In autonomous driving, the detection of all relevant road users is a key requirement for the sensors. In order to improve the detection capability and to increase the signal to noise ratio, transmit (Tx) beamforming can be applied. In this paper, the possibility of adaptive Tx beamforming in a multiple-input multiple-output radar imaging system using state-of-the-art radar transceivers is analyzed both in theory and by measurements. The imaging performance is shown for an exemplary intersection scenario, using an antenna system with a high virtual channel count exceeding the state-of-the-art. For the first time, high-resolution radar images are shown both with and without Tx beamforming. To enable the detection of a pedestrian in the proximity of a strong target, such as a car, the target response and side lobes of the strong target can be weakened by more than 11 dB, even at a large incident angle.