OpenAI's next-generation AI video model — generate long, cinematic, physically realistic videos from text or image prompts with superior motion and scene