Since depth is based on our own assumptions on any particular part of modeling (Lighting/shadow also work)

Is it viable to have only a side picture of what you wish to create? I've seen a tutorial on someone mainly using a side view to create a human looking head, but I am curious on peoples views of this approach.