I take the opportunity provided by this newly established workshop to review some of our more unconventional epitaxy work, all centering around selective area growth. While traditional planar layer growth certainly remains the most developed and important process, the introduction of masked areas or volumes can provide structures, materials, and material combinations not accessible otherwise. However, experimentally, the introduction of masked epitaxial growth results in many challenges, including selectivity, precursor transport, growth yield, material defects, and surface energies to name a few. I will start by exemplify some these points using results on selective growth of III-Vs within oxide masks and their structural characterization. While the successful implementation is more resource intensive, it can result in III-V quantum well structures integrated on Si  or lead to a precise tunability over the resulting crystal phase . Beyond that, and instead of using hollow masks templates for selective filling by epitaxial growth, we have recently explored the loading of masks with a group-III element before growth . In contrast to typical growth were group-III and group-IV elements are delivered to form a compound semiconductor, the synthesis of the semiconductors is here obtained solely by group-IV exposure. Indium was deposited within the mask before growth and exposed to an As or Sb flux in the reactor to form InAs and InSb within the mask templates. Finally, we have investigated an extension of the typical selective growth masks by the introduction of refractory metals and metal-nitrides, a family of metallic elements that are highly resistant to heat. More specifically, we have embedded TiN within a dielectric oxide template structure . The introduction of a conductor within the template allows to form an in-situ semiconductor-metal junctions during epitaxial growth. Here, the challenge was, apart from fabrication, to maintain nucleation and growth within the template, while suppressing nucleation directly on the exposed TiN surface. In summary, I hope this brief overview will stimulate further explorative work using MOVPE.