Improved data reduction techniques for three-dimensional (3D) data cubes obtained from Fabry-Perot integral field spectroscopy are presented. They provide accurate sky emission subtraction and adaptive spatial binning and smoothing. They help avoiding the effect analogous to the beam smearing, seen in HI radio data, when strong smoothing is applied to 3D data in order to get the most extended signal coverage. The data reduction techniques presented in this paper allow one to get the best of both worlds: high spatial resolution in high signal-to-noise regions and large spatial coverage in low signal-to-noise regions.