How to animate a photo with AI: turn any picture into a video
Step-by-step guide to animating photos with AI: choosing the right image, writing motion prompts, picking a model, and fixing common problems like warped faces.
Animating a photo is the easiest way into AI video. You already have the picture; the AI only has to add movement. Done well, a still portrait turns into someone who breathes, smiles and looks around; a product shot turns into a slow, glossy reveal; a landscape gets drifting clouds and moving water.
Step 1: Choose the right photo
The photo decides most of the result.
- Sharp and well lit. Blurry or dark photos get worse when animated. If yours is small, run it through Image Upscale first.
- Clear subject. One person, product or object in focus works far better than a busy group shot.
- The right shape. The video follows your photo's aspect ratio. Crop to 9:16 for TikTok and Reels or 16:9 for YouTube before you upload.
- Room to move. If you want a camera pull-back, leave some space around the subject.
Step 2: Write a motion prompt
The model can already see the photo, so do not describe what is in it. Describe what happens next.
| Instead of | Write |
|---|---|
| A woman in a red dress on a beach | She turns toward the camera and smiles, wind moving her hair and dress, slow push-in |
| A coffee cup on a table | Steam rises from the cup, soft morning light shifts across the table, static camera |
| A mountain lake | Clouds drift over the peaks, gentle ripples on the water, slow aerial forward movement |
| A dog | The dog tilts its head, then wags its tail and barks once |
Three rules: one main action, one camera move, and small, natural motion for faces.
Step 3: Pick a model
Open image to video and choose:
- Kling 3.0 for the best all-round result, 3–15 seconds, with sound.
- Veo 3.1 Fast for the most realistic people and speech.
- Seedance 2.5 for long shots, up to 30 seconds.
- MiniMax H3 if you want to set the last frame too, or need 4K.
- Seedance 2.0 Mini or Kling 2.5 Turbo to test ideas cheaply.
Not sure? Read our comparison of image-to-video models.
Step 4: Generate, compare, refine
Generate two or three takes. AI video is a little different every time, and one take is often clearly better. If it is not right:
- Face changes or warps: ask for smaller movement, use a sharper front-facing photo, or switch to Kling 3.0 or Veo 3.1 Fast.
- Nothing much happens: be more specific about the action and add a camera move.
- Too much happens: remove extra actions; shorten the clip.
- Strange hands or objects: keep hands still or out of frame, and simplify the prompt.
Step 5: Add a voice or effects
- Make them talk: add a script or your own recording with Lip Sync.
- One-tap viral effects: try Effects for looks like Earth Zoom and Liquid Metal without writing a prompt.
- Sharper result: finish with Video Upscale up to 4K.
Ideas to try
- Bring an old family portrait to life with a gentle smile and blink.
- Turn a product photo into a slow 360° orbit for an ad.
- Make a travel photo feel alive with moving water and clouds.
- Animate an AI-generated character for a short story.
- Turn a pet photo into a funny reaction clip.
Frequently asked questions
How do I make a photo move with AI?
Upload the photo to an image-to-video model such as Kling 3.0, describe the motion you want, choose a length and generate. The photo becomes the first frame of the video.
Can I animate an old family photo?
Yes. Upscale it first if it is small or blurry, then use gentle motion like a smile, a blink or a slow camera push-in for the most natural result.
Can I make the person in the photo talk?
Yes. Animate the photo, then use Lip Sync to add a script or your own recording, or use a model with native audio like Veo 3.1 Fast and write the line in your prompt.
Why does the face change during the video?
Large head turns, fast motion and low-resolution photos cause drift. Use a sharp, front-facing photo, ask for smaller movements, and try Kling 3.0 or Veo 3.1 Fast, which hold faces best.