Recent ultra-high precision observations of eclipsing binaries, especially data acquired by the Kepler satellite, have made accurate light curve modelling increasingly challenging but also more rewarding. In this contribution, we discuss low-amplitude signals in light curves that can now be used to derive physical information about eclipsing binaries but that were unaccessible before the Kepler era. A notable example is the detection of Doppler beaming, which leads to an increase in flux when a star moves towards the satellite and a decrease in flux when it moves away. Similarly, Rømer delays, or light travel time effects, also have to taken into account when modelling the supreme quality data that is now available. The detection of offsets between primary and secondary eclipse phases in binaries with extreme mass ratios, and the observation of Rømer delays in the signals of pulsators in binary stars, have allowed us to determine the orbits of several binaries without the need for spectroscopy. A third example of a small-scale effect that has to be taken into account when modelling specific binary systems, are lensing effects. A new binary light curve modelling code, PHOEBE 2.0, that takes all these effect into account is currently being developed.