The manufacturing freedom of additive manufacturing, and design freedom of topology optimization have proven to be a fruitful combination in creating and manufacturing designs. While research into applications of metal additive manufacturing has advanced from rapid prototyping to end-use components, metal additive manufacturing of compliant mechanisms has only been researched to a limited amount. This is despite the fact that additive manufacturing provides opportunities to manufacture highly advanced, monolithic compliant mechanism considered impossible before. Important considerations regarding geometry during additive manufacturing of compliant mechanisms are what minimum length scale and angle between overhanging features and build plate (the overhang angle) need to be imposed on the design. These considerations determine whether a compliant mechanism is successfully built by additive manufacturing and can ensure that designs are self-supporting. Since these constraints are most relevant in fragile hinge regions of compliant mechanisms where support material removal is unwanted, self-supporting, cross-axis flexural pivots are manufactured using selective laser melting. It is found manufacturability in additive manufacturing does not necessarily guarantee a predictable stiffness and yield strength of these flexural pivots. That is why it is recommended to print self-supporting regions of compliant mechanisms with a minimum length scale 25% greater than prescribed by the manufacturing process used and an overhang angle of at least 5 degrees greater than the prescribed minimum overhang angle. Next, these recommendations are implemented in compliant mechanism design using topology optimization, at locations where these recommendations are most relevant. This is done by redesigning hinge regions of compliant mechanisms, using static condensation to reduce the order of the problem. While adding only 5% extra computation time, small regions of a design can be redesigned successfully. To use the potential of this method fully, an extension to this method is introduced, which allows to redesign domains in more detail. Multi-point constraints are used to couple the original and finer redesign mesh. For a finer mesh, more computation time is needed, but as the design space is enlarged, other, more optimal solutions are found. Next, constraints on overhang angle, minimum length scale and stress are implemented, since these constraints are proven to be meaningful in hinge regions by the experiments with the cross-axis flexural pivots. Implementation is done successfully, and it is seen compliant mechanisms can be designed at relatively low additional computational cost and with minimal loss of mechanism performance. Thus, using this approach, redesigned compliant mechanisms can be manufactured with locally self-supporting hinges and predictable kinematics and stress concentrations can be ensured.