Fairly simple suggestion actually. It would make sense to have the output mapping initially match the input mapping. For a couple of reasons.
When you go nuts mapping input sections, you currently end up with all layers on top of each other in the centre of the output sections which you have to drag out and around, them all being behind each other.
When mapping an object where the input image and projector aren't drastically different to begin with this is a big time saver because you're immediately in fine-tune mode.
Even if you're input is a layout of imagery (not coherent to the output) it would make sense to be able to rapidly map the input and see it all in place in the output, rather than on top of each other, then begin your output mapping.