4 September 2026 · 6 min read

How to Turn a Photo into a Video with AI

A practical walkthrough of animating a still image into short-form video — what makes a good source photo, how to pick an aspect ratio, and the mistakes that waste credits.

Start with the right photo

The single biggest factor in how good an AI-animated video looks is the image you feed it. The model is inferring motion from one frame, so anything ambiguous in that frame becomes an artefact once things start moving.

Photos that animate well share a few traits: a clear main subject, reasonable separation between subject and background, and even lighting without blown-out highlights. Photos that struggle tend to be busy group shots, heavily filtered images, or anything where the subject's edges blur into the background.

If your source is small or soft, run it through the image upscaler first. Motion tends to exaggerate softness rather than hide it.

Choose the aspect ratio before you generate

Decide where the video is going before you spend credits. Cropping afterwards means throwing away pixels the model worked to produce, and it usually cuts your subject badly.

• 9:16 — Reels, Shorts and TikTok. The default for short-form.
• 1:1 — feed posts where either orientation may be shown.
• 16:9 — YouTube proper, or anything destined for a landscape player.

Describe motion, not just the scene

When you write the prompt, remember you are directing a shot rather than describing a picture. The image already establishes what things look like; the prompt's job is to say what moves and how.

"A woman standing in a field" restates the photo. "Slow push in, hair moving in a light breeze, grass swaying" tells the model what to actually animate. Naming the camera move — push in, pull back, slow pan left — gives you far more control than adjectives about mood.

Keep the motion modest

The most common mistake is asking for too much. Large, fast movements force the model to invent detail it never saw — the back of a head, an occluded hand — and that is exactly where distortion appears.

Subtle motion reads as real. Dramatic motion reads as AI. If you need a bigger movement, it is usually better to generate two shorter clips and cut between them than to ask one clip to do everything.

Add voice, then sync it

A silent clip rarely holds attention. Once the video works, write a short script and generate narration with text to speech. If your subject is a person facing the camera, run the result through lip sync so the mouth matches the audio — that one step is the difference between a moving photo and something that reads as a real talking-head clip.

A workflow that holds up

1. Pick or generate a strong source image; upscale it if it is soft.
2. Set your aspect ratio for the destination platform.
3. Write a prompt describing camera movement and subject motion.
4. Generate two or three takes rather than iterating on one.
5. Add narration, and sync it if there is a face on screen.
6. Trim the weakest half-second off each end — almost every clip improves.

The whole sequence runs inside AI Creator Studio on one credit balance, so you are not exporting between apps at each step.