by Suzanne Martin
Animation can be thought of as a "trick". The good tricks are the ones that
are not noticed unless they are pointed out. This write-up will describe four
types of tricks: design, implementation, economization, and production [BLIN94]
as well as tricks to animating characters [LASS94].
The basic intent of animation is to communicate something. There are four
types of design tricks: attraction tricks, distraction tricks, timing tricks,
and motion enhancement.
The direction of the viewer's attention to what the designer considers the
important parts of the image is one of the most important aspects of
designing either still or moving images. This is particularly important for
moving images, as they are not on the screen for a long time; the viewer does
not have the choice to study it for an extended length of time. One must get
the viewer looking at the place where the new action will occur before it is
all over. The following are ways of achieving this:
Appearing and Disappearing
Changes on the screen will get the eyes attention. Something changing from
existence to non-existence, or vice versa, is pretty dramatic.
Have something blink before it moves will get the viewer's attention.
Anticipation and Overshoot
Anticipate actions by having the objects backup a bit before moving in a
The See-saw effect
Lead the viewer into some such transformation by "rubbing the transformation
back and forth" a bit
To point up the connection between two things on different parts of the screen,
attract attention to them by shaking them both up and down at the same time.
Tension and Release
Tension is created by making shapes seem unbalanced, nearly tipping over.
Release comes from objects being firmly placed.
Make object(s) pause a little before moving them to their final positions.
This builds up anticipation.
Sometimes it is necessary to distract the viewer from something on the screen.
This could be a glitch in animation rendering or it might be a short-cut the
animator is using to avoid some lengthy or complicated computations. One
example is the old switcheroo; sometimes different models are needed for the
same object at different points in the animation.
These tricks apply to how long you make actions take to occur.
If a physical action happens too quickly to see, simply slow down the time
scale. This is similar to scaling an object to fit on the screen.
When flying in to objects that vary greatly in scale, it is useful to animate
the logarithm of the distance rather than the distance directly.
When to double/single frame
Double framing is the process of rendering every other frame and recording each
rendered frame twice. This makes a motion seem to move faster and is
sometimes used to add liveliness to the scene. Single framing is preferable
to double framing except it takes longer to render.
Having various actions overlap in time makes an animation seem more alive.
It is difficult to convey some motions without exaggerating them in some way.
To give the impression of something falling continuously, put some texture in
the background that scrolls as the object is tracked. Another technique is
adding some random fluctuation to the position on the screen. This gives the
impression of a falling object being tracked by a human cameraman.
To show a ball rolling down an inclined plane, add a simple line to the ball.
This represents a highlight. When the ball is rolled down the plane, the
highlight rotates with it looking just like a mark on the surface of the ball.
The spinning top
This is a problem of spinning a symmetric object. Placing a pair of black
marks near the apex of the top (in a "+" shape) gives the asymmetry needed to
follow the rotation. Another trick is the use of Gouraud shaded highlights.
The number of polygons is small so the irregularities in the image, especially
in the highlights, gives a nice impression of motion.
Many animation systems are based on some sort of keyframing system applied to
a nested transformation scheme. The animator must design the transformation
structure and then specify the values for the transformation (i.e. values for
translation, scale and rotation) and the keyframe numbers for them to have
those values. The following are ways to generate the control files:
Top down design
There are two ways to achieve this. The first way is to animate each keyframe
completely and then proceed to the next keyframe. The second way is to
animate the root level of the transformation tree for all keyframes and then
animate the next highest level for all keyframes, etc. This second method is
the easier to implement. That is, the top level motions of the centers of the
objects are animated first using linear interpolation between the frames.
The entire sequence of events is first layed out with no thought given to how
much time it takes for each event to occur. Once the sequence is set the
keyframe numbers are spread apart to specify timing.
Once the top level motions and time durations are set, the detailed motion of
sub-objects, i.e. the limbs of the character, are added. Then add anticipation
and overshoot to the linear interpolation used for the first approximation.
