“Pulling a key”

This phrase refers to the process of ‘Keying’ which is essentially using colour or tone to identify the difference between the foreground object or character and the background in an image.

Green Screen

Why green screen? And how does it work? Why green? 
Because people aren’t green. The reason to shoot anyone (or anything) in front of a green screen is to digitally separate your subject from the background, ready to be transplanted into another shot or scene. The colour green is used because in colour science it is the most distinct from human skin tones, no matter what your ethnicity. Green is far away from red in a colour wheel. We’re all made of red tones! If we chose red or yellow screen, for instance, we’d have a harder job getting a clear separation between the person standing in front and the screen itself.

Why have I seen a blue screen being used then?
Blue is next to green in the spectrum, so is also far away from skin tones. Blue was a lot more common when keys had to be created in film, before digital technology, because of the nature of the separate colour emulsions within celluloid. When digital processes came in, green tended to be favoured because of its higher luminance- it needed less lighting. Some digital cameras also record more information in the green channel via their sensors, so providing a smoother (and therefore easier to key) picture.

There’s nothing particularly wrong with blue, although if you are adept at Adobe Photoshop and can view individual Channels, you’ll know that blue is often a bit ‘noisier’ or grainy, and so this is reckoned to hinder more delicate keys .

I want to key Kermit the frog; or a potted plant, what are my options?

If you want to key something green, you can actually use any other colour. In some films red screen has been used on models, like spaceships (because they were lit rather blue for outer space conditions, so blue and green weren’t good ideas).

Can I key on Black or White backgrounds?

Yes, it’s called Luma Key (Luma = Brightness). The problem is this tends to be inexact: for instance if you were filming an actor against White, the eyes would also be keyed, and the teeth, and maybe the highlight on the forehead. It wouldn’t be convincing. However it can work fine for white text on a black background (or vice versa), or creating an experimental effect. Many stock effects are shot against black, as you saw in week 1, ready to be luma keyed.

Why learn keying anyway?

Here’s the thing that most people don’t realise at first; when we pull a key, through whatever method, the computer is really behind the scenes creating a Matte. It’s a term from the old days of film, equating to a mask. It’s a record of which parts of the image are see through, and which aren’t. This is often represented in the computer as a monochrome version of the image. The whitest areas are telling the computer which pixels in the image are opaque, and the blackest areas are those areas that are transparent- where the other image will come through. A range of grey pixels designate translucent areas. You’ll see this matte in the KEYING PROJECT.

This matte can be saved as part of the image, like saving an instruction to any VFX software about how the image is to be cut out and presented later. In VFX circles, you may hear this saved Matte referred to as the Alpha Channel. Put simplistically, many digital images can have four channels or separate sub-images inside, namely a Red, Green, Blue, and Alpha channel. The Red Green and Blue channels contain information on the colour of individual pixels in your image, whilst the Alpha Channel only contains information on the opacity or transparency of each pixel. Generally, you don’t see the Alpha Channel, it’s hidden away and is only viewable in certain software. Some file formats aren’t designed to contain Alpha Channels, like JPEGs, so if you want to save this keyed information into your image, you need to choose a TIFF, TARGA or EXR format, or others.

Sometimes, keying isn’t possible: welcome to the art of Rotoscoping

Rotoscoping, which as you’ll see (it’s regularly used by at least one of our guerrilla filmmaker guests), is used when you can’t pull a good key. It involves masking your foreground character frame-by-frame by drawing around it. You are essentially drawing a matte or mask, rather than getting one instantly when you pull a key. Although today’s rotoscoping software has a level of automation involved, rotoscoping is very time consuming, a bit like drawing frames to be animated. Most VFX artists will use a mix of keying and rotoscoping to ‘cut out’ their chosen foreground figure.

The big myth?

Actually, keying can be hard work. Getting the first 70% of a shot right is easy, it’s the last 30% that needs the hard work, and that’s often the convincing bit. There are always edge issues, and also what is called‘spill’- the phenomenon where green light from the screen bounces back onto our subject, especially if they are wearing white. Because the way your green screen is shot is such a major factor in the success of your VFX, it’s increasingly important that you pay attention to the shoot. Big VFX companies like MPC make sure they are always present when things are shot by the big studios, because a well lit green screen will save so much time later.

A short history of keying

Although we now live in a digital world where keying is concerned, the desire to transplant images from one film into another is as old as cinema. Well, actually older, because cinema only developed as a phenomena years after film was being used in fairgrounds and magic shows.

The image you see above is from one of Georges Méliès films. Méliès is the ‘godfather’ of VFX; in fact one of his most famous images, that of a rocket stuck into the eye of the moon- is the symbol of today’s Visual Effects Society, the organisation that looks after VFX worldwide. Méliès is held in high esteem to this day.

You could say it all started in 1888 when George Eastman developed sensitized celluloid that you could expose an image onto. The first successful system for filming and projecting moving pictures came in 1895 when the brothers Auguste and Louis Lumière exhibited the Cinématographe in Paris. They were businessmen not artists, and they thought the Cinématographe would be a passing craze. They made short mundane films by today’s standards - a woman combing her hair, fireman extinguishing a fire, a baby being fed. The idea of projecting a story came later, but there was a restriction on the duration of early films so this wasn’t possible at first.

