They should really look into using the Unreal's per-pixel shading methods that Arc System uses for the new Guilty Gear/Dragon Ball games, and do all their characters in 3D, then hand-animate the rigs per key frame like traditional cel animations.

This way they have all of their characters as 3D assets, a one-time investment, and they can setup and stage cinematic shots and do animations on the fly. It may be costly upfront to get the artists to design and rig the models, but after that, the costs dwindle greatly as you're only setting up backdrops and shots, and animating using pre-existing assets. The only serious costs would be in sound design and voice acting.