Most sheet metal components are made by deep drawing, which requires expensive tooling. Although many new flexible forming processes have been invented, they have largely not had industrial application, so it would be valuable if intelligent means to design new processes existed. This has not previously been attempted, although there has been work to classify both products and processes and to define optimal forming processes. A body of work in garment production examines the optimal flattening of garments, starting from their final form on a human body, to deduce the best cutting pattern from flat fabric. This paper develops a related approach for the first time, “un-forming” sheet metal from its finished geometry to a flat blank without prior specification of a process. An algorithm is developed that allows specification of process constraints and great freedom in implementing un-forming strategies, leading to a prediction of the strain history of the un-forming process. Reversing the direction of this history, allows prediction of the stresses in the workpiece required to form the target part, by use of an appropriate material model. The external forces (boundary conditions) required to maintain equilibrium with this stress state can then be calculated, allowing an iterative refinement of the constraints on un-forming until a physically achievable process has been designed. The approach is validated against a known process, and used to demonstrate how several previously untried forming strategies could lead to specification of new process designs. In future work, the method could be extended to allow an iterative specification of tooling to create the required boundary conditions, and hence to complete automatic process designs.