Shape is one of the important characteristics of nonspherical particles. Herein, the two-dimensional shapes of several micrometer/millimeter-sized particles with a variety of geometrical forms are retrieved from simulated interferometric out-of-focus images using the Hybrid Input-Output algorithm. The particle concerned can be either a single particle or a complex particle with separate parts. The impact of the axial size of the three-dimensional particle on the retrieval of the two-dimensional shape (i.e., the projection of the particle on the image sensor) is analyzed, showing that an increase of the axial size increases the deviation of low frequencies in the interferometric out-of-focus image and eventually degrades the quality of the reconstructed shape. This study demonstrates the capability of the interferometric out-of-focus imaging technique on shape information retrieval for micrometer/millimeter-sized nonspherical particles.