3D in movies : How it works

Everything in the world that has eyes has at least two of them, so clearly nature thinks that stereo vision is worth the investment. You live in a 3D world – you see in stereo all the time: what makes it so amazing in the movie theatre is the novelty of seeing 3Ds where you are used to seeing 2.

There is no practical way of presenting stereo in a theatre without some kind of glasses because what you call ’3D’ or ‘stereo’ is really something that is happening inside your head.


Stereo is that aspect of 3 dimensionality that we get when we match up inside our head two slightly different points of view of the same scene – the two points of view being separated by about two and a half inches – (75mm) – the distance between our eyes. Those two points of view – like the beams from two flashlights – converge on the object we are looking at. Our eye muscles tell our brain how much we have converged our eyeballs so then our brain is able to calculate how far away the object is and thus how big it is. Obviously this stereo effect only works for a limited distance. Once objects are a long way off our eyes remain effectively parallel, there is no convergence and no ‘stereo’ effect. We then depend on other clues – occlusion, size, texture detail, haziness etc.

A whole long list of qualities give us 3D clues telling us about the relationship of objects to each other and their size. Stereo vision – each of your eyes sending a slightly different picture to your brain – is just one of those clues and it is not even the most important one.

The strongest is occlusion: If a person passes between you and a car the person ‘occludes’ (prevents you from seeing) the car and you know that the person is closer . If the car occludes the person then the car is closer.

The size of the image on the retina is a strong clue, if (as is nearly always the case) the object is familiar. If the image of car ‘a’ on the retina is larger than the image of car ‘b’ then we know car ‘a’ is closer.

Some of the clues are more subtle but nevertheless very powerful: The apparent texture of the surface of the object, the shading and shadows, the amount of haziness in the air between viewer and object give us important clues to size and distance.

Antarctica offers a compelling lesson on the strictly limited role of bi-occularity in our perception of 3D. On that continent many of the clues we normally rely upon are missing. There are few objects of familiar size (no trees, no roads), few familiar textures (no grass, no bricks), quite often there are no shadows and there is usually a complete absence of haziness in the Antarctic atmosphere: Explorers with two perfectly good eyes have perceived big objects as small, small objects as big, far objects as close – at the cost of their lives.

Really high resolution – where the recorded image approaches the clarity of the image that our eyes give us of reality – is an immensely powerful 3D cue. Most people come out of high definition 2D presentations such as Showscan (70mm film shot and projected at 60 frames per second – instead of the usual 24 fps) and IMAX®HD (70mm shot and projected at 48fps) believing that they have been having a 3D experience – and they are more than half right.

One of the nice things about 3D is that it makes you appreciate 2D. “2D” is not really 2 dimensional – you don’t need two eyes to play tennis and you don’t need “3D” to be able to perceive that the actors, props and locations in a movie are not flat.

Computer animators have unabashedly appropriated the designation “3D” to mean animations that feature all of the 3D cues except bi-ocularity. When they talk about offering a different image to each eye they talk about “stereo”.

When we shoot stereo movies we actually use two cameras squashed together so that the two lenses are as close together as two eyes. We record two rolls of film simultaneously – one for each eye – then in the movie theatre we use two projectors (or one projector with two lenses) to screen both rolls of film simultaneously. If you take you glasses off you will see both images superimposed on top of each other on the screen. It looks a mess. The glasses sort this out by only allowing the right eye to see the roll of film shot by the right camera and the left eye only to see the roll of film shot by the left camera.

This used to be achieved by colouring one roll of film blue-green and the other one red and by giving the viewers cardboard glasses with one blue-green lens and one red lens. Through the red lens you could only see the red tinted image and through the blue-green lens only the blue-green tinted image – thus each eye got to see the appropriate roll of film. This doesn’t work very well mainly because the red filter is not very good at blocking out the blue-green light so the ‘red’ eye tends to get two images at once!

Better separation of the two images can be achieved by using filters to polarize the light carrying the image as it comes out of each projector lens. The left eye image is polarized vertically – the right eye image is polarized horizontally – the left eye of the glasses is also polarized vertically so the image that was captured by the left side camera and projected by the left side projector lens can only be seen by the left eye of the viewer. The converse is true for the right side image.

The story is further complicated in some theatres which use goggles with polarized lenses as described above but also include a liquid crystal layer in each lens that works as a shutter – opening and closing each lens alternately hundreds of times a second so that the left eye and right eye images are separated by time as well as space.

The stereo effect exists only inside a pyramid of vision with its apex at the eyes of the viewer and its base at the edges of the screen. The restrictions of this pyramid make it impossible to place (for instance) a full length actor two or three feet ( one metre) from the viewer. At that proximity an object of the size of a person just doesn’t fit inside the pyramid – and as soon as anything ‘breaks the pyramid’ it falls back inside the proscenium and appears to the viewer to be as far away as the theatre screen – no matter what you do (incidentally proving again that occlusion is more powerful than stereo as a 3D cue).

It will be obvious that the bigger 3D screens are the bigger the pyramid is and the more room you have to make the effect work. With a conventional small cinema screen the stereo pyramid has such a narrow base that the 3D effect is confined to thrusting spears and paddle balls.

If you look at the screen without glasses you may notice that the projector beams cross in the theatre: You will see that the left eye image is projected to the right of the screen and the right eye image is projected to the left of the screen – so your eyes are actually crossed as you watch the movie. The point in space where your eyes cross is where the object seems to be. When the object seems to get very close what is actually happening is that the images are being projected further and further apart so that your eyes are crossing more and more and your eye muscles are telling your brain that whatever you are looking at must be close! If handled clumsily by the film maker this can give you the headaches that 3D is famous for – but if done well its a lot of fun.

The Gulliver Effect

I mentioned earlier that if the two camera lenses are placed the same distance apart as our eyes then objects we photograph look “normal’ size. If we move the lenses further apart it is as though our eyes have moved further apart – as though we are bigger and the object correspondingly smaller. I pushed this to an extreme in “Imagine“.
By mounting the cameras more than two metres apart I effectively increased the size of the viewer by about 33 times and thus shrank what we were looking at by 33 times…..so the people in the scene look less than 50mm – 2 inches – high. To get the right sound for their speedboat we used an actual recording of a mosquito…..

John Weiley. Byron Bay.