Méliès was an enterprising theatrical magician who saw the potential of this new development for his stage act. When he couldn’t buy a system from the Lumière brothers, he hired a couple of engineers to build his own. It’s reckoned Méliès made over 500 films in his lifetime, but only a handful survive.

There were two big breakthroughs in the early days of film. One was thesubstitution cut where the camera was stopped and an object in the scene replaced or removed. This amazed early audiences, and Méliès discovered it when his hand-cranked camera broke down whilst he was filming. There was a gap of a minute in the street scene he was filming “I suddenly saw an omnibus changed into a hearse and men into women” he exclaimed. You can see an early example of this substitution cut if you search for Alfred Clarke’s “Execution of Mary Queen of Scots” (1895) on YouTube. At the time of course, this was all new and people were amazed.

Discovery of the matte

The second effect, and the one that concerns us most, was the matte used with a double exposure. It was discovered that if you exposed the film (by shooting a scene) and then wound the film back into the camera, you could expose a new image over the top. Filmmakers like Méliès saw that if you left parts of the scene dark in the first exposure, then you could align some action to happen in that space on the second exposure. The India Rubber Head (1901) was a case in point.

From the still at the head of this article you can get the idea. An alchemist/scientist places a version of his own head on a table and begins to inflate it with bellows. Conveniently there is a black shadowy archway where this double exposure happens. The head ‘expands’ by having the camera move closer in the second exposure. Filmmakers soon started experimenting with Matte boxes- simple bits of card over the lens that stopped portions of the film being exposed to light the first time, and then inverted this on the second exposure.

Travelling Mattes

As film processes developed, and film became more sophisticated, the weaknesses of simple mattes became apparent. It would only work if the camera and actors were still or stayed in a fixed area. A travelling matteallowed a freely moving person to be cut out and placed on top of a background without winding film back through a camera for double exposure. To achieve this, just like today’s green screen, an actor would perform against a contrasting screen. In film terms this meant creating a number of differently developed prints from the original, starting with a very high contrast film of our foreground character (let’s say, dancing around). Then a high contrast negative of this would be created too. Put simply, these two high contrast (black and white) filmstrips of our dancer would be used to create masks or mattes for both the dancer and the background. This film technique was like having a magic moving piece of card that ensures only the dancer is exposed on one film, and that the background has a perfectly synchronised moving black shadow where the dancer should be. These two films are optically printed (projected and refilmed) and combined together and, hey presto, our dancer seems to be dancing on the moon or whatever the chosen background image was! This method was prevalent from the 1920s till the 1960s.

Colour Separation processes

As film grew in commercial importance, there were many variants of this process created; but essentially it’s the antecedent of what happens today digitally. When colour film started gaining ground (around 1918-1936) clever new methods to separate foreground and different background were invented. One particular method was the Blue Screen Colour Difference process. Invented by Petro Vlahos it became the basis for the most widely used methods of getting a travelling matte until digital techniques were possible in the 80s. The process demanded a very pure blue backlit background. The film negative produced was copied onto back and white filmstock through a blue filter, which allows only red and green to expose the film, so the blue area is transparent. However there is always likely to be a bit of blue on the foreground character so this black and white film was combined with the original negative to make a new high contrast matte, but this time through a special green filter. By now the blue screen is clear and the foreground character totally black. This ‘male matte’ is then copied to a negative ‘female matte’. There are several other stages to this - each one with other film prints. Aren’t you glad we now do this digitally?

The advantage of this system was detail - around hair for instance, and it recorded semi-transparent objects like a glass of water or reflections on a window very well. The system was even used in Superman (1978) -although Superman’s costume was blue, they were able to pull a decent key through this method.

Now it’s all digital

In the digital age we still use some of these tricks to pull a decent key, although it all happens now at the tweak of a mouse, we don’t have to wait overnight for film to be processed through chemicals. However, we sometimes still need to look at the individual Red, Green or Blue channels of a foreground image to make a matte work. It can still be laborious to get your key as convincing as you want it to be.

YouTube is a great place to look for early cinema firsts, like Georges Méliès or Petro Vlahos’ work. It’s been said that many of the first pioneers of cinematography shared similar characteristics to today’s guerrilla filmmakers; the ability to experiment, make do with what they found around them, and inventiveness. Just like today, the field was wide open for newcomers to create things never seen before. As an example have a look at this film about the first colour film ever invented. If you find any interesting clips about any of the subjects in this article, why don’t you share them with all of us.

Setting up a green screen, guerrilla-style

Note how Scott is particularly keen to control the light of the space, and even uses separate lights for the foreground and background

Preparation for shooting a green screen is always important, it’s really not worth rushing because you’ll pay for it later with more time in front of your computer screen, tweaking at pixels, and cursing whoever shot your footage!



