One move per clip
AI video clips are short. Ask for one camera move and one action in each. To combine moves, cut between clips; don’t ask one clip to do both.
The moves
| Move | What the viewer feels | How to write it |
|---|---|---|
| Locked | Calm, watching | locked-off camera, no camera movement |
| Push-in | Attention, rising tension | slow push-in towards her face |
| Pull-out | A reveal, or loneliness | camera pulls back to reveal the empty hall |
| Pan | Following, discovering | camera pans left to follow him |
| Tilt | Scale | camera tilts up from his boots to his face |
| Orbit | Importance, a hero moment | slow orbit around the subject |
| Tracking | Energy, travelling with someone | camera tracks alongside as she runs |
| Rise | An opening or an ending | camera rises above the rooftops |
| Handheld | Urgency, a documentary feel | handheld camera, slight shake |
Write the motion prompt
[camera move and speed]. [what the subject does]. [what the world does: wind, rain, dust, light]. [what stays still].
Slow push-in. She lifts the torch and the flame flickers. Mist drifts across the floor. The statues behind her stay still.
Saying what stays still matters. Without it the model tends to invent movement you did not ask for.
In Midjourney video
Midjourney video starts from an image, which becomes the first frame. The text prompt is optional, so the still has to be right first.
- Length. Five seconds. You can extend by four seconds, up to four times, for 21 seconds.
- --motion low (the default): gentle movement. Best for slow push-ins and portraits.
- --motion high: bigger camera and subject movement, with more glitches.
- --raw: less added flair, so the clip follows your motion prompt more closely.
- --loop makes the last frame match the first. --end plus an image link sets the final frame yourself.
- Resolution. 480p on every plan; 720p on Standard and above.
- Look. Style references don’t apply to video. Lock the look in the still.
Midjourney’s own documentation lists no camera vocabulary and mentions no sound. The wording in the table above is plain film language; test each move on your own frame, and add sound in the edit.
When a move won’t work
- Cut the prompt down to one move and one action.
- Switch the motion setting: high for big moves, low for subtle ones.
- Compose the still for the move. Leave space in the direction of a pan, and room around the subject for a push-in.
- Generate the final composition as a second still and set it as the end frame.
- Do it in the edit. A slow scale-up on a locked shot reads as a push-in.
The same shot language carries over to other image-to-video tools. Each film breakdown names the tools it was made with.
Tools, prices and platform rules change often. Everything here was checked in November 2026; confirm on the tool’s own page before you rely on it.