Relighting, and by consequence, inverse rendering, have been widely explored classes of problems in computer vision and computer graphics, but yet the methods used in recent literature remain divided and, for the most part, particular for a class of objects [1, 2, 3]. In this work, we explore and research relighting methods specific for human shapes in a three-dimensional setting, and take further steps to increase the realism of a motion transfer process, which is to synthesize new videos of people in a different context where they were initially recorded, taking into account 3D shape, appearance, and motion constraints. We start by taking a precomputed radiance transfer approach, which uses spherical harmonics lighting to relight the image of the human model which is going through a motion transfer process. Three approaches are used by the authors for relighting the human shape, namely, a two-dimensional image-domain renderer, a custom OpenGL renderer and a Blender script. Finally, we conclude that relighting in 3D from inferred spherical harmonics coefficients is a feasible task and we plan on exploring it further in the next months.