How does 2D animation work?

2D animation is a flip-book made big. An animator draws a character again and again, each drawing a tiny bit different. Show the drawings fast, one after another, and your brain sees one moving character. (Why that works, the phi phenomenon, is in FilmClear.)

Cinema shows 24 frames a second. Drawing a new picture for every frame is called animating on ones: 1,440 drawings for one minute. Most hand-drawn films save work by shooting each drawing twice: on twos, 12 drawings a second. It still looks smooth for most actions. Very fast moves, like a whip or a spin, go back on ones. Slow, stiff moves can even go on threes.

Those numbers add up fast. A 90-minute film on twos needs about 65,000 drawings for just one character. Add friends, effects and backgrounds and a feature can need hundreds of thousands. That is why studios split the work, which the next chapter shows.

Nobody draws an action from the first frame to the last in one go. The key animator first draws the key poses: the drawings that tell the story. Here that is the crouch, the top of the jump and the landing. Winsor McCay worked this way on Gertie the Dinosaur in 1914.

Then an in-betweener (or assistant) draws the in-betweens. On the key drawing sits a little ruler called a timing chart. Its tick marks say exactly where each in-between goes. Ticks bunched near a key mean the move slows in to that key or slows out of it. A jump slows at the top, just as gravity does.

To draw a drawing between two others, you must see all three. On a light table the paper is lit from behind, so the sheets underneath shine through. Software copies this as onion skinning: earlier drawings show as red ghosts, later ones as green.

In the 1930s, Disney’s animators kept asking why some scenes felt real and others felt stiff. Two of them, Frank Thomas and Ollie Johnston, wrote the answers down in The Illusion of Life (1981): the 12 principles of animation. Animators still learn them first, for pencil, for software and for 3D.

Every student starts with a bouncing ball. With squash and stretch it looks soft. Slow in and slow out lets it hang at the top of each bounce, as gravity does. Arcs keep it on a curved path. Anticipation is the little crouch before it leaps, and follow-through is the tail that keeps swinging after the ball stops.

Timing tells us about weight. The same toss gives a rubber ball lots of bounces, a beach ball a slow, floaty drift (air drag matters when you are light and big), and a bowling ball one thud and a long roll. Tap the other cards on the board for staging, straight ahead vs pose to pose, secondary action, exaggeration, solid drawing and appeal.

Redrawing a whole scene 12 times a second would be madness. So from the 1910s, studios drew the moving character on a clear plastic sheet called a cel (Earl Hurd’s patent, 1914). The lines were inked on the front and the colours painted on the back. One background painting sat underneath and stayed still, while only the cels changed.

Flat cels have a problem: when the camera slides sideways, everything slides together, like a photo of a photo. Real depth needs parallax: near things pass faster than far things. Disney’s multiplane camera (1937) spread the artwork out on up to seven levels of glass, with the camera on top looking down. Levels near the camera slide across the frame much faster than the background.

Many artists drawing one character must all draw it the same way. The model sheet shows the character from several sides with its rules. A drawing that breaks the rules is off model.

Today almost all 2D animation is made on computers, in two main ways. Frame by frame is still drawing: every drawing is made by hand, with a pen on a tablet, in programs like TVPaint, Krita or Toon Boom Harmony. It can change shapes freely, but costs a drawing every frame or two.

The other way is a rig: the character is cut into pieces (head, body, upper arm, forearm) joined at pivots, like a paper puppet. This is cut-out animation, which is older than computers: Lotte Reiniger made a whole feature from paper cut-outs in 1926. Now software such as Harmony and Adobe Animate does it with vector shapes and bones, and many TV series are made this way.

With a rig you set keyframes: “arm down on frame 1, arm up on frame 12”. The computer draws the in-betweens: a tween. How it spaces them is set by an easing curve. A straight line gives even spacing and a robotic move. An S-shaped curve gives slow in and slow out, the same principle as the timing chart in chapter 2.

For a talking character the voice is recorded first. Then someone does a track reading: listening frame by frame and writing each sound on the exposure sheet (X-sheet, or dope sheet). It has one row per frame, 24 rows for every second, with columns for dialogue, action notes and the drawings to use.

Your mouth makes many sounds with the same shape, so animators use a small mouth chart. Here Aamu has eight shapes, called visemes: M B P (closed), A I (wide open), E, O, U, F V, L, and one for most other consonants. The animator picks the mouth for each frame from the X-sheet. Many put the mouth a frame or two before the sound, because a late mouth looks wrong and an early one does not.

A moving mouth alone looks like a puppet. Acting brings it to life: a head tilt on the important word, eyebrows that lift, a blink in the pause. The best animators study how real people talk, and some film themselves as reference.