By default, PE will be inactive and wait for any kind of input.
Whenever PE/Set is greater than 0, the layer will advance to Warmup, which is responsible to put the avatar in a posable state. The warmup transition is forked, depending on whenever Eyelook is engaged or not. The difference is whenever eyes should be included in pose space or not.
The track fork allows experimental partial tracking while posing.
Once that is done, we move on to the state selected with PE/Set. Dances and Poses are treated the same here, only difference being how they use the PE/Float value.
A direct transition from Warmup to Restore should ensure that the logic works and syncs reliably in case of fast or invalid inputs.
If PE/Set changes away from the current category, the layer will advance to Restore and loop back.
Changes between banks of the same category will be direct, without going to Restore first.
Each state machine contains a normal and a mirrored pose of each category’s banks, PE/Mirror controls which one is selected.
Restore re-engages tracking, essentially reversing changes done in Warmup
Introduce your own data using the modding prefab, more infos about that on the modding page.