Animancer - Less Animator Controller, More Animator Control

Hey everyone, Animancer v7.1 is now up on the Asset Store and itch.io. It has a couple of improvements to Mixer Transition Drawers, but mostly consists of minor bug fixes (including the .NET Framework version issue @itadakiass found).

 

Did something change with OnEnd in 7.1? Suddenly the code I call gets called permanently instead of once...

 

The first dot point in the Fixes section relates to End Events, but it doesn't seem like it should be the cause of what you're describing and all my Unit Tests still pass.

Can you send a minimal reproduction project to animancer@kybernetik.com.au so I can take a look at it?

 

I currently have no time to do a test case (sorry!), but I seem to have fixed it. In 7.0 and before I used animState.IsPlaying = false; inside the OnEnd() event to stop it from getting called more than once. I replaced this line with animState.Stop(); now and it works as before.

 

I've fixed that issue and re-uploaded v7.1. Didn't increase the version number since it shouldn't affect most people.

 

Is there a way to cheaply check if a layer has a state registered by key or better yet retrieve it (or null) without implicitly creating it?
I'm looking up transitions/clips from a list and want to avoid that cost when accessing the state to update it incase it already exists.
So far i only see the indexer and GetOrCreate.

 

There's only one dictionary so

animancer.States.TryGet

then check if its on the correct layer.

 

Ok, i guess my intention may just be wrong. Now that i think about it, it's probably better to have only overriding clips/transitions on the upper body layer and just fade it in/out accordingly, and not try to sync them otherwise.

 

I still have a conceptual hurdle. This means, if the character does a full-body animation, i play it on the base layer (upper body with 0 weight), and when he begins to move while still performing that animation, i have to stop it, play it on the second masked layer (weight 1), setting the time to the same progress value and start playing the movement animation on the base layer. When the character stops again while still performing the action, i have to stop the base layer and the upperbody layer, play/continue the action animation on the base layer again also transferring its active time value. This seemingly wouldn't allow me to fade the base layer's animations then (moving lower to fullbody action), or the upper body (as played in the base layer with 0 weight for upper body layer) would skip or fade too. The documentation shows how to enter the upper body action animation, but fading back to fullbody is my problem.
Do i miss a crucial step somewhere or is there an alternative way to achieve what i intend (smoothly transition in and out of fullbody-action and lower/upper movement+action separation)?

 

The Layers example now supports swapping between moving and idle while performing the action. Basically, it uses FadeMode.FromStart to create a new state if the existing one has non-zero weight so that the new state can fade in on the desired layer while the old state fades out on the old layer. It seems a bit hacky, but I haven't been able to come up with a better solution.

 

I've got my character set up with a mixer to blend between walk/idle/run animations, and I want to be able to change these animations at runtime when the character picks up an item. Is there a way to substitute the clips inside the mixer (like what would be done with an Animator Override Controller in mechanim)?

Also to the discussion on swapping between moving and idle - I solved this by always playing the action on the upper body layer and enabling/disabling the upper body layer mask when the character moves. There is no blending but my animations don't require it as the lower body movements in my action are subtle.

 

mixer.GetChild(x).Clip = ...

would do it.

Or if you're replacing all the animations, just make a separate mixer for each set.

 

Thanks, that did the trick!
For anyone else with the same issue - I was using a LinearMixerTransition rather than a LinearMixerState so I actually had to call

mixer.State.GetChild(x).Clip

to find the clips.

 

Ive been looking around at the animancer website and inside animancer source code but cant seem to find an answer to a question i have, so here i am.

I have animations set up with events. The events are set up inside the animation file itself through unitys animation import settings. AnimationA has events. AnimationB doesnt. I am interupting AnimationA with AnimationB, however the events from AnimationA are still triggering. Ive tried

animancer.Layers[layer].CurrentState.Events.Clear();

but it doesnt seem like its working. Im wondering if this is only for

AnimancerEvent

? if this is true, how/where exactly are the unity events that are set up through the import settings being triggered? any way I can stop them from triggering when interupting and blending to a new animation or should i be manually setting up

AnimancerEvent

instead of in the animation import settings?

