Midjourney Inc. introduced V1, a new AI model that can generate videos up to 21 seconds long from a single image.
Midjourney, launched in 2022 and now serving over 21 million users, previously focused on image generation. Its service operates on a subscription model and includes a gallery where users can store and view their creations. A new feature adds an “Animate” button below each image, enabling instant video generation using V1.
By default, V1 generates a five-second video clip. Users can extend the clip in four-second increments, up to four times, reaching the maximum duration of 21 seconds — slightly longer than Google’s Veo 3 and OpenAI’s Sora, which are capped at 20 seconds.
Users can choose automatic animation or guide the process with a text prompt. Two options control prompt interpretation: one for precise alignment with the input, and another for creative variation. Motion settings can also be adjusted. Low motion favors subtle camera and subject movement; high motion creates more dynamic scenes.
According to CEO David Holz, V1 is part of Midjourney’s broader plan to build models for interactive 3D simulations, which will require image, video, and real-time 3D generation capabilities.
Like other video generators, V1 uses diffusion techniques and includes a temporal module for frame consistency and logic to maintain correct frame order.
V1 follows the recent launch of V7, Midjourney’s latest image model, which improved speed and quality over previous versions.
