Structural studies in general, and crystallography in particular, have benefited and still do benefit dramatically from the use of synchrotron radiation. Lowemittance storage rings of the third generation provide focused beams down to the micrometre range that are sufficiently intense for the investigation of weakly scattering crystals down to the size of several micrometres. Even though the coherent fraction of these sources is below 1%, a number of new imaging techniques have been developed to exploit the partially coherent radiation. However, many techniques in nanoscience are limited by this rather small coherent fraction. On the one hand, this restriction limits the ability to study the structure and dynamics of non-crystalline materials by methods that depend on the coherence properties of the beam, like coherent diffractive imaging and X-ray correlation spectroscopy. On the other hand, the flux in an ultra-small diffraction-limited focus is limited as well for the same reason. Meanwhile, new storage rings with more advanced lattice designs are under construction or under consideration, which will have significantly smaller emittances. These sources are targeted towards the diffraction limit in the X-ray regime and will provide roughly one to two orders of magnitude higher spectral brightness and coherence. They will be especially suited to experiments exploiting the coherence properties of the beams and to ultra-small focal spot sizes in the regime of several nanometres. Although the length of individual X-ray pulses at a storage-ring source is of the order of 100 ps, which is sufficiently short to track structural changes of larger groups, faster processes as they occur during vision or photosynthesis, for example, are not accessible in all details under these conditions. Linear accelerator (linac) driven free-electron laser (FEL) sources with extremely short and intense pulses of very high coherence circumvent some of the limitations of present-day storage-ring sources. It has been demonstrated that their individual pulses are short enough to outrun radiation damage for single-pulse exposures. These ultra-short pulses also enable time-resolved studies 1000 times faster than at standard storage-ring sources. Developments are ongoing at various places for a totally new type of X-ray source combining a linac with a storage ring. These energy-recovery linacs promise to provide pulses almost as short as a FEL, with brilliances and multi-user capabilities comparable with a diffraction-limited storage ring. Altogether, these new X-ray source developments will provide smaller and more intense X-ray beams with a considerably higher coherent fraction, enabling a broad spectrum of new techniques for studying the structure of crystalline and non-crystalline states of matter at atomic length scales. In addition, the short X-ray pulses of FELs will enable the study of fast atomic dynamics and non-equilibrium states of matter.