Image to Video AI: turn a still into a moving shot
Upload a photo, product image, portrait, or generated frame. Your image sets the starting appearance; your prompt describes what moves. Choose an image-to-video model above and review the available duration, ratio, and audio options before generating.
Slow push-in, steam rising, raindrops sliding down the window, warm reflections flickering subtly.
The café example uses small movements around an established scene. You do not need to describe every object again. Focus on the action and the parts of the image that should stay steady.
Write an image to video prompt in three parts
01 / Subject movement
Choose one visible action: steam rising from a cup, a person blinking, fabric moving in a breeze, or a small turn of the head. Subtle movement is a useful first test because it gives you fewer changes to review.
The subject breathes gently; loose hair moves in a light breeze.
02 / Camera movement
Choose a fixed camera, slow push-in, gentle pan, or restrained orbit. A large orbit reveals surfaces the image never showed, which can introduce invented details. Start with less movement when exact product shape or facial appearance matters.
Slow camera push-in, keeping the subject centered.
03 / What stays steady
Name the important constraints: product outline, face, background, lighting, or lettering. These instructions guide the generation but cannot guarantee perfect preservation. Watch the whole clip before treating the result as ready to publish.
Keep the label, tabletop, and lighting consistent.
How to turn a photo into a video
Use a sharp source with clear edges and enough space around the subject. A still with visible defects is worth fixing in Image to Image AI before you animate it.
No starting frame yet? Create one with Text to Image AI. When you want the entire scene invented directly from a written brief, use Text to Video AI.
- Upload your image. Use the reference-image input in the generator. Check that the preview shows the intended file and that the crop leaves room for your planned movement.
- Describe the motion. Start with one subject action and one camera instruction. Avoid asking for several unrelated scenes or major costume and location changes in the same short shot.
- Review the settings. Pick from the model’s supported duration and output options. Audio, end frames, and additional reference inputs are model-dependent. Check the estimated credits before submitting.
- Generate and watch. Play the result from beginning to end. Look for flicker, changing edges, distorted faces, or a camera move that drifts away from the subject. Download the take that matches the brief.
Review an image to video result before sharing
Pause near the beginning, middle, and end. For a product, compare its silhouette and label. For a portrait, watch the eyes and mouth through the movement. For a landscape, look for structures that bend or disappear as the camera moves. A clean opening frame does not mean the entire shot is consistent.
If motion causes distortion, simplify the prompt and reduce the requested movement. If the shot barely moves, specify a visible action rather than adding more style adjectives. Keep your source image and prompt so you can compare attempts without changing several variables at once.
For several connected panels, continue in Storyboard to Video. If you still need to decide the shot order, use the AI Storyboard Generator to plan the sequence before rendering. A single image-to-video shot works best with one clear moment to animate.