Lipid membranes provide diverse and essential functions in our cells relating to transport, energy harvesting and signaling. This variety of functions is controlled by the molecular architecture, such as the presence of hydrating water, specific chemical compounds and microscopic structures, such as the local membrane curvature, as well as macroscopic properties, such as the fluidity of the membrane. To understand the chemistry of membranes, ideally one needs access to multiple length scales simultaneously, using probes that are noninvasive, label-free and membrane-interface specific. This dream is generally pursued by following either a top-down approach, introducing labels to real cell membranes or by following a bottom-up approach with well-controlled but simplified membrane monolayer or supported membrane models. This Perspective offers an alternative path that ultimately envisions bringing together both approaches. By using intermediate nano-, micro- and macroscale free-floating membrane systems in combination with novel nonlinear optical methods, one can advance the understanding of realistic membranes on a more fundamental level. Here, we describe recent advances in understanding membrane molecular structure, hydration, electrostatics and the effect of variable length scale, curvature and confinement for 3D nano- and microscale membrane systems such as lipid droplets and liposomes. We also describe an approach to image membrane hydration and membrane potentials in real time and space together with an application to neuroscience. In doing so, we consider the emerging role of interfacial transient structural heterogeneities that are apparent in both model membranes as well as in whole cells.