Luckily the baking feature has just recently been re-activated for newer Unity versions as well, so this is now again also an option. 🙂
Performance wise, the baked clips might even yield slightly better performance (depending on your skeleton setup), but there are some major drawbacks that are important to know:
- Only a subset of Spine's functionality is baked to clips.
- Animated Attachment visibility will be e bit off sometimes. E.g. in Raptor's gungrab animation, the gun will shortly be shown twice - the version with the hand will be shown at the same time as the gun-nohand attachment.
- Additional animation layers might not add up as expected (as previewed in the Spine Editor).
- Constraints are baked as is, so e.g. a aim-at-target constraint (as shown in the
4 Object Oriented Sample
scene) can no longer be modified at runtime.
- There will be a lot of additional keyframes for each animation, creating additional data. This is because not only the keyframes set by the user will be present, but instead every
bake-increment
of 1/60 second, there will be a keyframe (to prevent interpolation differences).
So in short: it is not really a clean solution to use the baked clips, uncless you use only a small subset of Spine's functionality. We would recommend baked clips only as a last resort. The StateMachineBehaviour
would most likely be the preferred solution in your case.