Creative Commons License

Velocity Graphs

Up until now, whenever we have keyframed motion for a layer it has been constant motion. From the moment the layer leaves one keyframe and goes to the next it is traveling at a single, constant velocity. This is fine except for one small problem - in the real world, nothing moves at a constant velocity...and unconciously we are aware of this. Motion with constant velocity looks mechanical and computer generated.

In the real world, things have mass which must be accelerated and decelerated in order to move or stop moving. A ball cannot be still one moment and then instantly moving full speed the must gradually progress through a constantly increasing speed. Once it's moving, for it to come to a stop it must progressively slow down first. Because everything in our physical world moves with acceleration and deceleration we must make our objects do the same in AE if we don't want the movement to look computer generated.

Constant, linear velocity and abrupt starts and stops are one of the most obvious signs of a beginning animator...because they are easy to do. Mimicking the real world, making movement look natural, requires two things - an ability to learn from watching the world around you, and Velocity Graphs.

Without velocity graphs, if you wanted to slow an object down before it stopped you would have to set several keyframes, each moving slightly slower than the previous one...this could get very tedious, very fast. Even if you could get the motion you wanted this way, if you wanted to change it you would have to change every single keyframe. With velocity graphs we get around all this extra work. Velocity Curves give you the ability to vary motion velocity between 2 keyframes.

A Velocity graph is similar to any 2 axis graph. On the vertical axis is your parameter value, such as position. On the horizontal axis is time...starting at one keyframe, and ending at the next. Imagine at the first keyframe your layer is at position 0, and 2 seconds later (on the last keyframe) it's at position 100. If you draw a straight line between those two points you get a diagonal line. This represents a constant rate of speed as the object moves from one point to the next.

If we leave the two keyframes where they are, but change the shape of the line between them, suddenly we get variable speed movement between the two points. If we make the curve convex the object will start out really slow, gradually picking up speed, then come to a sudden stop at the second keyframe. If we make it concave the object will leap forward very quickly, but then gradually slow down as it approaches the second keyframe. If we make an s-curve out of the line, the object will slowly accelerate, travel quickly when it's halfway between the 2 keyframes, then gradually slow down until it comes to rest at the second keyframe. Suddenly the motion between our two keyframes is no longer abrupt and has begun to move like things do in the real world.

Unfortunately, learning the mechanics of manipulating the curves is the easy part...the hard part is learning to translate what you see around you into those curves, and the only way I know of to do that is practice...

Manipulating Velocity Graphs

Any property which changes over time (i.e. anything that can be keyframed) has a curve that can be changed. All of the curves work in a similar manner, although there are some differences in how they affect the various properties. To access a properties curve:

One thing to keep in mind is how much time you have between keyframes. If 2 keyframes are only 15 frames apart, you do not have enough frames to make very subtle changes in speed. After a certain point acceleration will be too fast and the layer will appear to jump. Likewise, if your keyframes are to far apart the progress may be too slow for you to create natural looking motion. However, you do have plenty of frames for creating very subtle changes in velocity.

The effects of your changes in the velocity graph can also be seen on the motion paths in your comp window. These paths are comprised of dots which each represent a single frame. When motion is linear all dots between 2 keyframes are spaced evenly. Once you start to change the graph, the dots will shift closer or father apart from each other to visually represent the change in motion that the graph produces.

Position and anchor point each have only a single graph - velocity. However, rotation, opacity and scale get a little more complicated. Each has two graphs - velocity, or how fast a layer gets bigger or smaller, and value. The Value Graph represents the property value itself, how big or small an item gets, how far it turns, or how visible it is. You'll want to start by working with velocity, because that is the mor intuitive of the two, and then try value once you start to get comfortable comfortable with curves in general.

The tricky thing to remember with value grpahs is that you can go above or below the keyframe values at either end of the graph. So if you set a layer to rotate one time, but then change the curve so that it raises above the point for a single rotation, your layer will spin more than once and then have to spin backwards to land on the value set in the second keyframe. This kind of unexpected change of direction is something you need to watch out for with value graphs. Again, start out with very minor changes to the curve until you understand how those changes affect your layer.