EDIT*
the more i think about it the more it seems like these events have nothing to do with animancer and they are simply unity events. i think i will go ahead and try implementing these events as animancer events instead

EDIT2*
replaced unity events with

AnimancerEvent

. using animancer events instead solved my issue. now animations that are interupted also have their events interupted.

 

Yeah, Animation Events are entirely separate from Animancer Events.

Animation Events keep getting triggered while the animation has any weight as it fades out.

Animancer Events are Cleared Automatically whenever you play something.

 

I'm using Version 7.1 but when in

Update()

I call:

Code (CSharp):

1. animancerComponent.Play(walk);

I get this error:

ArgumentOutOfRangeException: AnimancerLayer.GetOrCreateWeightlessState has created 5 states for a single clip. This is most likely a result of calling the method repeatedly on consecutive frames. This can be avoided by using a different FadeMode or calling AnimancerLayer.SetMaxStateDepth to increase the threshold for this warning.

Is this the same issue or did I do something wrong? Before 7.1 if I remember correctly I could call

Play()

multiple times with the same clip and it wouldn't restart it.

 

If walk is a transition with its Start Time field enabled and it's already playing, it will create a new copy of the state to fade in while the old one fades out. So if you don't want to restart it, just untick the Start Time.

 

Ah, thanks! I noticed that Start Time is checked by default now. Unchecking it solved my issue. Btw, thank you very much for C# in Unity! It helped me a lot in learning C#.

 

Good to know, i had the same problem and for now thought i had to find the error somewhere in my code.

 

Do you think it's better to have it checked by default or should I revert it to unchecked? I find myself needing it checked more often, but I suppose it is more likely to cause problems if left on unintentionally than if left off unintentionally.

 

Hi, I was wondering if it is possible to ease the pausing of an animation? I guess one way would be lerping the speed, but is there a built-in function that I haven't found yet? I'm pausing the animation if a specific button is *not* pressed, and it would be nice if it wouldn't pause immediately but with some easing.

 

Well, it caught me by surprise and I couldn't find the cause of the issue so I looked for help on the forum. When default was unchecked and I tried to re-start animation and it didn't work it was clear that by default the animation is not restarted and I just needed to read the docs. Also first animations I started using were running and idle so I guess it's a better default for beginners.

 

Another question, it seems that transitions are not affected by Time.timeScale. Is there another way to see them in slow motion? The animations themselves are played in slow motion, however. It's just transitioning between them that happens at the same speed.

UPD: Fixed by removing

UNITY_2021_1_OR_NEWER

in

AnimancerPlayable.UpdateAll

. (I was using 2020.3)

 

@theghostronaut There's nothing inbuilt, but you could make a class that implements IUpdatable to lerp the speed over time like the MixerParameterTween script.

@kostyaunity Which version of Unity are you using? There's a bug in 2021 that causes it to not take time scale into account so I had to manually multiply it, but if the bug is also in an earlier version then my fix won't be applied. It's in AnimancerPlayable.UpdateAll.

 

looks like new transition preview window not work with odin

upd:
yep,
[DrawWithUnity] fixed it

 

Try putting Odin's [DrawWithUnity] attribute on your field. I've added a documentation page about Odin.

 

I'm using Unity 2020.3.14f1 and removing the

if UNITY_2021_1_OR_NEWER

condition has fixed my issue. Thanks for the help!

 

With AnimancerLayer.CreateState<T> removed, what is the proper way to add a state? In the older version I was adding all states prior to playing them at initialization to work around a Unity bug with AnimationRigging resetting values when the playable graph changes. I do not see a way to add LinearMixerState states anymore.

Edit: looked up the old source, looks like I should use AddChild()

 

You can just make a

new LinearMixerState()

then use

animancer.Layers[x].AddChild

to attach it to a specific layer.

