The plasma mirror (PM) is an ideal model system to study relativistic optics; its geometry is simple, and its dynamics is rich and non-linear. Emphasized by high-order harmonic generation (HHG), relativistic PMs are a promising next-generation EUV source, unbounded in brightness and band-width. The applicability of these sources, however, is impeded at present because of the stringent requirement on laser intensity and temporal contrast. To-date, PM-HHG at the relativistic regime are only generated using post-compression contrast enhancement, commonly in the form of a PM-optical-switch. The complexity and low efficiency of this approach impose even stronger requirement on laser peak-power. I will present our progress towards PM-HHG in the relativistic regime using the newly commissioned 20 TW laser system at Tel-Aviv University. The laser's architecture is based on Picosecond Optical Parametric Chirped Pulse Amplification (Ps-OPCPA) for most of the system gain, followed by a traditional Ti:Sapphire power amplifier. In Ps-OPCPA the seed pulse is amplified in a picosecond window, enhancing contrast and eliminating pre-pulses associated with the use of classic regenerative amplifiers. These result in temporal laser contrast better than 1010 on 50 picosecond time scale. Owing to this pristine contrast, we demonstrated PM-HHG directly without post-compression contrast enhancement. I will present our method for controlling the spatial phase properties of the harmonics by tailoring the focused laser intensity profile. This method is based on a beam shaping technique which employs a two-optical-paths mirror that adds a half-cycle phase to the center of the beam. A variable aperture controls the fraction of laser energy outside the phase-shifted region. By changing the aperture diameter, the focal spot profile can be manipulated from gaussian to flat-top to donut shape. Owing to strong correlation between laser intensity and HHG phase, the tailored intensity profile is mapped into the HHG spatial phase profile, changing the far field properties of the EUV beam. Our aim is to minimize the angular divergence of the generated harmonics, forming the next step towards making these sources applicable. Preliminary results towards this goal will be presented.