Like learning to *fall off* a bike

Question

I am very new to Blender, but not to 3D. My ultimate question is what are some alternate options for rigging a bike/rider system in Blender and what are the pitfalls to watch out for...

---

The goal

I have a Rigify-rigged character Armature. I have a custom bicycle rig Armature. I need this character to mount the bike, ride the bike, then fall off the bike.

---

Current Solution

On a previous Blender project, I used Child-Of Constraints on the character Armature. Hand and foot IK handles were assigned Child-Of Constraints to Empties. Those empties were parented to joints within the bike armature. (Blender vets: Is this a bone-headed move?)
The character's Root was given a child-of Constraint to an Empty on the bicycle armature placed at the seat position. My belief was that empties parented to handlebar and pedal bones allowed for easier fine-tuning of the rider's hand and foot positions. (Misguided? Right idea?)
I then chose a frame where the fall would occur and keyframed the Influence on each of those Constraints to zero, shifting the character back into global coordinate space to separate them from the bike.

I was absolutely brand new to Blender and tried the first tools/processes I found.
Are there better solutions for a dynamic bicycle and rider setup?
Are there any advantages to them being in the same armature? Caveats?
What level of the character is best for constraining (Armature Object level versus some root joint)?

Answer 1

In my opinion you're on the right way. My advice is to leave all object level data into a default rest position (000, 000, 111) and perform all move rot and scale animation and constraint in pose mode, so that they get stored directly into the action.

You can add some extra bones in edit mode and use them instead of empties, so that everything is stored into two actions only (Character action and bicycle action).

Joining the two armatures could result in managing one action only but, as it's possible to select both armatures and enter pose mode and animate both, I don't think it's so useful, unless you need to export to another software one single mesh (Bicycle and Character) with one single action.

When I build thoose kinds of setup, often after finishing the animation I make a bake of the constrained bones (visual keying, clear constraints, overwrite current): it's a destructive editing but it lets you get rid of all constraints and results in a simple exportable animation (pose mode > pose menu > animation > bake action).