Google's paper on "Tune-A-Video: One-Shot Tuning of Image Diffusion Models for Text-to-Video Generation" - https://tuneavideo.github.io/ shows generating simple YouTube videos with just text prompts is going to be a reality soon.