When mixing or post-producing audio scenes, it is often desirable to modify the spatial characteristics of the sound field. In this article we present a method for controlling the angular spread of a sound source or a sound scene encoded in Ambisonics. To do this, we develop a mathematical approach allowing the norm of the energy vector to be continuously varied, by simulating fractional Ambisonics orders. We derive this approach for the 3D and 2D cases, then present the objective results in particular in the form of equivalent directivity patterns. The technique proposed here corrects and completes two previous publications on related topics [1, 2].