Parallel-hole SPECT collimators have traditionally been manufactured by stacking sheets of lead foil or by casting. These techniques significantly restrict our options in terms of collimator geometry. However, recent developments in metal additive manufacturing are making novel collimator designs possible, giving rise to new opportunities in SPECT imaging. In this paper we propose an innovative type of collimator for stationary SPECT, using parallel-holes whose collimation direction depends on their axial position. Its main advantage compared to current stationary SPECT systems (which are based on pinholes) is that, using only axial bed translations, we can achieve complete angular sampling of an increased portion of the transaxial area of the collimator bore. This allows the system to be much more compact than current stationary SPECT systems that image objects of the same size. We describe three possible designs, for full-body, brain and small-animal imaging, respectively, and test their feasibility using simulations. The system modeling method is validated against realistic Monte Carlo simulations, and then used in the evaluation of the systems’ performances and reconstructions. The simulations show that the system is able to reconstruct objects occupying the predicted field of view (75% of the transaxial area of the bore) without sampling artifacts. In particular, we perform reconstructions from noisy projection data obtained for an activity and scanning time similar to standard protocols for the three applications, and the resulting images indicate the possibility of using the proposed systems in practice.