Consider incorporating machinema and v2v routes

A suggestion - add a machinima-style workflow where ai-planned scenes can be rendered in a real-time engine, then imported as rough shots for iteration and refinement. This would make it much faster to edit individual scenes in the middle of a sequence, generate rough cuts faster than pure text-to-video, and use engine renders as inputs for optional V2V refinement. It would be especially useful for long-form projects where shot-by-shot control, continuity, and fast re-rendering of selected scenes matter.