Midjourney launches image-to-video AI model: All the details

Midjourney

Midjourney, known for its imaginative and high-quality image generation, has unveiled its first video generation model. In a new announcement, the company revealed that it’s releasing Version 1 of its video model, bringing image-to-video animation to its users. Midjourney says it’s building the essential blocks: static images, animated videos, 3D spatial understanding, and speed—all to culminate in fully immersive, dynamic virtual worlds.

Dubbed “Image-to-Video,” this feature lets users animate any image generated within Midjourney—or even one uploaded externally. With a simple click of the “Animate” button, users can bring their creations to life. Two modes are available:

•Automatic: The system generates motion prompts for you.

•Manual: You define exactly how the scene and objects should move.

You can also choose between high motion (lots of movement and camera shifts) and low motion (gentler, more ambient animation). Each animation starts at five seconds, with up to four extensions possible, bringing the total to 20 seconds.

The cost? Each video job will be priced at about eight times that of an image job, which translates to one second of video costing roughly the same as one upscaled image—a surprisingly affordable rate, and reportedly over 25 times cheaper than previous industry benchmarks, claims Midjourney.

Initially, the feature is web-only, with a “video relax mode” rolling out soon for Pro subscribers. As the company gauges usage and server load, pricing and access may evolve.

More than just a fun toy, Midjourney believes this tech has deeper creative and practical implications. Whether it’s animating old memories, building new cinematic worlds, or prototyping virtual experiences, this is just the beginning.

“We’re giving you the pieces now,” the company said, “and over the next year, we’ll start putting them all together.”