A successful solution was found to almost the entire problem of registration or at least its major weaknesses. This was done simply by changing the optimisation so that it searches a valid range of possible warps (the space of warps in essence) and runs for a long enough period of time to benefit from computational savings. Also, an increasingly higher tolerance was sought for the optimiser, whereupon registration using statistical models finally become possible and even successful. More on these solutions and successful heuristics are to be listed in the remainder of this broad subsection.