How to animate a photo with AI (step by step)
Image to video takes a still you already own and generates the frames that follow it. There is no masking, no keyframing and no parallax layering — the model infers depth, light and plausible movement from the picture itself.
Step 1: pick the right photo
Sharp, well-lit images with one obvious subject animate best. Heavy motion blur, dense crowds and very dark frames give the model too little to work with.
- Good: product on a clean surface, portrait with soft light, landscape with sky
- Harder: busy group shots, screenshots of text, low-light phone photos
- Format: JPG, PNG or WebP under 5 MB
Step 2: scan the image
Scanning sends the photo to a vision model that describes the subject and proposes camera and subject motion — for example a slow orbit around a product with shifting highlights. The description lands in the prompt box, and you can rewrite any part of it.
Step 3: keep the prompt about movement
Your photo defines the look, so use the prompt for motion only: "slow push-in, gentle parallax, light drifting across the surface". Asking for new objects, a new background or a different outfit invites the model to redraw what you wanted preserved.
Step 4: choose length and shape before rendering
- 9:16 for TikTok, Reels and Shorts
- 16:9 for YouTube, landing pages and ads
- 5 seconds for a hook, 8 to 10 for a product reveal
Step 5: review, then iterate
Watch the first and last second closely — that is where drift shows. If the subject warps, simplify the motion: a static camera with subtle light movement is more convincing than an ambitious orbit on a complicated photo.
Where photo animation pays off
Ecommerce listings, property tours, album art, event promos and archive photos all become scroll-stopping without a reshoot. One good photo can supply a week of social posts once it moves.