabs in The Abyss

Think of using the proper tool for the task, and consider the following; a key moment in The Abyss. The crew of the undersea Deepcore oil rig have been drafted by the US Navy to find the sunken USS Montana submarine. A wide shot requires that we see the entire sub, the divers at work, and the ‘Cabs’ which are the submersibles used to drive around at depth.  

This particular shot is a combination of three techniques, each used where it is most effective. In the foreground is a live action plate, shot underwater on the small section of the full sized submarine that was built in a tank in North Carolina. The divers are at work, and the one full-sized Cab is hovering. In the deep background is a shot of the miniature Montana (about 1/48th scale) with three miniature Cabs working around it. And in the middle, the two components are blended by a matte painting; in this shot, this is a real painting, done with paint and brushes on hardboard by the great Bob Scifo; no digital matte painting in 1997!

You see here how the VFX team has used each process to best advantage. The foreground plate gives us the live divers in action; and since there was only one Cab, the cluster of Cabs in the distance is handily provided by the miniature (filmed in incredibly dense smoke).

There are issues with scaling the real world that make some types of shot very difficult. Probably the worst miniature shooting you will see is in wartime films (and some even quite recent films) about naval ships. The motion of those racing destroyers, and the oily looking water, the giant drops and splashes, are horrid to see. Why is this?

It is possible to scale the size of a real object, the destroyer. But the slice of the world, which has dynamic properties or viscosity, does not scale. Among those unscalables are the viscosity of water, the motion of snow and flame and dust, and the behaviour of tree branches.

To create miniature water that would shoot realistically, the size of the ship and the viscosity of the water would both have to be scaled. Otherwise, the destroyer is basically floating in treacle. And this viscosity can’t be overcome; although the suggestion of using petrol or kerosene instead of water has been suggested (but not very seriously).

Even in the flooding of Isengard at the end of The Two Towers, which was filmed in a huge 1/6th scale, the raging water looks stiff and sluggish. Those scenes were produced with a 200,000-litre water dump tank. But 1/6th scale is right at the edge of where water just begins to behave in miniature; and usually, if there were a chance, the miniaturist would always go to 1/4 or 1/2 scale. In fact, the foaming and cresting waves had to be added by digital animation to the leading edge of the Isengard water, specifically to cover up the ‘stickiness’ of the water.

In shooting King Kong, the Miniatures Unit was given the brief by Peter Jackson that we were to shoot the miniature sets with 1/10th scale trees with real leaves, and with the leaves and branches moving in the breeze. Now, red beech tree leaves look almost perfect at 1/10th scale; but although the branches are ten times smaller, the stiffness of the wood is just as great as if the tree were full size. How to get realistic motion in these miniature branches?

The solution: cut the branches into multiple sections and then join the sections with tiny linear springs. These were composed of thin, twisted music wire strands, which made it possible to ‘tune’ the stiffness. It was possible to make a linear spring that had the true scale stiffness of the branch: just stiff enough to hold the weight of the scale branch, flexible enough to move with the slightest touch.

Looking back, it’s funny that so much of what has been done to create these scenes in miniature, which enhance the story and expand the scope of movies, are just done with such simple methods. Simple, but applied by dedicated hands and willing hearts. It has been a joy to be part of that world.


Bio: Born in Santa Barbara, CA. A fourth generation Californian, Alex has been a cinematographer for over 30 years. He studied biochemistry for several years in college, but in the early 1960’s he switched to UCLA Film School. He subsequently worked with Stephen Burum, ASC. He spent eleven years with The Office of Charles and Ray Eames (shooting over 30 films at that time). His first effects project was Battlestar Galactica, at first with Apogee and then at Universal-Hartland. He founded Precision Film Group in the late 1970’s, doing effects for TV and features. Later, at Las Palmas Production, he co-designed the first-ever system for pin-registered transferring of film to video. Although he is a specialist in miniatures and visual effects, he also shoots live action of all kinds. He has taught advanced cinematography courses at both UCLA (full-time faculty for two years) and Loyola University. He has been invited to give lectures on visual effects photography in Italy and Norway.