blgjdg is right that a mesh deformer or "cage" is not the same as weights. Weights can do much of the same job, but the cage is a simplified envelope with different controls. It hides some of the complexity of both the mesh and weights. It is also easier to apply a cage to different images or to multiple images, which is one of the largest benefits.
For the cage, under the covers the images still need to be meshes. Those can be generated automatically, as they are typically a uniform grid. That can hide performance implications compared to creating a mesh explicitly. Also a little more processing is needed to map the cage controls to deforming the mesh.
We don't currently have plans to add cage-style mesh deformations. It's a cool feature, but we have other features we want to get to first and for now you can achieve similar results in other ways.
