This paper is concerned with the use of a spherical projection model to design optimal visual features for visual servoing. Here two special targets are considered: a sphere marked with two points on its surface and a sphere marked with a tangent vector to a point on its surface. For these targets we propose a new minimal set of six visual features which can be computed on classical perspective cameras. Using the new set, a classical control method is proved to be globally stable even in the presence of modeling error. In comparison with the previous set originally proposed for the second target, the new set draws a better camera trajectory. Finally, simulation and experimental results confirm the validity of the proposed theoretical results.