Emu Video is a cutting-edge tool for text-to-video generation, utilizing diffusion models to streamline the process into two efficient steps. By first generating an image based on a text prompt and then creating a video using the prompt and the generated image, Emu Video stands out for its effectiveness and simplicity. This innovative approach allows for the training of high-quality video generation models with just two diffusion models, producing impressive 512px, 4-second videos at 16fps. In comparison to other text-to-video generation models, Emu Video excels in both quality and faithfulness to the prompt, as confirmed by human raters. With state-of-the-art results, Emu Video outperforms prominent models like Make-a-Video (MAV), Imagen-Video (Imagen), and others across various metrics. Developed by a team of dedicated authors and supported by numerous collaborators, Emu Video represents a significant advancement in the field of text-to-video generation.