AI Image-to-Video Prompts: 7 Templates + Formula (2026).

The image-to-video prompts that work name one motion and stop — the image already supplies the scene. Below: 7 templates built on that formula, plus the cost — 3 credits at signup, free.

Jump to: Portrait dolly-in · Product orbit · Ambient still life · Landscape crane-up · Portrait gesture · Handheld action · Cinematic style pairing

The formula (sourced, not guessed)

Kling calls it subject + movement + background. Runway's Gen-4 guide says re-describing visible details cuts motion. Google's Veo docs confirm the image becomes the video's first frame — the prompt only covers what's next.

Quick reference: which move for which shot

Need a still first? The AI image generator makes one; the image-to-video explainer covers the step.

1. Portrait: slow dolly-in

Slow dolly-in toward the subject, hair drifting gently in a light breeze, calm and steady, warm evening light.

Why this works: one motion, no re-description — safest for a face.

2. Product: static orbit

Slow 360-degree orbit around the product, everything else still, clean studio light, no camera shake.

Why this works: one locked subject, one camera path — what product shots reward.

3. Still life: ambient motion only

Locked-off static shot, steam rising slowly from the cup, fabric settling, nothing else moves.

Why this works: zero camera motion isolates the one moving element.

4. Landscape: crane-up reveal

Slow crane-up revealing the horizon, clouds drifting steadily, grass moving in the wind, calm pace.

Why this works: wide shots hide small errors — the most forgiving template to test first.

5. Portrait: natural gesture

Subject slowly turns their head toward camera and smiles, a natural blink, everything else still.

Why this works: one gesture, no camera move — isolates face handling.

6. Action: handheld push-in

Subtle handheld push-in, documentary feel, a crowd moving in the background, the subject holds steady.

Why this works: camera move plus background motion, subject held still — the riskiest pairing here, test last.

7. Style pairing: cinematic slow motion

Cinematic slow motion, shallow depth of field, golden-hour light, a gentle drifting camera, subject barely moving.

Why this works: one style layer, same four slots — the formula carries mood too.

When not to run these on getvivix

Already inside one model's app with no plan to compare? Prompting there is simpler. getvivix, the AI video studio at getvivix.com, earns its keep running one prompt across Kling, Veo, and other models without separate accounts — see the full model comparison if deciding.

Frequently asked

Which image-to-video prompt should I start with?

Portrait dolly-in (#1) or landscape crane-up (#4) — one camera move, one motion, least likely to warp a face.

What does it cost to run these prompts?

3 credits at signup, free. Paid tiers run 1,000, 1,500, or 5,000 credits monthly — exact cost shows on Generate before you click (July 2026).

Does the same prompt work on every model?

Wording ports across models — Kling, Runway, and Veo all read subject-plus-motion prompts alike, though motion amount varies. Test one prompt on two models to confirm fit.

Try these prompts on getvivix — animate a photo across 100+ models, cost shown before you click. Or read the full photo-to-video workflow.

Published by the getvivix team · Last updated July 28, 2026