Midjourney Video

Midjourney's image-to-video: click Animate on any Midjourney still for 5-second clips (extendable to 20 s), gorgeous style consistency, no audio, Pro plan and up.

What Midjourney Video is

Midjourney's video model does one thing and does it beautifully: you make a still in Midjourney, press Animate, and get a 5-second clip that looks exactly like the still — same grade, same brushwork, same face — now moving. You can extend it four times to reach 20 seconds. Motion comes in low (subtle, camera-led) or high (subject moves a lot, sometimes too much), and you can let the model write the motion prompt or write your own.

What it does not do: text-to-video, sound, 1080p, or photoreal motion. And it only animates Midjourney's own images. It is a wonderful finishing touch for illustrators on a Pro or Mega plan, and a dead end for anyone who needs a talking clip or a product ad with sound.

How Midjourney reads a motion prompt

Manual mode wants a short motion sentence: one camera move, one subject action, one environmental detail. The community's best results read like "Slow dolly forward. Woman turns her head toward camera. Dust motes drift in sunlight." That exact style works on Runway Gen-4.5 image-to-video — and Runway follows it more precisely.

Slow dolly forward. The woman turns her head toward the camera. Dust motes drift in the sunlight. Nothing else moves.

Midjourney Video vs. the Lab's image-to-video

Animate any still, step by step (Lab version)

1. Make the still: in the image studio (Flux 2 Pro for painterly looks, Ideogram 4 for graphic and typographic stills, Nano Banana Pro for photoreal).
2. Open the video studio, pick Runway → Gen-4.5, upload the still as the start image.
3. Write only the motion — one camera move, one subject action.
4. Generate at 5 seconds (96 credits). Extend by generating the last frame forward if you need more.
5. Need sound? Use Grok Imagine Video or Kling 3.0 instead of Runway; same still, add an audio line.

Handheld shake, medium shot. The detective lights a cigarette against the wind; smoke tears away to the right.

12 Midjourney-style motion prompts, pointed at the right Lab model

Subtle motion (Runway Gen-4.5)

Static shot. The candle flame flickers. Shadows stretch and shrink across the wall behind.

Slow push-in. Snow falls past the window; the reading lamp glows steady; the cat's tail moves once.

Gentle parallax: foreground branches sway, the mountain behind stays still, mist drifts across the lake.

The astronaut's visor reflects passing stars; her breath fogs the glass once; the camera holds.

Stylized worlds with sound (Grok Imagine Video)

Animate the reference illustration: the paper city's windows light up one by one at dusk; a tiny tram crosses the bridge. Sound: a distant bell, wind.

Animate the reference: the watercolor fox blinks and turns its head; leaves drift down. Sound: forest ambience, a soft rustle.

Animate the reference poster: the neon title flickers on; rain streaks the glass; a silhouette passes behind. Sound: rain, neon buzz.

Characters that speak (Kling 3.0)

Start frame provided: the knight in the painting lowers his visor, looks at the camera and says: "Not today." Slow push-in. Sound: armor creak, wind.

Start frame provided: the anime girl on the rooftop turns, smiles and says: "You came." Cherry petals drift. Sound: wind, petals, a soft piano note.

High-motion pieces (Wan 3.0, cheap)

Animate the reference: the crowd in the stadium illustration rises in a wave; flags whip; confetti falls. Sound: roar.

Animate the reference: storm clouds boil over the desert, lightning forks twice, sand streams across the dunes. Sound: thunder, wind.

Animate the reference: the koi in the ink painting swim in a slow circle, ripples spread, ink blooms. Sound: water.

Automatic vs. manual motion, and when each wins

Midjourney offers two ways to animate. Automatic reads the still and invents a sensible motion prompt — a slow push, a breeze, a blink. It is the right choice for landscapes, interiors and abstract pieces, because the model already knows what usually moves in that kind of picture. Manual is you writing the motion sentence yourself, and it is the only way to get a specific beat: a head turn toward camera, a door opening, a lamp switching on.

The Lab's models are all manual, and that is an advantage once you know the sentence pattern: camera move · subject action · environmental detail. Runway Gen-4.5 follows that sentence with the most discipline; Grok Imagine adds sound; Kling 3.0 adds dialogue and, with a second image as the end frame, tells you exactly where the shot lands — something Midjourney cannot do at all.

Extension works differently too. Midjourney extends a clip by four more 5-second passes, each drifting a little from the original look. In the Lab you extend by taking the last frame of a clip as the start frame of the next; because the frame is a real image, the style does not drift, and you can change the model between segments — a quiet Runway opener into a talking Kling close-up.

Keeping the look: the two rules

Midjourney's magic is style lock; you can have the same lock anywhere by following two rules. One: in image-to-video, describe motion only — the moment you re-describe the picture, the model repaints it. Two: keep motion small for painted and illustrated stills; high motion breaks brushwork. When you need big movement, generate three 5-second clips with small moves and cut them, rather than one wild clip.

Gouache illustration of a lighthouse on a cliff at dusk, thick visible brushstrokes, warm window light, 16:9.

Slow push-in. The lighthouse beam sweeps once across the water. Waves move gently. Brushstrokes stay still.

Retro concert poster, bold condensed type reading "NIGHT SHIFT", teal and orange, halftone texture, 3:4.

Animate the reference poster: the halftone shimmers, the title pulses once with a light flicker. Sound: a vinyl crackle, a bass hit.

Mistakes that waste money (on Midjourney and in the Lab)

• High motion on faces. Warps. Use low, or generate the face still on Nano Banana Pro and animate on Runway.
• Describing the scene again in image-to-video. Motion only.
• Expecting audio from Midjourney or Runway. Use Grok or Kling for sound.
• Paying a Pro plan for 120 seconds a month. That is 24 Midjourney clips; the same 24 clips on Runway cost about $23 in the Lab, and the credits do not expire.

FAQ

Related tools

Runway Gen-4.5 · Grok Imagine Video 1.5 · Kling 3.0 · Wan 3.0 · Flux 2 Pro and Ideogram 4 for the stills. Same studio, one wallet.