Slow zoom-in on the subject's eyes over 4 seconds, no subject movement, ambient particle drift in the background, contemplative mood.
Image to Video Prompt Guide
Image-to-video prompts are different from text-to-video: the still already determines composition and identity, so the prompt should describe motion and camera language only. These prompts cover the seven motion patterns Kubricon's image-to-video flow handles most reliably across Sora, Kling, Seedance, and Veo — zoom-ins, orbits, dollies, product spins, handheld walks, atmospheric drift, and character micro-motion. Paste the prompt next to your source still in Kubricon.
Camera move + Speed + Subject motion (if any) + Environmental motion + Mood
Use this skeleton as a starting point. Substitute your subject and setting, keep one idea per clause, and lead with the most important visual so the model anchors on it before adding camera and lighting detail.
Zoom moves
5 promptsQuick zoom-out from a single product detail to reveal the full product on its pedestal, 2-second move, no other motion.
Slow dolly-zoom: subject stays the same size while the background compresses, classic vertigo effect, 5-second duration.
Snap zoom-in followed by a hold: aggressive 0.5-second zoom into the subject's face, hold for 2 seconds, designed as an ad hook.
Punch-in zoom on a hand-held object (a phone screen, a price tag, a logo), 1.5 seconds, kinetic and short-form-ready.
Orbit shots
5 promptsSmooth 180-degree orbit around the subject, level horizon, constant speed, 6-second duration, subject does not move.
Full 360-degree orbit at a slight tilt-down angle, the subject is a sneaker on a turntable, 8 seconds, studio lighting unchanged.
Half-orbit (90 degrees) around a car parked in a desert, low angle, 4-second move, sun stays in frame as the camera moves.
Slow orbit around two characters facing each other, the camera passes between them, 6 seconds, dramatic mood.
Reverse orbit: camera rotates while pulling back simultaneously, revealing more of the environment, 5-second duration.
Dolly moves
5 promptsSlow forward dolly into a doorway, 3-second push, no subject motion, light shifts subtly as the camera advances.
Lateral dolly from left to right across a landscape, 5-second sweep, distant clouds drifting in parallax.
Pull-back dolly revealing a wider context (subject was framed tight, now full body), 4-second move.
Forward dolly down a long corridor toward a distant light source, 6-second push, subtle perspective compression.
Slow dolly arc combining forward motion and a slight curve to the right, 5 seconds, narrative reveal.
Product spin
5 promptsSubject (product) rotates 360 degrees on a turntable, camera locked, 6-second full rotation, studio lighting unchanged.
Product spins while the camera does a slow vertical tilt, revealing the top of the product mid-rotation, 8 seconds.
Product spins counterclockwise as a single beam of light sweeps across it, 5-second loop-ready motion.
Product hovers and rotates simultaneously, soft floating motion, 6-second duration, magical reveal mood.
Macro product spin: extreme close-up on the surface, slow 180-degree rotation, lighting shifts to emphasize texture.
Handheld / walk-through
5 promptsHand-held POV walking forward into the scene, subtle natural sway, 6-second push, no zoom.
Steadicam walk-through of a room, camera weaves slightly left then right, 8 seconds, natural pacing.
Hand-held lateral walk across a market street, subject (a person) stays centered as the background scrolls, 7-second duration.
POV running shot, faster pacing, slight vertical bounce, 4-second sprint, kinetic mood.
Slow gimbal walk away from the subject, the subject stays in frame growing smaller, 6 seconds, melancholic mood.
Atmospheric drift
5 promptsSubject is static, but add slow drifting clouds across the sky and faint dust particles in the air, 6 seconds, contemplative.
Subject is static, add gentle water rippling on a nearby lake, subtle wind in the trees, 8 seconds.
Subject is a candle, add a flickering flame and faint smoke rising, 5-second loop-ready motion, dim lighting.
Subject is a city skyline at night, add blinking lights in distant windows, faint traffic flow streaks, 6 seconds.
Subject is a misty forest path, add slow fog drift and faint sun rays filtering through trees, 7 seconds.
Character micro-motion
5 promptsCharacter's portrait stays centered; add gentle breathing, slow blinks, and a faint smile forming over 4 seconds. No head movement.
Character standing in profile: add a slow head turn toward the camera over 3 seconds, hold for 2 seconds.
Group shot: each person blinks at slightly different intervals, one person tilts their head to the side, 5-second duration.
Character looking down at a phone: thumb scrolls slightly, eyes track the screen, ambient room motion in the background, 5 seconds.
Character laughing: shoulders rise and fall, eyes crinkle, hair moves slightly, 4-second authentic moment.
Frequently asked questions
Why are image-to-video prompts shorter than text-to-video prompts?
In image-to-video the still already controls composition, subject, lighting, and identity. The prompt should describe motion only — camera move, subject motion, and ambient drift.
Which models support image-to-video on Kubricon?
Sora 2, Kling 3, Seedance 2.0, Veo 3.1, and LTX 2.3 Pro all support image-to-video to varying degrees. The prompts here are tuned to be model-agnostic; refine per model after testing.
What is the best image to use as a source still?
A clean, sharp, high-contrast image with clear subject-background separation works best. Avoid stills with heavy motion blur, low resolution, or compressed JPEG artifacts.
Can I combine camera and subject motion in one prompt?
Yes, but keep it light. Aim for one camera move (e.g. slow orbit) + one subtle subject motion (e.g. blink, breathe) per clip. Stacking too many movements often confuses the model.
How do I keep the source still's look intact?
Add "keep the lighting and background identical to the source still" at the end of the prompt. Avoid describing the subject or scene again — only describe what moves.
Bring these prompts into the Kubricon editor
Copy any prompt, run it on the model you want, then add captions, hooks, beat-sync, and vertical reframing before exporting for TikTok, Reels, and YouTube Shorts.