 

Wondering if anyone has added async/await support to Animancer? I didn't see anything about so figured I would give it a shot and it was surprising easy to implement.

I added this script to the source code.

Code (CSharp):

1. using System;
2. using System.Runtime.CompilerServices;
3.  
4. namespace Animancer {
5.     partial class AnimancerState : INotifyCompletion {
6.         private Action _continuation;
7.  
8.         public bool IsCompleted {
9.             get {
10.                 return !IsPlaying && TargetWeight == 0;
11.             }
12.         }
13.  
14.         public void GetResult() { }
15.  
16.         public void OnCompleted(Action continuation) {
17.             _continuation = continuation;
18.         }
19.  
20.         private void OnRequestCompleted() {
21.             Events.OnEnd -= OnRequestCompleted;
22.             _continuation();
23.         }
24.  
25.         public AnimancerState GetAwaiter() {
26.             Events.OnEnd += OnRequestCompleted;
27.             return this;
28.         }
29.     }
30. }

and now I can write async code, here's an example:

Code (CSharp):

1.         async Task PlayLandedAnimationAsync() {
2.             await Animancer.Layers[HumanoidEntity.LowerBodyLayer].Play(landingAdditiveAnimation);
3.             Animancer.Layers[HumanoidEntity.LowerBodyLayer].StartFade(targetWeight: 0f, fadeDuration: 0.25f);
4.         }
5.  
6.         void Landed() {
7.             PlayLandedAnimationAsync();
8.         }

I will experiment with using UniTask next as it works much nicer in Unity (no allocations, tied into runtime much better). This might be nice to add to the asset

 

What happens if you play something else during that animation? Would it cancel the fade out properly?

And are you sure it doesn't allocate any garbage? I'm pretty sure

+= OnRequestCompleted;

will always allocate a new delegate and

-=

might too.

Also, why not just use the OnEnd event directly instead of adding the _continuation field? I'm not familiar with async stuff so there could be something I'm missing, but if you're accessing the OnEnd event anyway it seems like you might as well just use it directly.

 

Yeah, the above code was done really quickly without much testing, it does have some problems, but I think they are solvable. Probably more changes to Animancer would be required to support async/await in a better way.

I was using OnEnd as a way to simply know when to call

_continuation

once an animation completes. This is how the async/await generated code knows how to continue execution after being blocked. I would like to avoid using this but haven't dug through enough of the code to know where to modify. All I need to know is a specific place in Animancer to call the

OnRequestComplete

and I will no longer need to deal with the Action event.

 

I think I see what you are saying. This might actually work.

Code (CSharp):

1. using System;
2. using System.Runtime.CompilerServices;
3.  
4. namespace Animancer {
5.     partial class AnimancerState : INotifyCompletion {
6.  
7.         public bool IsCompleted {
8.             get {
9.                 return !IsActive;
10.             }
11.         }
12.  
13.         public void GetResult() { }
14.  
15.         public void OnCompleted(Action continuation) {
16.             Events.OnEnd = continuation;
17.         }
18.  
19.         public AnimancerState GetAwaiter() {
20.             return this;
21.         }
22.     }
23. }

 

I gave that a go and it seemed to mostly work, but according to the Profiler the following script is allocating 480 bytes every frame:

Code (CSharp):

1.     public AnimancerComponent animancer;
2.     public AnimationClip action;
3.  
4.     async void Update()
5.     {
6.         await PlayAnimationAsync();
7.     }
8.  
9.     async UniTask PlayAnimationAsync()
10.     {
11.         await animancer.Play(action);
12.     }

I'm also unsure how it's supposed to know if the awaited task has been cancelled. If you play something else the old End Event will be cleared (unless

AnimancerState.AutomaticallyClearEvents

is set to false), but it isn't doing anything to inform the task that the

continuation

delegate will never get called. The only possibility I can think of is that it simply allows the delegate to get garbage collected, but in that case it's obviously impossible for the system to have no allocations.

