An image model as an iterative creative tool
Midjourney describes itself as a small, self-funded independent research lab focused on design and AI. Its product puts generative images in a workflow where creators supply prompts and references and keep refining a visual direction. [1][2]
The documentation exposes parameters for style references, stylization, image influence and model selection. These controls affect how a model interprets a request; their availability and behavior depend on the selected version. They give creators a vocabulary for steering generation without making a sampled image fully deterministic. [2]
Adding motion to an existing image
Midjourney released its V1 video model on June 18, 2025. Users could animate a generated or uploaded starting image, either letting the system choose motion or describing it manually. Low- and high-motion settings offered different trade-offs, and clips could be extended. [3]
The release began on the web and produced short clips. Its discussion of future interactive 3D worlds was a research ambition, separate from the image-to-video product that users could access that day. Keeping those states distinct makes the company's video contribution concrete without turning a roadmap into a released system. [3]
Sources
- Midjourney Magazine
Midjourney · retrieved Oct 23, 2026
- Parameter List
Midjourney · retrieved Oct 23, 2026
- Introducing Our V1 Video Model
Midjourney · Jun 18, 2025