Midjourney Launches AI Video Generation Model

Midjourney has released the first version of its video generation model to the public. This innovative tool currently enables users to generate short videos based on images uploaded or created on the platform. Midjourney has indicated plans to expand its capabilities in the future.

Features of the New Video Generation Tool

After creating an image using Midjourney, users will see a new “animate” button. By pressing this button, they can create a 5-second video clip based on a text prompt. Additionally, users have the option to add an uploaded image as a “starting frame” for the video. The tool generates a generic prompt that simply animates the image by default. However, a “manual” button allows users to specify how they want the motion to appear.

Users can extend the video duration by four seconds up to four times, resulting in a total video length of 21 seconds. The tool also offers high and low motion settings, which determine whether both the subject and camera move or just the subject.

Access and Subscription Details

Currently, Midjourney’s AI video generator is accessible only via the web and through the startup’s Discord server. A subscription is required to use the service, starting at $10 per month for 3.3 hours of “fast” GPU time, which equates to approximately 200 image generations. The startup has stated that it will charge “about 8x more for a video job than an image job,” which translates to around “one image worth of cost” per second of video.

Legal Concerns and Copyright Issues

Midjourney is currently the subject of a lawsuit from Disney and Universal, which has raised concerns regarding the potential implications of launching a video generator. Critics argue that Midjourney functions as a “virtual vending machine,” producing endless unauthorized copies of copyrighted works from companies like Disney and Universal. The ongoing development of the video generation model has been first announced in January, with Disney and Universal asserting that the training process likely infringes upon their copyrighted materials.

Future Developments

In a post announcing the new generator, Midjourney founder David Holz described this initial version as merely “a stepping stone.” The startup aims to develop “models capable of real-time open-world simulations.” Other companies, including Google, OpenAI, and Meta, have also introduced AI video generators that can create videos based on text prompts.