 

C#'s implementation of async/await will allocate a mini state machine so that is probably what you are seeing. I believe if tried with UniTask this will be resolved as it supports async/await without allocating. It is also has cancelation support, but I have not dug deep into it yet.

 

In this case, what is the states' key for duplicates if i wanted to access/check for them from somewhere else?, i thought key's had to be unique.

 

Each duplicate uses the previous state as its key so they form a chain. It's implemented in the AnimancerLayer.GetOrCreateWeightlessState method.

 

@Kybernetik
[SOLVED]
I had a particular condition misfiring.
Thank you


-----

Hi there!
Occasionally my avatar goes into the T-Pose, and I am unsure why ( I must have some leak somewhere). What I am wondering is, can I notify myself when this occurs, some way?

Thank you

 

G'day! Just updated to 7.1. I'm running into this issue with events,

NullReferenceException: Object reference not set to an instance of an object
Animancer.AnimancerState+EventDispatcher.NextEvent (System.Int32 playDirectionInt) (at Assets/Plugins/Animancer/Internal/Core/AnimancerState.EventDispatcher.cs:643)
Animancer.AnimancerState+EventDispatcher.CheckGeneralEvents (System.Single currentTime) (at Assets/Plugins/Animancer/Internal/Core/AnimancerState.EventDispatcher.cs:530)
Animancer.AnimancerState+EventDispatcher.Animancer.IUpdatable.Update () (at Assets/Plugins/Animancer/Internal/Core/AnimancerState.EventDispatcher.cs:264)
Animancer.AnimancerPlayable.UpdateAll (Animancer.Key+KeyedList`1[T] updatables, System.Single deltaTime) (at Assets/Plugins/Animancer/Internal/Core/AnimancerPlayable.cs:1194)
UnityEngine.Debug:LogException(Exception, Object)
Animancer.AnimancerPlayable:UpdateAll(KeyedList`1, Single) (at Assets/Plugins/Animancer/Internal/Core/AnimancerPlayable.cs:1199)
Animancer.PostUpdaterepareFrame(Playable, FrameData) (at Assets/Plugins/Animancer/Internal/Core/AnimancerPlayable.cs:1293)

I'm not sure what's causing it but seems to be related to this -

Code (CSharp):

1.     public void Test()
2.     {
3.         anim.Play(clips.testClip, 0.15f).Events.Add(new AnimancerEvent(0.5f, () => TestEvent(0.25f)));
4.     }
5.  
6.     public void TestEvent(float blendTime)
7.     {
8.         anim.Play(clips.eventClip, blendTime);
9.     }
10.    

All of the clips exist, do you know what might cause that?
Thanks!
Pete

 

Try this fix.

 

That seems better, thanks!
Hey one other thing I've noticed, the animancer component doesn't seem to update very well in the editor for me while playing the game. It only seems to update once if I mode the mouse over the component. Is there a way to force that to update, it would be handy for debugging stuff.

 

Right Click any state or layer in the Inspector and enable Display Options -> Repaint Constantly. It should be enabled by default though.

 

Hey thanks, that's on but it doesn't seem to make a difference.
Some cool options I didn't know about in there though.

 

Try it in an empty project, maybe something else is interfering with it.

 

Yep, cool I’ll do that and get back to ya!

 

When I have triggered an animation using state.Play(), do I need to stop the animation at some point? I noticed the Normalized Time counter in the inspector keeps running after the animation has finished.

 

If you have a non-looping animation and you want to leave it on its last frame for a while you could call animancerComponent.Playable.PauseGraph() to save a bit of performance, but otherwise there's no real harm in just letting it continue.

 

Thank you!

 

Hey @Kybernetik, is there a way to exclude a specific object from updating with Animancer? I have a few objects in the rig that I use purely for secondary animation (applied after everything), and they seem to be affected sometimes when I play a new clip.

 

Not that I know of. The Playables API always controls every field that any clip in the graph affects regardless of what's playing and I've never seen anything that would let you exclude certain objects.