I am trying to prototype an advanced third-person controller in Godot for humanoid characters. Currently, the controller uses two concurrent state machines: one for locomotion states and one for posture. I also plan to add an "Ability System" that will control all character actions (attack, reload, etc.).

I have already implemented basic states:

LocomotionSM:

InPlace
Walk
Run
Sprint

PostureSM:

Stand
Crouch

Now I want to add animations to the character’s movement, but I’m unsure how to approach this in a way that will keep the system easy to extend with new states. The main problem is how to combine locomotion animations with posture.

My initial idea was to assign standard animations to locomotion states and, in the PostureSM, only use idle animations for standing and crouching poses. Then I would blend animations from both state machines - for example, legs moving as in the walk state while the whole character is crouching. However, I realized this is not a good approach, because I cannot guarantee that the blended animations will match properly.

My second idea was to prepare animations for each possible combination of locomotion and posture. Based on the states in LocomotionSM and PostureSM, I would select the appropriate animation. However, this also seems suboptimal, as it introduces a two-dimensional matrix of animations, which could grow rapidly as I add more states - even if some locomotion states are not valid for certain postures.

Now I’m wondering how this should be done to avoid complexity as the system grows. I assume that blending animations will work well for combining locomotion/posture with abilities, since abilities mostly involve upper-body actions. I previously created a prototype of an isometric shooter in Unreal Engine 5, and blending worked well there, so I believe it could help avoid a three-dimensional animation matrix. However, I’m still unsure how to handle movement animations properly.

I am not an artist, so I want to minimize the amount of animation work by using assets from Mixamo and some purchased animation packs.

I also wonder whether I should assign animations directly to states and play them on state changes, or use an AnimationTree with parameters (from a blackboard or state machines) to control animation transitions. I’m a bit concerned that the graph in the second approach might become overly complex.

Questions:

1st question: How should I combine locomotion and posture to minimize the number of required animations? Should I blend them together, or prepare animations for each combination? Or is there another approach?

2nd question: For a controller with concurrent state machines, should I assign animation resources directly to states and play them on state changes, or should I use an AnimationTree driven by parameters from the state machines?

Gamedev

Message History

Delivery option(s)