A common problem is the need to change the structure of the transformation
tree dynamically during animation. A common solution is to have the object
appear in the transformation tree at both positions, and utilize some trick
to make only one copy visible at any time. For example: John gives an apple
to Mary. The apple is at the end of the transformation tree of John's arm.
When Mary holds the apple, the apple should be at the end of Mary's arm in
Mary's transformation tree.
Squash and stretch
This is commonly used to give life to inanimate objects. The problem with
this is that objects are usually modeled centered on their center of mass.
Any squashing requires a scale factor about this center. This makes it
difficult, for example, to keep the bottom of an object on the table before
it jumps off. A better way is to provide positional handles on each side of
the object. It is more intuitive to animate these two locations separately,
typically with similar motions but just displaced in time and position.
Some ways to produce scenes cheaply where a full simulation of the situation
would be too hard or too slow:
Translucent or cloudy objects are hard to animate. To get around this the
following can be used:
Scaling and Fading
Animate the object getting larger while making it progressively more
This effect can be made by using a 3D model of lines radiating randomly from
a center. Then make the transparency of each line interpolate from opaque at
the center to transparent at the endpoints. The resultant shape is then given
a large rotation velocity around a couple of axes This will result in a spray
of lines from the origin that is radically different from the one frame to the
Motion blur must be done to properly portray motion with film or video. This
can be difficult since the rendering system needs to know, the position of each
object in scene as well as the speed, and maybe the acceleration. One way of
achieving this is using speed lines and streaks. Motion can be enhanced by
explicitly drawing lines trailing a moving object.
"Production" refers to the process of rendering the frames and recording them
to tape or film. The following are methods to actually produce the animation.
They are mainly bookkeeping techniques to keep production from getting out of
Start at frame 1000
Start at frame 1000 instead of 0. It makes it easier to add frames at the
beginning if needed. Since the frame number is part of the file name, trying
to insert a frame before 0, i.e. using negative numbers make invalid file names.
A four digit frame number allows getting a sorted list of frame files and
having them come out in order. It avoids frame 10 coming before frame 2.
Skip identical frames
If motion sequences are interspersed with pauses, it is silly to re-render
frames that are identical during a pause. Create an automatic mechanism that
scans the animation control file and detects frames for which all rendering
parameters are identical.
Binary search rendering order
To debug the rendering scene, it is often useful to render several frames
scattered through it. If these are ok, render the rest. It is a shame to
re-do frames that have already been done. One solution is to render for
every 32 frames. Then render every 32 frames halfway between these, then
every 16 frames halfway between these, then every 8 frames between these, etc.
An advantage is getting a quadruple framed version done which is important
during any time critical operations. If time permits, a double framed version
can be made by rendering the between frames.
Multiple machine parallelism
This is basically the standard way to render, especially when machines are
connected on a network.
Tricks to Animating Characters
Traditional animation is basically one trick after another. Whatever it takes
to get it working right on the screen is fair game. The following are trade
secrets [LASS94] to animating characters.
Most commercially available computer animation systems are based on animating
with keyframes. In hand-drawn animation, work on basic poses of scene first,
drawing poses of entire character so timing and acting can be worked out with
a minimum of drawings created. Once the poses are finalized, in-between
drawings are created to complete the action. With computer animation,
keyframes are values at certain frames for the articulation controls of a
model, which are usually set up in a hierarchy. The computer calculates the
in-between values based on a spline curve connecting keyframe values. Working
down the hierarchy of a model, creating keyframes for different controls at
each level of the hierarchy. Controls at some levels may need only a few
keyframes whereas at other levels keyframes needed on practically every frame.
An important point with this approach is to have a clear idea of the action
you want to achieve before starting.
2-D vs. 3-D
The biggest difference between hand-drawn animation and computer animation is
computer animation is truly 3D. Animation of a scene or parts of the
animation can be reused and you cannot tell that it's the exact same motion.
