A classical problem in the Calculus of Variations is to minimize an integral functional in a certain class of functions. In this talk we present several simple examples which illustrate why and how non-convexity of the integrand can result in the non-existence of minimizers or the formation of small-scale oscillations.