How to Animate a Photo with AI — Make a Picture Come Alive
MyDreamVid · 2026-09-11
Image-to-video is the fastest "wow" in AI video: you upload one still picture, describe how it should move, and a few minutes later the picture is alive — the camera drifts in, light sweeps across the scene, steam rises from the cup. Here is how to do it well.
The three-step recipe
- Upload the photo. JPG, PNG or WebP, up to 10 MB. Product shots, pets, landscapes, illustrations, old paintings — all work.
- Describe the motion, not the picture. The model can already see the image. Your prompt should say what changes: "slow camera push-in, morning light sweeping from the left, gentle steam rising, shallow depth of field."
- Pick duration and resolution, check the price, generate. The exact credit cost is shown before you start. The uploaded image itself costs nothing — you pay only for the seconds of video generated.
Getting the frame right
By default the output follows your photo's own proportions (the adaptive aspect ratio). That's usually what you want for a living photo. But if you need a specific format — 9:16 for a phone screen, 16:9 for YouTube — select it explicitly, and know that a landscape photo forced into 9:16 will be cropped. Compose accordingly.
What kind of photos work best
- Products — a slow turn or light sweep makes a static product shot feel premium. See How to make product videos with AI.
- Pets and animals — head tilts, ear twitches, fur in the breeze all read naturally. One tested tip: if you want your pet to "talk", ask for a voice-over ("the kitten's voice is heard as a voice-over; its mouth stays a natural cat mouth with only tiny movements"). Asking the model to lip-sync an animal tends to produce eerily human mouths.
- Landscapes and cityscapes — clouds, water, traffic and light are the model's home turf.
- Illustrations, paintings, old drawings — parallax and subtle motion bring artwork to life beautifully.
The one hard restriction: real faces
Every Seedance model refuses to animate a photo containing a real, recognizable human face — including your own. This is an upstream safety rule shared by all models, so switching models does not help. Illustrated or stylised characters are fine; photographs of people are not.
What it costs
Billing is per second of finished video, quoted exactly before you start. As a reference point, a 5-second 720p clip on Seedance 2.0 Mini costs about 54 credits — roughly $0.54. Credits never expire and failed generations are refunded automatically. Full rates per model and resolution are on the pricing page.
Frequently Asked Questions
Can the animated photo have sound?
Yes. Audio is on by default: ambient sound matches the scene, and if you put dialogue in quotes in the prompt, it gets spoken. On Seedance 2.0 Fast and Mini audio adds nothing to the price; on 2.5 and 2.0 it roughly doubles it.
Why was my photo declined?
Most often because it contains a real human face (see above), or because it tripped content review. Content rules cover the same categories as text prompts. Try a different photo or reword the motion prompt.
How long does it take?
Typically one to three minutes for a 5–10 second clip at 720p.
Can I animate an AI-generated image?
Yes — images generated in the MyDreamVid image workspace have a one-click Animate button that sends them straight into image-to-video, no re-upload needed.