Spine provides two APIs for applying animations.
AnimationState supports applying animations over time, queuing animations for later playback, mixing (crossfading) between animations, and applying multiple animations on top of each other (layering).
AnimationState is stateful. It stores the animation times and other parameters for applying the animations. While a single AnimationState can pose multiple skeletons, it's somewhat rare to want multiple skeletons in exactly the same pose. Typically a single AnimationState instance will be used for each Skeleton instance.
AnimationState is built on top of the Timeline API and can handle most animation playback needs, except for playing backward. If reverse playback is needed, the Timeline API can be used directly or the animation can be duplicated and reversed in Spine using box select scaling.
update method takes the time that has elapsed since it was last called and updates the internal state. The
apply method takes a skeleton and applies the appropriate animations.
// Every frame:
Creating an AnimationState instance requires supplying an AnimationStateData. When the AnimationState changes the current animation, it will automatically mix (crossfade) between the animations using the mix durations defined in the AnimationStateData, resulting in smooth animation transitions.
stateData.setMix("walk", "jump", 0.2);
stateData.setMix("jump", "walk", 0.4);
stateData.setMix("jump", "run", 0.25);
stateData.setMix("walk", "shoot", 0);
AnimationState state = new AnimationState(stateData);
Rather than manually specifying the mix duration for every animation pair, a default mix duration can be set. Also, the mix duration can be set on a case-by-case basis using the TrackEntry mixDuration property:
entry.mixDuration = 0.6;
As indicated by the "Data" suffix, AnimationStateData is stateless. The same AnimationStateData instance can be used with multiple AnimationStates.
Tracks allow animations to be applied in layers. Each track stores an animation and playback parameters. Tracks are numbered increasingly starting at zero (track index is an array index internally). When the AnimationState is applied to a skeleton, the track animations are applied in order, starting with the lowest track number.
Tracks have many uses. For example, an animation that does not key everything can be played on a higher track, overriding the lower track only for what is keyed. Track 0 could have a walk, run, swim, or other animation and track one could have a shoot animation that keys only the arm and gun firing. Also, setting the TrackEntry alpha for a higher track allows it to mix with the tracks below it. For example, track 0 could have a walk animation and track 1 could have a limp animation. The alpha for track 1 is increased as the player takes damage, so they limp more and more.
Setting the animation for a track is done by calling setAnimation. This replaces the current animation and any queued animations on that track with the specified animation. If a mix duration is defined between the previous animation and the current animation, then the current animation will be mixed out over the mix duration so the transition between animations is done smoothly.
setAnimation returns a TrackEntry which can be used to customize playback in many ways.
By default, the animation will continue to be applied until another animation is played or the track is cleared. To stop the animation after a specific amount of time, the TrackEntry trackEnd time can be set.
To queue animations to play in the future, call addAnimation. This schedules the animation to play after the current or last queued animation for the track. If the track is empty, it is equivalent to calling
addAnimation returns a TrackEntry which can be used to customize playback.
When an animation is set on an empty track, it begins playing right away. Similarly, when a track is cleared, the animation stops being applied. To cause an animation to mix in or out, an empty animation can be specified which is an animation that has no timelines. The empty animation is used as a placeholder so a mix duration can be set. For convenience, the setEmptyAnimation and addEmptyAnimation methods are provided for setting or queuing an empty animation.
To mix an animation in from the setup pose, set an empty animation, add the animation to mix in, then set the mix duration:
TrackEntry entry = state.addAnimation(track, "run", true, 0);
entry.mixDuration = 1.5;
To mix an animation out to the setup pose, set or queue an empty animation specifying a mix duration:
state.addEmptyAnimation(track, 1.5, 0);
When an animation reaches the TrackEntry trackEnd time, properties that the animation keyed are set to the setup pose and the track is cleared. It may be desired to use
addEmptyAnimation to mix the skeletons back to the setup pose, rather than have it happen instantly.
The methods that set or queue animations return a TrackEntry which can be used to further customize playback. Refer to the TrackEntry API reference for the many settings that are available.
It is possible to keep a reference to the TrackEntry, for example, to adjust the
timeScale properties over time. However, care must be taken not to hold the reference past when the
dispose listener event occurs.
An application can register a callback to be notified of TrackEntry lifecycle events.
addListener on the AnimationState registers the listener for all TrackEntry events. A listener can also be set on a specific TrackEntry to receive only events from that entry.
The events that are possible are listed on AnimationStateListener and are guaranteed to occur in the order specified. Events are queued during the internal processing done by the AnimationState
apply methods and listeners are notified afterward, just before the method returns. This makes it safe for listeners to manipulate the AnimationState, for example to set animations or clear tracks. However, all events that have already been queued will still be fired unless clearListenerNotifications is used.
Changes made to the AnimationState in a listener, such as setting a new animation, are not applied to skeletons until the next time AnimationState
apply is called. This can be done in the listener, but care must be taken to first call
state.setAnimation(0, "jump", false);
state.update(0); // Advance internal state.
apply method does not change any internal state, allowing a single AnimationState to be applied to multiple skeletons. Calling
update lets the AnimationState know all the applying is complete and subsequent
apply calls are for the next frame. If
update is not called,
apply may trigger the same listener notification resulting in an infinite loop and stack overflow.
The Timeline API is the lowest level API for applying animations and consists of the Animation and Timeline classes. These classes are stateless, so the time and other parameters for applying the animation must be stored and manipulated externally. This API provides the most control over animation playback, but requires more work to manage the playback state yourself. Most users will prefer to use the AnimationState API.
An Animation is very simple, it has a name and a list of Timelines. Each timeline knows how to modify a specific skeleton property over time. Applying an animation to a skeleton is done by calling apply for each timeline in the animation.
alpha = 1; // For mixing between the current or setup pose (0) or the animation pose (1).
blend = MixBlend.first; // How the current or setup pose is mixed with the animation pose.
direction = MixDirection.in; // Whether mixing out to the setup pose or in to the animation pose.
for (Timeline timeline : animation.timelines)
timeline.apply(skeleton, lastTime, time, events, alpha, blend, direction);
// The events list contains any events fired between lastTime and time.
// Process them here, then clear the list.
lastTime = time;
Animation has an
apply convenience method for doing this which has a
loop parameter and simply calls
apply on each timeline.
animation.apply(skeleton, lastTime, time, loop, events, alpha, blend, direction);