Upload the images that define your opening, ending, or key moments. Add one clear motion prompt, choose a model, and create a short transition you can review and refine.
Place visual anchors along a shot and watch the model fill in the movement between them. This example follows a rooftop character from a still pose to a leap above the city, with key moments you can inspect along the way.
Use two images when the transition is simple; add intermediate references when a pose, product angle, or setting needs to stay on course. The more clearly you define the important moments, the easier it is to review the result.
Give the clip a clear destination. Upload the image that starts the shot and the one that ends it, then describe the movement between them. This works well for a product reveal, a change of scene, or a subject moving into a new pose.
A glossy red apple opens, glows from the center, and becomes a column of orange flame.
For a longer or more structured shot, place several references in order. Use them to mark a new pose, camera position, product angle, or background, then explain the path from one moment to the next in a single prompt.
Reference thumbnails are arranged in sequence so the direction of a generated clip can be checked at a glance.
References can guide identity, framing, and style, but the output still needs a human review. Use images with similar lighting, keep one main action in the prompt, and add another anchor when the scene changes too much for one transition.
A reference-led clip follows the same character as the pose and camera path change.
Start with the moments you already know, explain the move in plain language, and inspect the generated clip before you decide whether to export or try another pass.
Upload a first and last image, or add intermediate references when the shot changes pose, angle, product view, or setting.
Name the subject action first, then add camera movement, pace, light, and details that should remain recognizable.
Follow the task in history, preview the clip, and rerun it when the motion or details need another pass. Available download settings depend on the selected model.
This workflow is most useful when you know the important visual moments: a storyboard beat, product reveal, scene change, or vertical social cut. Use the output to test direction before spending time on a full edit.
Use rough drawings or finished frames to check the movement between important poses. Add references when the character, camera, or environment changes, then treat the result as a motion study that can guide the next animation pass.
A rough character sequence is used as a visual guide while the generated clip adds movement between the chosen poses.
A few practical choices affect the result: how many images to upload, how to write the prompt, which ratio to use, and how to handle a clip that needs another attempt.
Image-to-video usually starts with one image and leaves most of the movement open to the model. A reference-led workflow gives the shot more structure: you can define the opening, destination, and important moments in between, then describe how the scene should travel from one to the next.
Upload an opening image and a final image, or add several references for a more structured shot. Select a model, ratio, duration, and quality, then write the action and camera direction. The generated clip appears in your history for review.
Yes. Two images are enough when the shot has a clear beginning and end, such as a product reveal or a scene change. If the subject needs to change pose or location halfway through, add an intermediate reference so the model has another point to follow.
Use more anchors when the shot changes pose, camera angle, product view, or setting in a way that two images cannot explain clearly. The available number of uploads depends on the selected model and input mode, and the composer shows those limits.
Start with the subject and its main action. Then add the camera move, pace, lighting, and anything that should stay stable, such as a product shape, face, outfit, or art direction. One focused action is usually easier to direct than several unrelated changes at once.
Yes. Describe the subject's direction, speed, and ending pose, then add a camera instruction such as push-in, pull-back, pan, orbit, tilt, tracking, or a locked-off shot. Intermediate references help when the movement has several important beats.
Use clear references with similar framing and lighting. Ask the model to preserve identity, proportions, colors, and product details, and keep the prompt focused on one transition. Text, logos, hands, and fast interactions can still vary, so review the clip before publishing.
Yes. Choose 9:16 for vertical social platforms and 16:9 for landscape video, presentations, or websites. Compose the reference images with the final canvas in mind so the subject stays in the visible area.
The selected model determines resolution, duration, aspect ratio, audio, and file details. The composer shows the available choices before you submit, and generated clips can be previewed or downloaded from your history.
The credit cost depends on the model and settings such as duration, resolution, audio, and quality. Check the estimate shown in the composer before submitting. Your plan or credit pack determines the balance and any included commercial rights.
Timing varies with the model, duration, resolution, queue, and traffic. You can follow pending and processing tasks in your history. If a task fails, consumed credits are returned automatically; contact support if the balance does not update.
Uploaded references are linked to your account and used to process the task. Avoid uploading confidential material unless your organization permits it, and review the Privacy Policy for storage and provider terms. Commercial use depends on the plan or credit pack that you choose.
