For small objects, the cage can transform to reflect the adjustments on nodes and handler, but returns the some default forms (a circle in this case) when being activated again, depending on the shape:
For large objects, we only need to offset the control from object a bit:
Not the brightest example here, the whole point is in the future I hope AD dev consider inventing less intrusive ways to work with objects. These current way of distributing controls right on objects (Move, Node tool etc.) have been around for ages, I believe they can be changed in a better way.
Let's try assuming the best case when you editing an object:
1. The controls should not interfere with the view on object. There are indicators to know what you're working on, but the object remains intact in term of color, shape, stroke etc.
2. User should not lose perpective on the object. It stays at the same place on the screen
, no need to move/zoom in/out document for the sake of handling some controls.
On the ipad version, there's a preview displayed when holding nodes, my thought is what if we switch the role by allowing controls on the preview and see the object transforms instead?