Looking at a scene of animation from a different camera angle, may appear to
look completely different. Similar scenes may be recreated by varying the
timing of the motion or changing the motion of an arm or head. This technique
makes it look different from the original. This is great for crowd shots,
where re-use of animation is an easy way to keep the crowd alive.
Weight & Size
The computer has the ability to create images that look absolutely real.
It's animation that gives the object its physical properties. The timing of
movement of an object defines the weight of the object. Two objects which are
identical in size and shape can appear to have two vastly different weights
by simply manipulating the timing. Light objects have much less resistance to
change of movement and thus require much less time to start moving. The way
an object behaves on the screen, the effect of the weight that is gives,
depend mostly on the spacing of the poses and less on the poses themselves.
No matter how well rendered a bowling ball may be, it does not look like a
bowling ball if it doesn't behave like one when it is animated. Proper timing
of motion can also contribute greatly to the feeling of size and scale of an
object or character.
The Thinking Character
When animating characters, every movement, every action must exist for a
reason. If a character were to move about in a series of unrelated actions,
it would seem obvious that the animator was moving it, not the character
itself. To convey the idea that the thoughts of a character are driving its
actions, a simple trick is anticipation; always lead with the eyes or head.
If the character has eyes, the eyes should move first locking the focus of
its action a few frames before the head. The head should move next, followed
a few frames later by his body and main action. If a character has no eyes,
such as inanimate object like a Luxo lamp [PIXA86], it is even more important
to lead with head. The number of frames to lead the eyes and head depends on
how much thought precedes the main action. The character's thought process
must be understood before any given action. The only time the eyes or head
would not lead action would be when an external force is driving the
character's movements, as opposed to his thought process. For example, if a
character is hit in the back by a low flying sheep, the force of the impact
would cause the body to move first, snapping head back and dragging it behind
the main action of body.
In hand-drawn animation, it is very common to animate action, then slow it
into a pose and hold the drawing of that pose for several frames, then move
into action again. In 3-D computer animation, as soon as you go into a held
pose, the action dies immediately. So instead of having every part of the
character stop, have some part continue to move slightly in the same direction,
like an arm, a head, or even have whole body. Even the slightest movement
will keep character alive.
Sometimes an action that feels believable in traditional animation, looks too
cartoony in computer animation. One needs to be aware of how far to push the
motion. The motion should match the design of the character and world. A
pitfall of using motion capture devices to create final animation is that
motion capture from human actors will always look realistic....for a human.
Apply motion capture to a chicken and it will look like a human in a chicken
suit. Motion capture data should be used as a starting place. Tweak the
timing and poses to make it more caricatured, then apply it to the chicken and
the motion will match the design of the character.
The personality of a character is conveyed through emotion. This is the best
the indicator as to how fast an action should be. When a character is happy,
the timing of his movements will be faster. Conversely, when a character is
sad, movements are slower. An example of this can be seen in Luxo Jr.
The action of Jr. hopping when the chasing ball, he is very excited and happy.
His head is looking at the ball, the timing of his hops are fast as there is
very little time spent on the ground between hops because he can't wait to
get to the ball. After he pops the ball, his hop changes drastically,
reflecting his sadness that the object of all his thoughts and energy of just
a moment ago is now dead. As he hops off, his head is down, the timing of
each hop is slower, with much more time on the ground between hops.
To make a character's personality seem real to an audience, he must be
different than other characters on the screen. A simple way to distinguish
personalities of the characters is through contrast of movement. No two
characters would do the same action in the same way. For example in Luxo Jr.,
when Dad & Jr. are hitting the ball with their heads, Dad being larger and
older, leans over and uses only his shade. Jr., however, is smaller, younger,
and full of energy, whacks the ball with his whole shade, putting his whole
body into it.
