Generate videos from reference images while keeping characters, objects, and scene identity consistent using Multi-Entity Consistency. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.