Midjourney has released its first AI video model, allowing users to animate still images into short video clips. The company announced the launch of its V1 Video Model to nearly 20 million users worldwide.
The new feature works through an “Image-to-Video” workflow. Users create or upload an image, then press an “Animate” button to generate movement. The system offers automatic animation that creates motion prompts independently, or manual controls where users describe how elements should move.
Two motion settings are available. Low motion works best for ambient scenes with subtle movement, while high motion creates more dynamic animations of both subjects and camera angles. Each video job produces four 5-second clips, which users can extend up to 20 seconds total.
Midjourney founder David Holz described the video model as a stepping stone toward more ambitious goals. The company aims to eventually create real-time interactive simulations where users can navigate through AI-generated environments.
Pricing and competition
The video feature costs approximately eight times more than generating a still image. Since each job produces 20 seconds of content, the cost per second equals roughly one image generation. Midjourney states this pricing is 25 times cheaper than previous market offerings.
The launch positions Midjourney against established AI video competitors including OpenAI’s Sora, Google’s Veo 3, and Runway’s Gen-4. Unlike Google’s Veo 3, Midjourney’s system does not generate audio tracks, requiring users to add sound separately.
Additional source: VentureBeat