Midjourney Launches AI Video Generation Model

Midjourney has unveiled its initial version of an AI video generation model, enabling users to produce short animated clips from images within the platform. Accessible via the web and Midjourney’s Discord server, this feature currently requires a paid subscription.

Overview of the Video Generation Features

This first release allows users to create five-second clips from images that they either generate or upload. Once an image is created, an “animate” button appears, guiding users through an animation process based on prompts. By default, the system incorporates motion using a standard prompt, while a manual option offers custom movement descriptions. Users have the ability to provide an initial image to steer the animation process.

Expanding Animation Length

Midjourney provides an option to extend animations in four-second increments, allowing users to do this up to four times, resulting in a maximum video duration of 21 seconds. The platform features both high and low motion modes, which enable users to control the animation of the subject, camera, or both.

Pricing Structure

The subscription pricing is linked to GPU time, commencing at $10 per month for 3.3 hours of rapid GPU usage—approximately equivalent to 200 image generations. For video content, Midjourney estimates that producing video will be around eight times costlier than generating a single image, with an associated cost of one image per second of video.

Introducing their V1 Video Model, Midjourney promotes its offering as fun, easy, and visually appealing. The subscription is now available for $10 per month, marking it as the first video model accessible to a wide audience.

Midjourney’s founder, David Holz, remarked that this feature is merely a starting point. He expressed the company’s ambition to develop more advanced models in the future that could facilitate real-time open-world simulations.

Legal Challenges

The release of this feature occurs amidst ongoing legal issues, as Midjourney is currently facing a lawsuit from Disney and Universal Studios. These companies have raised concerns regarding the implications of Midjourney’s video generation capabilities. The lawsuit characterises Midjourney as a “virtual vending machine” that reproduces copyrighted works without permission, specifically highlighting the video generator as a point of concern. The studios contend that the model’s training likely contravenes their intellectual property rights.

Midjourney joins a growing roster of technology firms, including OpenAI, Google, and Meta, that are navigating the AI video generation landscape. Each company has developed tools that transform text prompts into video, as the race to innovate content creation tools intensifies.