Most of the exoplanets known today have been discovered by indirect techniques, based on the study of the host star radial velocity or photometric temporal variations. These detections allowed the study of the planet populations in the first 5–8 AU from the central stars and have provided precious information on the way planets form and evolve at such separations. Direct imaging on 8–10 m class telescopes allows the detection of giant planets at larger separations (currently typically more than 5–10 AU) complementing the indirect techniques. So far, only a few planets have been imaged around young stars, but each of them provides an opportunity for unique dedicated studies of their orbital, physical and atmospheric properties and sometimes also on the interaction with the ‘second-generation’, debris discs. These few detections already challenge formation theories. In this paper, I present the results of direct imaging surveys obtained so far, and what they already tell us about giant planet (GP) formation and evolution. Individual and emblematic cases are detailed; they illustrate what future instruments will routinely deliver for a much larger number of stars. I also point out the limitations of this approach, as well as the needs for further work in terms of planet formation modelling. I finally present the progress expected in direct imaging in the near future, thanks in particular to forthcoming planet imagers on 8–10 m class telescopes.