I guess it’s just a matter of personal preference really, and personally, I prefer to use armatures every time I can. The main reason why I prefer armatures is animation work flow, it makes life easier. Working with the animation editors in blender is object based.
When you create an animation, blender creates an action for the object. Key the loc/rot of the camera on a frame, key the loc/rot of the camera on a different frame. Simple enough, and you have created an action for the camera. Say you want to try out a different camera angle, to see how it looks. Simple enough, create a new action for the camera and insert keys as needed. Easy enough. You can see your keys in the dopesheet editor, but to switch actions on the camera from one action to the other, you’d have to switch from the dopesheet editor to an action editor, and change the action there. Still simple and easy enough.
Let’s say instead of animating the camera, you’re animating a mechanical leg that has a foot, lower leg, and upper leg, 3 mesh objects total. As you animate, each mesh object gets an action created for it. 3 actions created. Now you want to create a new animation, you need to create a new animation for 3 objects, so in the action editor, you have to select each mesh object and add a new action. That’s a little bit more work, select each object and change it’s action. (btw, I didn’t watch the videos posted, but I assume he uses empties as controls and also as targets for some constraints. But empties are objects, so if it’s used as a control, it get’s its own action.). This is where human error creeps in. You have three objects, each one can have 2 actions. You’d need to make sure all 3 objects are using the 1st action created, and when you wanted to try the second animation, you’d have to switch the action on 3 objects.
What if you missed not changing one action on one object? 2 objects using action ‘a’ and 1 object using action ‘b’ just wouldn’t work as they are different action sequences.
You rig the same mechanical leg using a 3 bone armature object, you create 1 action that has keys for all three bones. Because animation is object based, your armature object gets 1 action created for it with keys for 3 bones. Now when creating a new action for the armature, you only create 1 new action for the 3 bones. You only have to change the action on the armature object to switch between action ‘a’ and ‘b’, instead of 3 objects and switching 3 actions.
There are other advantages as well for using an armature. I’d like to see 2 complex animation sequences created without using an armature on jet shown in the video…
Just my thoughts…