Attach the source
Choose an image from your project assets or upload a new reference. Keep the subject unobstructed and avoid relying on detail that is too small to survive motion.
Upload a character, product, or location image. Describe what moves, what stays fixed, and how the camera behaves. The workbench then lets you set the clip length and frame before quoting the run.
The strongest image-to-video result usually starts with a restrained brief. Decide what the shot needs to communicate, then describe one readable movement instead of packing several unrelated events into a few seconds.
Choose an image from your project assets or upload a new reference. Keep the subject unobstructed and avoid relying on detail that is too small to survive motion.
State the subject action, camera movement, and atmosphere. Mention stable details such as wardrobe or product shape when consistency matters.
Select duration, orientation, resolution, and candidate count. Confirm the visible parameter summary and requested credit quote before generating.
A still already establishes composition and identity. The prompt should add the missing temporal information. For a portrait, a slight turn toward camera may be more convincing than a large gesture. For a product, a controlled orbit can reveal shape while preserving the label. For an environment, moving light or drifting atmosphere can add life without changing the scene beyond recognition.
Describe camera behavior separately from subject behavior. “The woman turns her head while the camera pushes in slowly” gives the generator two readable instructions. “Make it cinematic and dynamic” leaves the result open to interpretation. Concrete direction also makes the next iteration easier: if the camera is the problem, adjust that clause instead of rewriting the entire prompt.
The editor accepts reference media inside the same composition surface as the prompt. Audio can establish pace when the intended motion should follow a beat or mood. Project assets can be reused when you are producing several clips with the same character or product.
Marketing teams can start from an approved product or lifestyle frame and test several short movements without a new photo shoot. Character creators can keep a consistent face and costume while exploring small gestures. Storyboarders can bring a keyframe closer to the intended timing before assembling a sequence. Ecommerce teams can test a vertical teaser and a landscape placement from related source material.
Results still need review. Motion can introduce artifacts around hands, fine text, reflective surfaces, and complex edges. Use a shorter duration when testing a new source, keep the most important subject action near the beginning, and compare candidates before committing to a larger resolution. The case library shows the kind of prompts and formats this workflow is designed for.
No. Generation adds motion and may reconstruct details. Use explicit prompt language for the features that must remain recognizable.
Begin with a lower-cost test when you are still exploring motion. Raise the output quality after the direction is stable and the quote has been reviewed.
Yes. Authenticated project assets can be bound from the editor. Visit pricing to understand how generation and optional prompt rewriting are charged.
Attach the image, write the movement, and review the exact quote in the workbench.