Readability of Actions
Proper timing is critical to making ideas readable. It is important to spend
enough time (but no more) preparing the audience for anticipation of the
action (follow through), the action itself and reaction to action (the follow
through). If too much time is spent on any of these, the audience's attention
will wander. If too little time is spent, the movement may be finished before
the audience notices it, thus wasting the idea. The faster the movement, the
more critical it is to make sure the audience can follow what's happening.
The action must not be so fast that the audience cannot read it and understand
the meaning of it. An idea or action must be made unmistakably clear. The
audience's eye must be led to exactly where it needs to be at the right moment.
It must not miss an idea or action. It is important that only one idea at a
time is seen by the audience. If a lot of action is happening at once, the eye
does not know where to look and the main idea will be overlooked. The object
of interest should be significantly contrasted against the rest of the scene.
In a still scene, the eye will be attracted to movement. In a very busy scene,
the eye will be attracted to something that is still. In most cases, an action
should not be brought to a complete stop before starting another action; the
second action should overlap the first. This slight overlapping maintains
flow and continuity between whole phrases of the actions. In Luxo Jr. it is
very important that the audience is looking in the right place at the right
time, because the story, the acting and emotion is being put across with
movement alone, in pantomime, and sometimes the movement is very subtle. If
the audience misses an action, an emotion would be missed, and the story would
suffer. An action has to be timed and paced so that only Dad or Jr. is doing
an important action at any one time, never both.
A Story Trick
In storytelling, the timing of ideas and actions is important to the audience's
understanding of the story at any point in time. It is important that
animation be timed to stay either slightly ahead of the audience's
understanding of what's going on with the story, or slightly behind. It makes
the story much more interesting than staying even with the audience. If the
animation is too far ahead, the audience will be confused; if the animation is
too far behind, the audience will get bored; in either case, their attention
will wander. When the action is timed to be slightly ahead of the audience,
it adds an element of suspense and surprise; it keeps them guessing about what
will happen. When the action is timed to be slightly behind the audience, the
story point is revealed to the audience before it is known to the character.
Many of these tricks can be used together in any scene in order to achieve the
strongest impact on the audience. In Red's Dream [PIXA86] at the end of the
dream sequence, Red juggles three balls and catches them with big a finish; the
crowd goes wild, and Red takes his bows. Slowly the circus ring dissolves to
the interior of the bike shop. The sound of applause fades into sound of rain.
Red unaware continues to take his bows and at this point, the audience has
not caught on to what's happening because the timing of the action is
slightly ahead of the audience. As the room reappears, so does the large "50%
OFF" tag hanging from Red's seat. The animation of the tag is timed to be
light in weight. It flops around more actively than anything else in the
scene. This contrast in action directs the audience's attention to the tag
which is a
subtle reminder that Red is still in the bike shop. The audience is now ahead
of the character and watches Red discover where he really is. Red's actions
were timed to be slow, emphasizing his sad emotion.
In every step of the production of the animation, the story, design, staging,
animation, editing, lighting, sound, etc., ask why. Why is this here? Does it
further the story? Does it support the whole? To create a successful
animation, you must understand why an object moves before you can figure out
how it should move. Character animation is the design of an object moveing
like it is alive. This is achieved when it looks to the audience like it is
thinking and all of its movements are generated by its own thought processes.
Thinking gives the illusion of life. Every single movement of the character
should be there for a purpose. It should support the story and personality of
the character. As soon as something looks wrong or out of place, the audience
will step back from your story and think about how weird that looked and you've
lost them. The goal is to create a personality of a character and a storyline
that will draw the audience in and keep them entertained for the entire length
of the film.
[BLIN94] Blinn, J., "Animation Tricks", SIGGRAPH 94 course notes and video:
Animation Tricks, July 1994.
[LASS94] Lasseter, J., "Tricks to Animating Characters with a Computer",
SIGGRAPH 94 course notes: Animation Tricks, July 1994.
[PIXA86] PIXAR, "Luxo Jr.", 1986.
[PIXA87] PIXAR, "Red's Dream", 1987.