For games you have little choice to reframe for higher sub d meshes, but recently with games I have seen much higher poly counts. Film they use higher res meshes because they are not real time but do use poxy meshes during animation stages. A good example of this is in the film Avatar , see the making of videos and your see they use quite med range meshes but in either case one thing stands to reason, at articulation point they must deform well.
As I mainly produce animation within C4D and render it out also in C4D I stick to Med res to hold the meshes own volume, but use displacement during render time. Close up shots id use higher res displacement, higher sub d, and further away shots I can get away with normal maps, but in all cases the silhouette/volume must be within the cage mesh. This all changes when you have weaker resources and multiple characters. you have to basically adapt and think on your toes. As I go solo and don't have access to huge render farms, I don't have tons of characters, and my animations are short I stick to that workflow.
Dan