After a series of tests and comparisons we ultimately decided to go with a 3D approach for the character generation, as it would allow us to give the image more depth, especially as shadows travel across the uneven surface of the background, and offer us possibilities with lighting that a 2D approach would not give us.