Midjourney debuts AI video tool to animate images into short clips

Midjourney has introduced its inaugural AI-driven video generation model, enabling users to convert static images into dynamic short clips. This innovative feature, available exclusively to paying subscribers on both Midjourney’s website and Discord, signifies a significant expansion of the platform's creative generative capabilities. Currently, users can produce five-second videos from images, whether they are created within the platform or uploaded directly. A newly integrated 'animate' button becomes accessible post-image rendering, leading users through a prompt-based animation process. The tool offers a default motion sequence but also provides a manual option for users who want to customize the movement of their images. Moreover, Midjourney allows users to extend their animations in four-second increments, reaching a maximum length of 21 seconds. It features two distinct motion modes—low and high—that provide varying levels of dynamism, enhancing how the subject or camera is animated. The pricing structure operates on a GPU time model, starting at $10 per month, which grants approximately 3.3 hours of fast GPU usage, equating to around 200 image generations. However, rendering video is considerably more resource-intensive, with each second of video consuming the GPU resources equivalent to a single image. “This is merely the beginning,” stated David Holz, founder of Midjourney, in a formal announcement. Holz expressed that future iterations of the tool might pave the way for real-time, open-world simulations, reflecting the company’s broader aspirations in immersive media creation. However, this launch occurs amid increasing legal scrutiny. Major entertainment companies, including Disney and Universal, have filed a lawsuit against Midjourney, claiming that the platform facilitates unauthorized reproduction of copyrighted materials. The lawsuit describes the platform as a "virtual vending machine" for derivative content and raises concerns about the legality of the training data utilized for the AI model. As the realm of AI-generated content evolves, experts predict that the tension between innovation and copyright law is just beginning to unfold.