A new AI model demonstrates significant progress in generating realistic and coherent video sequences from simple text prompts. The system, developed by researchers at a leading technology institute, uses a novel diffusion-based architecture that processes video frames in both spatial and temporal dimensions. This approach allows for improved consistency of objects and motion over time compared to previous methods. Initial results show the model can create short clips depicting complex scenes with multiple moving elements. While the output resolution and duration are currently limited, the researchers highlight this as a key step toward more general-purpose video synthesis tools. The team has released a technical paper and examples of the generated videos for peer review. Read the full article for detailed technical analysis and sample videos at https://technologyreview.com/2024/05/15/ai-video-generation-advance.