There’ll come a time when you just can’t key an object. Either the background is just not distinct enough from the foreground, or you need to key something out of a shot that actually hasn’t got a plain background at all. In cases like this you need to resort to Rotoscoping, or ‘Roto’ as it’s often called.

However, most of the times you’ll use roto as a technique is as an additional assistance to your key. Roto’ing creates a mask by drawing shapes onto a layer. If you think about first principles, what you are doing when you pull a key is making a matte. Now you may never see that matte, but it’s there. You'll see it briefly in the Car Project - the black and white image that shows which areas of the driver and car were white (opaque) and which parts of the image let the background through (the window). So drawing roto shapes does the same thing, except you are drawing a matte by tracing over something in your shot.

Conceptually, think about having a pair of scissors and a book of a film of someone dancing from a musical. Let’s say they are shot against an ugly theatrical background which you wish wasn’t there. Now imagine each picture in this book is an individual frame from the original movie. With patience and all the time in the world, you could go through the book, cutting out individual frames with your scissors. Then, you could take all those cut outs of the dancer and stick them under a rostrum camera in order, and re-shoot them onto another more pleasing background image. It sounds like an ordeal, but it can be fun, and luckily it’s a bit more automated these days than we’ve made it sound, as you’ll see.

Roto History

To understand Roto, let’s go back in time. In 1917, animator Max Fleischer (The Fleischer Brothers were famous in the pre-Disney era with work like Betty Boop) developed the process as an aid to the laborious activity of drawing cartoons. Max had a rostrum camera pointing downwards that was adapted with the addition of a lamp so it could also be a projector. Then, frame by frame, a previously shot film was projected onto the table. Let’s say it’s the dancing person we’ve just been talking about. Paper is placed underneath and the dancing figure is traced, one frame at a time. Thanks to peg bars on the table, as each consecutive frame is traced onto a separate sheet, the images are all registered in the same place. This way, the natural movement of the filmed dancer can be copied, drawn, and re-filmed.

Now Max cleverly got his brother Dave to pose in a clown suit for the creation of some live action reference footage. This gave birth to the first rotoscoped cartoon character “Koko the Clown”. The cartoons were immediately popular. A reviewer for the New York Times wrote in 1919,“After a deluge of pen-and-ink ‘comedies’ in which figures move with mechanical jerks with little or no wit to guide them, it is a treat to watch the smooth action of Mr. Fleischer’s figure”

The still at the head of this article features legendary band leader and singer Cab Calloway performing “Minnie the Moocher” in 1932. You can see the animation here At the start you see footage of Calloway’s idiosyncratic act, and then at various times like 1:52 or 2:25 you’ll see a walrus creature whose movement has been rotoscoped or traced from Calloway’s performance.

This technique was eventually copied by Disney, who famously used it in the animation Snow White. Disney hired a dancer, Marjorie Belcher.Rotoscoping her movements sped up production and gave the animators a realistic reference movement. This tracing of real action soon got adopted elsewhere and took off.

Famous animator Walter Lantz explained “I would take the old Charlie Chaplin films and project them one frame at a time, make a drawing over Chaplin’s action, and flip the drawings to see how he moved. That’s how most of us learned to animate.”

VFX and Roto

Now as VFX matured and film makers started using mattes, rotoscoping became a great way to juxtapose and combine different characters and scenes. You could now draw a matte if you couldn’t key one.

In Return of the Jedi (1983) animated walking vehicles (called AT-ATs or All Terrain Armoured Transport) were matted into real forest scenes. A‘hold out matte’ of some of the trees in the forest of Endor were hand rotoscoped frame by frame to create a matte. These were protected during optical printing, so the AT-ATs appear to walk between trees, adding a sense they were really there, in the midst of the action.

Alfred Hitchcock’s The Birds (1963) needed scores of birds to attack a small town. It wasn’t possible to control the birds of course, or even key them against a background, and the nature of the shots meant the birds and a luma key against a sky wasn’t possible (some were white birds), especially as many were originally shot against cliffs, beach, or sea. So, tracing the birds was the only option. Rotoscoping the 500 frames in one shot of the attack took two artists 3 months to complete. Of course these are pre-digital film examples, but even today roto’ing big shots in feature films can take weeks.
Roto in the digital age

Most VFX based programmes have rotoscoping and masking capabilities these days. Some people prefer dedicated specialist software likeSilhouette or Mocha when they have tricky work to do.

Roto in the digital age is not all about making mattes. The same tools can digitally paint and be used to remove objects from the film- like pylons, boom mics, or the public straying into shot. In the context of this week, roto is a great aid to strengthening keys. Maybe your keyed figure looks great except for one arm- in that case a well-drawn roto of that arm can be added to the matte created by the key, thus saving the key. In one of this week’s Blaine Brothers clips you’ll see Chris using masks to protect and show breath coming from a character’s mouth. That kind of subtlety at that kind of speed was undreamt of by those who had to roto film footage before the digital era. However the same skill of a good eye and attention to detail combined with tons of patience is still needed today. With VFX, if you want good results, you need to put the time in.