Next-generation high-energy batteries require rechargeable metal anodes, but hazardous dendrites tend to form during recharging, causing short-circuiting risks and capacity loss. In this presentation, we will exploit our understanding of lithium whiskers and dendrites to investigate the interfacial instabilities of Na metal anode and Zn metal anode, through the intimate combination of operando microscopy and novel transport models. Our results demonstrate the necessity to carefully avoid the transport-controlled scenario, in which fast-advancing dendritic growths would become inevitable to short the cell. Within the reaction-controlled regime, however, resolving the interfacial instability and metal whiskers leads to ideally smooth, non-porous, ingot-type Na metal plating, which enabled the anode-free Na metal full cells with a record-high capacity retention rate of 99.93% per cycle at 3C charge and discharge. Chronopotentiometry tests of this ideal metal electrode revealed an inverse correlation between the penetration overpotential and the pore size of the separator, which can be captured by a new model we named as the Young-Laplace critical overpotential. Following a similar principle, we investigated the growth of Zn dendrites in operando in our capillary cells. We discovered for the first time that seemingly negligible pore-wall charges, far away from the bulk electrolyte, can affect the morphology of Zn dendrites, which suggests that by depositing negative and positive charges on the pore walls, a bipolar separator not only can ensure uniform cation flux to improve the interfacial stability but also can stop localized dendrite penetration autonomously toward ideal safety.