How to Control Midjourney Video Motion, Loops, and End Frames

Once you know how to turn Midjourney images into videos, the next challenge is controlling how the motion looks. By default, Midjourney decides what moves and how much. But with four parameters you can specify motion intensity, create seamless loops, define explicit end states, and control how smoothly frames blend together.

This guide covers each video control parameter in isolation, then shows how to combine them for specific use cases. Every section includes copyable prompt examples so you can test each technique directly.

The four video control parameters

• --motion low or --motion high: controls how much movement appears in the clip
• --loop: makes the video end where it began for seamless repetition
• --end [image_url]: defines a target frame the animation moves toward
• --bs [value]: blend strength controlling how aggressively frames interpolate

All four require --video to be present. They modify video behavior but do not trigger video generation on their own.

Motion intensity with --motion

The --motion parameter accepts two values: low and high. It determines how dramatic the movement in your clip will be. Low motion creates gentle environmental effects: slight camera drift, subtle particle movement, slow atmospheric changes. High motion creates visible camera sweeps, subject movement, and larger-scale transformations.

When to use --motion low

• Portraits or face-focused images where distortion is a risk
• Ambient backgrounds for websites or presentations
• Product scenes where you want subtle life without distracting movement
• Architectural shots where camera shake would look unnatural
• Any scene with fine details you want preserved

[upscaled_image_url] --video --motion low

When to use --motion high

• Action scenes where dramatic movement fits the subject
• Nature scenes with weather, water, or wildlife in motion
• Abstract art where distortion adds interest
• Wide landscapes where camera sweeps create cinematic drama
• Scenes with large open areas and few fine details

[upscaled_image_url] --video --motion high

Comparing low and high on the same source

A good test is to generate the same source image twice: once with --motion low and once with --motion high. This shows you exactly how much range the parameter provides for a given scene composition.

Source: a foggy mountain pass with a lone hiker, aerial perspective --ar 16:9
Variant A: [image_url] --video --motion low
Variant B: [image_url] --video --motion high

Seamless loops with --loop

The --loop parameter tells Midjourney to make the last frame of the video match the first, creating a clip that repeats without a visible seam. This is essential for background animations, social media posts, digital signage, and ambient displays.

Not all subjects loop equally well. Repetitive patterns (waves, rain, fire, clouds) loop cleanly. Subjects with strong directional movement (a person walking forward, a car driving past) are harder to loop without a visible reset.

[image_url] --video --loop

[image_url] --video --motion low --loop

Best subjects for looping

• Water: ocean waves, rain, waterfalls, ripples in a pond
• Fire and smoke: campfires, candles, incense, industrial chimneys
• Weather: snow falling, clouds drifting, fog rolling
• Abstract: geometric patterns, particle systems, color gradients
• Foliage: trees swaying, grass blowing, leaves falling in a cycle

Source: a campfire in a dark forest clearing, ember particles rising, warm orange glow --ar 16:9
Video: [image_url] --video --motion low --loop

End frames with --end

The --end parameter accepts a second image URL as the target destination for the animation. Instead of Midjourney deciding where the motion goes, you specify both the starting state and the ending state. The system then generates frames that transition between the two.

This is useful for before/after transitions, time-of-day changes, seasonal shifts, or any scenario where you want to show a transformation from one state to another.

[start_image_url] --video --end [end_image_url]

Tips for effective end frames

• Use the same subject and composition in both frames. Only change the element you want to animate
• Keep aspect ratios identical between start and end images
• Avoid radical subject changes. The system interpolates, it does not teleport
• Pair with --bs to control how smoothly or abruptly the transition happens
• Generate both frames from similar prompts to maintain structural consistency

Example: day-to-night transition

Start: a mountain village in bright afternoon sunlight, blue sky, warm tones --ar 16:9
End: the same mountain village at night, warm window lights, starry sky --ar 16:9
Video: [start_url] --video --end [end_url] --bs 50

Blend strength with --bs

The --bs parameter controls how aggressively the animation interpolates between frames. It accepts a numeric value. Lower values produce subtle, gradual transitions. Higher values push the system toward more dramatic frame-to-frame changes. This parameter is most impactful when used alongside --end, where it determines how quickly the transition reaches the target state.

[start_url] --video --end [end_url] --bs 30

[start_url] --video --end [end_url] --bs 80

A --bs of 30 will produce a slow, gentle morph. A --bs of 80 pushes the transition to happen quickly and visibly. Experiment with values between 20 and 80 to find the sweet spot for your specific content.

Combining parameters for specific outcomes

Ambient background loop (low motion, looped)

Source: abstract gradient with floating particles, soft purple and blue tones --ar 16:9
Video: [image_url] --video --motion low --loop

Dramatic transition with defined endpoints

Start: a flower bud covered in morning dew, macro lens, soft backlight --ar 16:9
End: the same flower fully open, warm golden light, macro lens --ar 16:9
Video: [start_url] --video --end [end_url] --bs 60 --motion high

Subtle product hero animation

Source: a perfume bottle on a marble surface, studio lighting, soft caustics on glass --ar 4:5
Video: [image_url] --video --motion low

The --raw parameter for video

The --raw parameter reduces Midjourney's automatic beautification and artistic interpretation. When applied to video, it produces movement that is more literal and less stylized. This can be useful when you want the animation to feel documentary rather than cinematic, or when automatic enhancements are adding unwanted flourishes to the motion.

[image_url] --video --raw --motion low

Troubleshooting motion control

• Video feels too static even with --motion high: the source image may lack depth or implied movement. Regenerate the source with more environmental activity
• Loop has a visible jump: the subject may have strong directional movement. Switch to symmetric subjects like fire, water, or particles
• --end transition looks unnatural: the start and end images are too different. Make both from similar prompts with one changed element
• --bs too high causes flickering: reduce blend strength to 40 or below for smoother results
• Faces distort in motion: always use --motion low for portraits. High motion is unreliable on fine facial features

Planning motion at the image stage

The best video results come from images that were composed with motion in mind before any --video parameter was applied. When generating source images, think about what direction the camera could move, what elements have natural kinetic energy, and whether the scene has enough depth for parallax effects. A flat composition with everything in one plane gives the motion engine very little to work with.

• Add foreground and background layers to give the system parallax opportunities
• Include elements with natural movement: water, fabric, smoke, foliage
• Frame the subject with space for the camera to drift into
• Avoid perfectly centered, symmetrical compositions which leave no direction for motion to travel

Limitations to keep in mind

• Image Prompts, Style Reference (--sref), and Omni Reference are not compatible with --video
• Personalization (--p) does not apply to video generations
• Video clip length is fixed and cannot be extended with parameters
• SD video in Relax mode requires Pro or Mega plan
• HD video uses significantly more Fast time than SD

Next steps

Related reading: /blog/how-to-turn-midjourney-images-into-videos covers the foundational workflow of getting from a still image to your first video clip. Once you have motion control locked in, the next step is building consistent visual systems for the source images themselves, so every video you generate matches a cohesive brand look.