Neutron-deficient 177−185Hg isotopes were studied using in-source laser resonance-ionization spectroscopy at the CERN-ISOLDE radioactive ion-beam facility in an experiment combining different detection methods tailored to the studied isotopes. These include either α-decay tagging or multireflection time-of-flight gating for isotope identification. The endpoint of the odd-even nuclear shape staggering in mercury was observed directly by measuring for the first time the isotope shifts and hyperfine structures of 177−180Hg. Changes in the mean-square charge radii for all mentioned isotopes, magnetic dipole, and electric quadrupole moments of the odd-A isotopes and arguments in favor of I=7/2 spin assignment for 177,179Hg were deduced. Experimental results are compared with density functional theory (DFT) and Monte Carlo shell model (MCSM) calculations. DFT calculations using Skyrme parametrizations predict a jump in the charge radius around the neutron N=104 midshell, with an odd-even staggering pattern related to the coexistence of nearly degenerate oblate and prolate minima. This near-degeneracy is highly sensitive to many aspects of the effective interaction, a fact that renders perfect agreement with experiments out of reach for current functionals. Despite this inherent difficulty, the SLy5s1 and a modified UNEDF1SO parametrization predict a qualitatively correct staggering that is off by two neutron numbers. MCSM calculations of states with the experimental spins and parities show good agreement for both electromagnetic moments and the observed charge radii. A clear mechanism for the origin of shape staggering within this context is identified: a substantial change in occupancy of the proton πh9/2 and neutron νi13/2 orbitals.