No, it doesn't perform any image gen. It orchestrates other models for asset gen + tts + stt and combines everything together