How-to · 6 min
Image to Video Guide: Turn a Still Image Into Motion
Image to video lets you start from a picture you already have and add motion to it. Instead of describing a scene from scratch, you upload a still and Seedance Mini animates it, keeping your composition while bringing in camera movement and subject motion. This is the most controllable way to use AI video, because you decide exactly what the frame looks like.
On SDMini, image-to-video is available to registered users on the free daily credits. This guide covers picking the right source image, writing a motion prompt for it, and using first-frame and last-frame conditioning to steer the result.
Choosing a source image that animates well
The cleaner your source, the better the motion. Pick a sharp, well-lit image with a clear subject and some breathing room around it. Photos with an obvious foreground and background animate more convincingly because the model has depth to work with.
Avoid heavily cluttered frames, tiny faces in the distance, or blurry shots. The model preserves what is in your image, so flaws come along for the ride. If your subject is partly cut off at the edge, expect the motion near that edge to look less stable.
Writing a motion prompt for your image
With image-to-video you are not describing the whole scene, since the image already provides it. You are describing what should move. Keep it focused: "gentle wind moves the grass, slow push-in, clouds drift in the background." That tells the model how to animate without fighting the picture.
Match the motion to what is plausible in the frame. If the photo shows still water, ask for soft ripples, not crashing waves. If it shows a person standing, ask for a subtle shift in weight or a slow camera move rather than a full action. Believable motion comes from respecting what the still already implies.
First-frame and last-frame conditioning
Seedance Mini supports first-frame and last-frame conditioning. Using your image as the first frame means the clip begins exactly on your picture and animates forward, which is ideal when the opening shot is the one you care about.
Setting a last frame instead, or both, lets you define where the motion ends. This is powerful for transitions and reveals: lock a start and an end, and the model interpolates a smooth path between them. Use it when you need the clip to arrive at a specific composition rather than wander.
Settings and export
Choose your length and aspect ratio to suit the destination, the same as text-to-video. If your source photo is vertical, render 9:16; if it is wide, render 16:9. Matching the ratio to the image avoids awkward cropping.
Registered users can output up to 1080p and export with no watermark on paid plans. Because the model adds visuals only, layer any audio afterward. A single animated photo with a music bed makes a striking, shareable clip.
Frequently asked questions
- Do I need an account for image to video?
- Yes. Image-to-video is available to registered users, who get free daily credits. The free anonymous tier covers text-to-video at 480p.
- What kind of photo works best?
- A sharp, well-lit image with a clear subject and some space around it. Clean photos with foreground and background depth animate most convincingly.
- What is first-frame and last-frame conditioning?
- It lets you lock the starting image, the ending image, or both. The model then generates smooth motion between your fixed frames, which is great for reveals and transitions.
- Will the clip look exactly like my photo?
- The composition stays recognizable because your image anchors the clip, but the model adds motion and camera movement, so the frame changes as it plays.