Versy When a skeleton's animation gets stuck, it is often due to the animation being repeatedly re-set on the same track.
From the screenshot you provided, it is unclear exactly how this occurs, but you may be able to work around it by inserting a Branch node.

For example, as shown below, the Spine -> Animation -> Get Current node retrieves the TrackEntry for a specific track, and the Spine -> Track Entry -> Get Animation Name node obtains the animation name. If the retrieved animation name does not match the next animation to be set, the Set Animation node will be executed:

Could you try ensuring that the same animation is not set consecutively to see if that clears the stack?

Versy Thanks for the additional clarification. It certainly looks like this is not a problem with the animation being applied repeatedly, but rather a problem with poses that should be reset not being reset. Could you please email us a minimal UE project that can reproduce this problem?: contact@esotericsoftware.com
Please include the URL of this forum thread in the email so we know the context. Then we can take a look at what's wrong.

Versy I'm glad you were able to narrow down the problem to some extent! However, the code you showed us does not seem to be wrong, would you still be willing to email us a minimal UE project that can reproduce the problem? Then we can look into it in more detail.

Versy Thanks for providing the UE project files! I wasn't sure at what level you were supposed to check, so I just wanted to make sure I was correct that the behavior you see in the following video is the problem you're encountering?:

The procedure to reproduce this is to press the H key to make Spineboy walk to the left for a moment, then release the H key.
Some of Spineboy's poses will then be incorrect, but the idle animation itself seems to be applied, and you can see that some bones are animated correctly.

As far as I could tell from several attempts, this problem did not seem to occur all the time, just occasionally. It certainly seems to be a problem with the mix, as you mentioned that if you set the mix time to 0, this problem does not occur. Once we can confirm that this is the problem you are experiencing, we will investigate the cause.

Versy Thank you for your response! Our runtime development team is currently engaged in high-priority tasks, and we anticipate that investigating this issue may take some time. We will provide an update in this thread as soon as we have further information.

Versy

It took long, but we finally should have discovered the issue. There is a fix to port to C/C++ runtimes that we still have in our queue. Here you can find the GitHub issue.
You can manually port it, or just wait until tomorrow when I will port it. But you'll need to copy it to your project eventually.

There's an additional consideration. Now that you are reducing the framerate, you could consider skipping updates where delta is 0. There are currently consecutive state and skeleton update invocations when Step is 0, leading generally to a waste of resources. If you place:

if (Step <= 0.0f) return;

in your code, you won't have the issue you reported even if you don't port the fix if you return when Step <= 0.0f.

Little update, we've just ported the fix

Versy

Are you sure you didn't make any other modifications to our runtime? The fix we implemented is in the spine-cpp runtime, which is a requirement for spine-ue and should be manually copied as described in step 6 of the Installation paragraph in the spine-ue guide.

If you've done all of the above, I need to ask you to again send over a project that reproduces this issue, as I cannot reproduce the slow motion problem you're experiencing on your end.

Versy

I’ve been thinking more about your code, and I suspect the slow motion you’re experiencing might be caused both by your implementation and potentially by your machine.

When you calculate Step like this:

Step = Accumulate - FMath::Fmod(Accumulate, ActualFPS);

If, for any reason, Accumulate exceeds twice the ActualFPS, you’re only advancing your timeline by one step, when in fact you should be advancing it by two steps.

I also remembered we have a Unity example that does something similar here. I adapted the logic to C#:

float Step = DeltaTime;
if (bDisableInterpolation)
{
	Accumulate += DeltaTime;

	const float FrameDeltaTime = 1.0f / DiscreteFPS;
	constexpr int32 MaxFrameSkip = 5;
	int32 FramesToSimulate = 0;

	while (Accumulate >= FrameDeltaTime)
	{
		FramesToSimulate++;
		if (FramesToSimulate > MaxFrameSkip)
			break;

		Accumulate -= FrameDeltaTime;
	}

	if (FramesToSimulate == 0)
		return;

	Step = FramesToSimulate * FrameDeltaTime;
}

The MaxFrameSkip should prevent steps from getting too large.

Let me know if this version works better.

Versy

Your computer is a beast, so that’s definitely not the issue. It might have been a problem if it were a slower machine.

There must be some difference in our configurations. I even asked @Misaki to test the latest version of the plugin, and everything works correctly for her as well.

Are you sure you're recompiling the plugin after making changes?

What’s suspicious to me is this part:

Versy New plugin / Modded w/ my original code: disable interpolation doesn't work, BUT no slowdown

But disable interpolation should work with the new plugin and your original code!

Also, make sure to check the value of the Discrete FPS field under SpineSkeletonAnimation in the Unreal Engine UI.