Midjourney video prompts work best as motion instructions for an image you already have: describe what visibly moves, how it moves, and—if it matters—how the camera moves. Midjourney’s official guide describes turning a single image into a five-second video; a text prompt is optional. For manual animation, you can edit the motion prompt before generating.
Production API decision: The official materials reviewed here explain using Midjourney through its website and Discord, but do not document a production video API, endpoint, or request format. Midjourney announced in July 2025 that it was investigating an Enterprise API; that announcement is not API documentation or confirmation that a video API is available. If an API is a firm requirement, treat Midjourney availability as unconfirmed until Midjourney provides current official API details.
A practical prompt pattern
Use this as a writing aid, not a guaranteed formula:
Subject + visible action + pace or manner + optional secondary movement + optional camera movement
For example, if your starting image shows a woman by a window:
The woman slowly turns toward the window. Her scarf lifts gently in the breeze as the camera tracks left.
The still image provides the opening visual; the text tells Midjourney what you want to happen next. Keep the main action clear, and add secondary movement or camera direction only when it helps define the shot. This pattern adapts Midjourney’s manual-animation workflow, which lets you describe how an image should move and how the scene should develop.
Steps to make a video
- Choose or upload a starting image. Midjourney’s Video guide describes selecting an image from your uploads or uploading a new one, then dragging it into the Animate section.
- Choose manual animation if you want to edit the motion prompt; automatic animation generates a motion prompt for you.
- Describe the action, not another still image. For instance, “the cyclist pedals slowly past the trees” gives a motion instruction; “a cinematic cyclist among trees” mainly describes the scene.
- Choose a motion setting. Use
--motion lowfor subtler movement or--motion highwhen you want more movement. Midjourney lists both as video parameters; its V1 launch announcement cautions that high motion can also lead to visual mistakes. - Generate and assess the result. If the action is unclear, simplify the instruction or change one element—such as the motion setting—on the next attempt.
The starting image establishes the first frame. If a subject or object must be visible from the outset, include it in that image. Midjourney’s guide also says several image-reference types—including Style Reference and Image Prompt—aren’t compatible with video generations, so don’t assume every image workflow transfers to animation.
What to know before choosing Midjourney video
- It is an image-to-video workflow: Midjourney describes animating a still image, including images uploaded from outside Midjourney.
- Five seconds is the documented initial video length—not a confirmed maximum. The official Video guide describes a five-second video. In its June 18, 2025 V1 launch announcement, Midjourney said videos could be extended roughly four seconds at a time, up to four times. That announcement documents the launch-era extension behavior; the current Video guide does not specify whether that extension limit remains in effect. Don’t treat the launch figure as a confirmed current maximum for production planning.
- Motion controls are available: documented video parameters include
--motion lowand--motion high. These are Midjourney prompt parameters, not evidence of API fields. - API suitability remains unconfirmed by the materials reviewed: Midjourney’s Enterprise API announcement described an investigation, not API documentation. Don’t treat unofficial wrappers or browser automation as an approved production integration.
Bottom line: Midjourney’s documented video workflow is worth exploring when you can provide a starting image and generate through its supported interface. If your requirement is to submit video jobs programmatically, obtain current official API documentation and access terms first; the prompt flags above do not establish API access.