I was wondering if the Dead Blending animation blueprint node is still considered experimental, or if it can be regarded as stable at this point. It has been very effective in resolving a number of artifacts in our animation system that the default inertialization doesn’t address.
I haven’t tried using normal blending yet, but since our setup is based on Lyra’s system, the blend artifacts occur when transitioning between cycle animations (e.g., WalkForward → WalkLeft, WalkForward → SprintLeft, etc.). As far as I know, the "Set Sequence" node—used to dynamically update the sequence on a Sequence Player—only provides an inertialization variant and doesn’t support normal blending.
Do you have any thoughts or recommendations on this?
Dead blending is stable and was removed from experimental status in May 2024.
As far as I know, the “Set Sequence” node—used to dynamically update the sequence on a Sequence Player—only provides an inertialization variant and doesn’t support normal blending.
In the Lyra graph, what that call does specifically is signal a blend time to the inertialization node downstream in the graph. This happens automatically and both the normal inertialization and dead blend nodes should pick it up. I’m not sure what you mean by normal blending in this context, but I’m guessing that you’re saying swap the animation and it will just not blend at all?
We have examples that use the newer blendstack node. It creates blends based on the animations you feed into it and serves as a base node that will be used for many animation features in the future. That would do the type of thing you are indicating. If you want to see examples of that, I would look at the Gameplay Animation Sample Project where we use it directly for the motion matching setup.
The reason I thought the feature was still experimental is because the documentation still labels it as such, and I couldn’t find any information indicating that it had ever been officially released as stable.
Also, to clarify what I meant by “it doesn’t support normal blending”: this approach of dynamically assigning a new sequence to the sequence player seems to only allow inertialization requests. It doesn’t appear to support the default blending method (I assume linear blending?), where you can define a curve and a blend time (crossfade duration).
I’ll definitely take a closer look at the BlendStack node you mentioned, as it might be exactly what I need.
Thanks for the advice and the fast response once again.