Although molecular dynamics (MD) simulations are a widely used tool in (bio)sciences, they often actually fail to provide a veritable description (e.g. matching NMR data within experimental accuracy) of the conformations and the dynamics of the molecules simulated—a crucial prerequisite for using MD to intuitively interpret such high-resolution experiments. These failures can largely be attributed to the quality of the underlying MD models (force fields), improvement of which is hindered by the lack of comprehensive comparison to experiments, and outdated approaches, such as hand-tuning parameters. The High-fidelity Biomolecular Modelling Group employs open-science approaches to test and data-driven optimization methods to improve biomolecular force fields. We strive to make such force-field-optimized MD a tool for Integrative Structural Biology: Combining data from multiple experimental techniques to generate physics-based high-fidelity structural models with realistic structure and dynamics. I argue that such approach could eventually evolve MD into a bona fide Ångström-resolution microscope.