Guide · updated August 24, 2026
30 image to video prompts you can copy
Image-to-video prompts are a different craft from text-to-video prompts, and using the wrong style is why so many first attempts fail. In text-to-video you are describing an entire world. In image-to-video the world already exists — your photo decided it — and your only job is to say what moves, how the camera behaves, and how much.
Below are thirty prompts, grouped by subject, that you can paste straight into the video studio with your own image. Every one of them follows the same four-part formula, which is explained first so you can write your own.
The formula: subject, movement, camera, pace
Subject anchors the model to what is in the frame — one short phrase, no need to describe the whole scene. Movement is the specific thing that changes. Camera is what the lens does: push in, orbit, drift, hold. Pace is the intensity dial, and it is the most important word in the prompt.
Two lines maximum. Long prompts fight the image. And when in doubt, add "subtle motion" — restraint is what separates a clip that looks filmed from a clip that looks generated.
- ·Subject: "a woman in a red coat", "a ceramic mug on oak", "a golden retriever".
- ·Movement: "hair moving in the breeze", "steam rising", "ears twitching", "leaves shifting".
- ·Camera: "slow push in", "gentle orbit to the right", "static shot", "slow drone pull back".
- ·Pace: "subtle motion", "gentle", "cinematic slow", "minimal movement".
Portraits and people
1
a woman in a tailored coat, hair moving gently in the breeze, a soft blink and slight smile, camera pushes in slowly, subtle natural motion
2
a man at a cafe table, he lifts the cup and takes a sip, steam rising, background pedestrians blurred and moving, static shot, natural pace
3
a portrait against a plain backdrop, the subject turns slowly from profile toward the camera, studio light unchanged, slow and controlled
4
a person walking toward the camera on a city street, coat moving with each step, shallow depth of field, camera tracks backward slowly
5
a musician holding a guitar, fingers moving on the strings, head nodding slightly to the rhythm, warm stage light flickering, gentle handheld motion
Pets and animals
6
a golden retriever looking at the camera, ears twitching, head tilts to one side, tail wagging out of frame, gentle handheld motion
7
a cat sitting in a sunlit window, tail flicking slowly, eyes blinking, dust motes drifting in the light, static shot, very subtle motion
8
a dog running through tall grass toward the camera, grass parting, ears flapping, camera tracks backward, energetic but smooth
9
a horse standing in a field at golden hour, mane moving in the wind, slow head turn, grass swaying, slow drone push in
10
a pet portrait in a royal costume, the fabric shifts slightly, a slow blink, a small dignified head turn, camera almost still, subtle motion
Products and e-commerce
11
a perfume bottle on a stone pedestal, camera orbits slowly to the right, light glinting across the glass, everything else perfectly still
12
a pair of sneakers on a concrete floor, camera pushes in low and slow, faint dust drifting through a shaft of light, static product, cinematic pace
13
a wristwatch on dark fabric, the second hand moves, a highlight travels across the polished case, extreme close-up, slow camera drift
14
a skincare bottle on a marble surface, a drop of liquid rolls slowly down the side, soft studio light, camera holds still, minimal movement
15
a coffee bag standing upright, beans falling slowly in front of it in the foreground, warm light, slow push in, controlled motion
Landscapes and nature
16
a mountain valley at sunrise, mist drifting between the trees, clouds moving slowly overhead, gentle drone push forward, cinematic pace
17
waves rolling onto an empty beach, foam receding across wet sand, seabirds crossing the frame, static wide shot, natural pace
18
a forest path in autumn, leaves falling gently, light shifting through the canopy, camera drifts forward slowly along the path
19
a city skyline at dusk, windows lighting up one by one, clouds moving, traffic streaming below, timelapse feel, slow motion of the camera
20
a field of lavender in the wind, rows rippling like water, distant trees swaying, slow drone pull back, gentle and continuous
Old and restored photographs
21
the person breathes gently and blinks once, the faintest smile, head turns two degrees toward the camera, camera pushes in very slowly, subtle natural motion
22
both figures breathe softly, one blinks, a small shared glance between them, everything else still, extremely gentle camera drift
23
a family group photo, minimal natural movement, slight shifts in posture, one or two blinks, no large gestures, camera almost static
24
an old street scene, faint movement in the background figures, light shifting slightly, dust in the air, very slow push in, restrained motion
25
a child in an old photograph, a soft blink and a small natural smile, hair moves slightly, camera holds still, extremely subtle and warm
Food, fashion and interiors
26
a bowl of ramen, steam rising in soft curls, chopsticks lifting noodles slowly, warm overhead light, camera pushes in gently
27
a chocolate cake on a stand, a slice being lifted away, crumbs falling, soft window light, slow orbit to the left
28
a model in a flowing dress, fabric moving in a light breeze, a slow turn, hair lifting, camera circles slowly, editorial pace
29
a living room with large windows, curtains moving gently, sunlight shifting slowly across the floor, camera drifts forward through the space
30
a kitchen counter with fresh ingredients, steam rising from a pan in the background, light shifting, camera pans slowly across the surface
How to adapt these to your own image
Swap the subject phrase for what is actually in your photo and keep everything after it. The movement, camera and pace clauses are the reusable part — they encode restraint, and restraint is the thing most people get wrong.
If a prompt produces too much motion, add "minimal movement" or "camera holds still". If it produces almost nothing, remove the word "subtle" and name the movement more specifically. Two or three runs is normal; models are non-deterministic, so the same prompt gives you different motion each time.
Draft on LTX at 5 credits until the motion is right, then run the same prompt on Kling or Seedance for the take you keep.
See how the draft-then-finish workflow works →Paste a prompt. Upload an image. Generate.
Kling, Veo 3.1, Seedance and LTX in one studio, from $9.99/month with 200 generations.
Frequently asked questions
How do I write a good image to video prompt?
Subject, movement, camera, pace — in that order, in one or two lines. The photo has already decided the scene, so your job is only to say what changes. "Subtle motion" is the most useful phrase in the whole vocabulary.
Why are image to video prompts shorter than text to video prompts?
Because half the work is already done. In text-to-video the model invents the entire scene from your words; in image-to-video it inherits composition, lighting and subject from your image and only needs direction about movement. Extra description just introduces conflict with what is already in the frame.
Do these prompts work on every model?
Yes, the formula is model-agnostic. Kling 2.5 and Seedance Pro respond particularly well to explicit camera direction. Veo 3.1 will also generate matching audio from the same prompt. LTX is the cheapest place to test one at 5 credits.
Can I ask for specific camera moves?
Yes, and you should — it is the difference between a clip that reads as filmed and one that reads as warped. Push in, pull back, orbit left or right, tilt up, track alongside, or hold static all work as instructions.
Why does the same prompt give different results?
These models are non-deterministic by design: the same input produces different motion on every run. That is a feature when you are exploring and an annoyance when you already had it right. Draft on a cheap model and re-run the winning prompt on a premium one.
Can I control exactly where the clip ends?
With Kling 2.5, yes. It accepts a start frame and an end frame and interpolates between them, which turns a non-deterministic process into something close to controllable. It is the only model in the studio that does this.