Image to video

Image-to-Video Prompt Guide: Subject Motion and Camera Moves

The reference image establishes a starting composition. Your prompt should explain what changes over time. For a first attempt, choose one visible action and one camera decision instead of trying to describe an entire film in one clip.

Omni AiLab guides · Published and updated October 15, 2026

  1. Choose the still
  2. Describe motion
  3. Review settings
  4. Inspect the clip

1. Choose a still that supports the movement

Start from an image with a clear subject and a composition that can accommodate the intended action. A hand already cut off at the frame edge is a difficult starting point for a gesture. A very tight portrait leaves little room for a broad body movement.

Look for ambiguous details, distorted anatomy or unclear product edges before uploading. Animation can make those problems more noticeable. If you need a specific composition, improve the still first rather than asking the video step to solve every visual problem.

2. Separate subject motion from camera movement

Describe the subject action in one sentence: a person turns toward a window, fabric moves in a breeze, or a product stays still as light shifts across its surface. Then add the camera instruction, such as a slow push-in or a locked-off frame.

A camera push-in is different from the subject walking toward the camera. State which one you mean. You can also specify what should remain steady: the face direction, the background or the placement of the product. These instructions describe intent; exact motion still depends on the selected model and input.

3. Review the model inputs and output settings

Open the video studio and choose a model that supports your intended reference workflow. Add the still using the available image reference control, then review the current duration, resolution and other output settings. Available controls can differ between models.

Keep the action small enough to be readable within the selected clip. If the story needs a wide shot, a close-up and an ending, plan them as separate shots. Check the task quote before generating, then use the returned task state rather than clicking again while acceptance is unclear.

4. Watch the entire result, including the first and last frames

Review the clip at normal speed and pause around the action. Check whether the camera follows the intended direction, whether the subject remains coherent and whether the background changes unexpectedly. For a product, inspect labels and edges through the motion.

If a result is close, simplify the next prompt around the part that failed. Keep the reference stable and change one motion instruction. For a sequence, choose an ending composition that gives the next shot a clear starting point; a clip does not need to contain every transition itself.

Prompt examples to adapt

These prompts illustrate a creative brief. They are not claims about a measured generation result; adapt them to your reference and the model controls available in the app.

A restrained portrait movement

Use the reference image as the starting composition. The person slowly turns their gaze toward the window while keeping a relaxed posture. A gentle movement in the hair from a light breeze. Locked-off camera, stable background and soft window light. Keep the action subtle and continuous.

Begin with a small action when testing a portrait reference.

A slow camera move around a product

The referenced product remains stationary on the table. The camera makes a slow, shallow move from front-left toward the center, keeping the product fully in frame. Soft reflections move naturally across the surface. Keep the background quiet and avoid adding new objects or lettering.

If the shape or label changes, improve the reference and reduce the camera movement.

An atmospheric scene

Start from the supplied street scene. Light rain falls and small ripples appear in the puddles. The camera slowly pushes toward the illuminated doorway. Maintain the original composition and restrained color palette. One continuous shot with no cuts or sudden changes of viewpoint.

Describe environmental motion separately from the camera instruction.

When a result needs another pass

The clip contains too many competing actions

Reduce the brief to one subject action and one camera decision. Split additional story beats into separate shots.

The camera moves in the wrong direction

Use a concrete direction and name the subject that stays in frame. Avoid combining contradictory camera terms.

The scene drifts away from the reference

Try smaller motion and a simpler background. Review whether the chosen model supports the reference inputs you supplied.

Before the next creation step

  • The starting image is clean and readable.
  • Subject motion and camera motion are separately described.
  • The action fits one shot.
  • The entire clip has been checked before reuse.

Open the video studio

Plan a cinematic shot · Explore video examples

Reading this guide does not start a generation. Sign in and review the selected model, settings and quote in the creation tool when you are ready.

Workflow guidance is compiled from the creation tools currently available in Omni AiLab. Controls and availability can change; the current interface is the reference for a specific task.