A new AI system developed by researchers at Stanford University demonstrates the ability to generate highly realistic and coherent video from simple text descriptions. The model, named ‘DreamScene’, uses a novel diffusion architecture that builds scenes frame-by-frame while maintaining temporal consistency and physical plausibility. Initial tests show it can create short clips of animals, landscapes, and simple human actions with impressive detail. The researchers acknowledge current limitations in video length and complex scene generation but highlight the technology’s potential for film pre-visualization, game development, and educational content. The full paper and sample videos are available for review. Read the full article at https://technologyreview.com/2024/05/15/dreamscene-ai-video-generation.