Abstract: A challenge in extragalactic astronomy is that we cannot watch what happens to galaxies before and after they are observed. In particular, it remains debated whether galaxy mergers or internal processes drive supermassive black hole growth, trigger luminous starbursts, and shape the population of galaxies we see today. However, given increasingly available computer resources, it is now possible to predict how galaxies might evolve according to a huge variety of observations. With hydrodynamical simulations followed by dust radiative transfer, I examine the formation of elliptical galaxies through three putative phases: dust-obscured starburst, transition object, and red spheroid. I derive an idealized JWST-accessible mid-infrared diagnostic using model spectra from simulations of merger-induced starbursts. I use similar models to reconcile the numbers of optically selected post-starburst galaxies with independent estimates of the galaxy merger rate. To conclude, I describe efforts to build a aEurooemock observatoryaEuro from large-volume cosmological hydrodynamical simulations, with which observations of many types can be brought to bear to constrain the physics of galaxy formation.