40 AI Image-to-Video Prompts You Can Copy and Paste (2026)
A working library of 40 image-to-video prompts for portraits, products, real estate, pets, and cinematic shots, plus how to write your own inside ImageToVideoAI.
Most people blame the model when their image-to-video clip comes out wrong. Stiff motion, melted faces, a camera that drifts when it should hold still. Nine times out of ten the model is fine. The prompt was vague.
A still image already locks your composition, colors, and subject. So the prompt has a narrower job than text-to-video: you're directing what moves and how the camera behaves, not building a scene from nothing. That changes how you should write it.
This is a copy-paste library. Forty prompts you can drop straight into ImageToVideoAI, plus the reasoning so you can adapt them to your own photo. Grab one, swap a noun, hit Generate.
Every solid prompt pulls four levers. Miss one and the model guesses for you.
Subject / identity: what stays consistent. For people this is the most fragile part. Say "keep facial features unchanged" and the model fights its own urge to morph the face.
Motion: what physically moves. Hair, fabric, water, steam, a blink, a turn.
Camera move: push in, orbit, pan left, or static. This is the single biggest lever for "cinematic" versus "phone footage."
Lighting / mood: flicker, golden-hour warmth, soft window light. Subtle, but it sells realism.
Here's the difference in practice.
Weak:
Make the woman move and look nice.
Strong:
A woman turns her head slowly toward the camera and smiles, hair shifting gently in a light breeze, soft natural window light, slow push-in, facial features unchanged.
Same photo, completely different result. The second one tells the model exactly what to animate and what to leave alone.
A few rules hold across every category:
One main motion per clip. Stacking five actions into 4 seconds produces chaos.
Describe speed. "Slowly," "gently," and "rapid" all change the output.
Name the camera move, or you'll get a random drift.
Match duration to motion. A slow orbit needs 5s; a quick blink works in 4s.
Each one is written to paste as-is. Adjust the subject noun to fit your image. The best model per category is noted so you can swap it in the model picker before generating.
For faces, identity preservation beats everything. Best model: Kling. For scanned or damaged photos, pair these with Animate Old Photos.
A man looks directly at the camera, blinks naturally, then breaks into a warm smile, subtle head tilt, soft front lighting, facial features unchanged.
An elderly woman in a vintage photo slowly turns her head toward the viewer and smiles softly, a faint breeze in her hair, warm sepia tone preserved, gentle slow push-in.
A young woman laughs lightly, shoulders shifting, loose strands of hair moving in a soft breeze, natural daylight, camera holds static.
A man in a suit nods once and shifts his gaze off-camera as if listening, slight shoulder movement, even studio lighting, identity preserved.
A 1950s family portrait gently comes alive, each person blinking and shifting weight subtly, warm film grain intact, no camera movement.
A woman closes her eyes, takes a slow breath, then opens them and looks up, calm expression, soft golden window light, slow push-in, features unchanged.
The goal is motion that flatters the product without distorting it. Best model: Veo 3.1 for polish, Seedance for camera moves. There's more setup in the product photo to video guide.
A perfume bottle on a marble surface, slow 180-degree orbit around the bottle, soft studio reflections sliding across the glass, shallow depth of field.
A sneaker on a clean white background, slow rotate to reveal the side profile, subtle rim light tracing the silhouette, product shape unchanged.
A coffee cup with steam rising and drifting upward, warm morning light from the left, camera holds still, gentle focus pull onto the rim.
A skincare jar with the label facing forward, slow push-in toward the lid, soft diffused lighting, water droplets glistening on the surface.
A wristwatch face catching light, slow tilt down across the dial, second hand ticking, reflective highlights moving, deep black background.
A handbag on a pedestal, slow dolly around to the front, soft top light, leather texture sharp, no warping of the shape.
Match the prompt type to the model before you generate. Cost scales with model, duration, resolution, and clip count, so the live number on the Generate button updates as you change settings.
Pick a photo, copy the closest prompt above, and tweak the subject to match. Set your aspect ratio, choose the model from the table, and watch the credit cost before you hit Generate.