Choose your route before writing
If you want to use our fixed scene, open the Flamingo Float AI preset. That route does not require a written prompt. Watch the sample, choose an authorized portrait, and review the cost shown before you start.
The prompts on this page are for a separate workflow in a creative tool that accepts images and text instructions. First, construct a still image with the rider already seated on the float. Then animate that image. Separating those decisions makes it easier to spot whether a problem began with the composition or the movement.
Use a clear portrait of an adult who has agreed to appear. Pick a tool whose current interface supports the inputs you need; an image reference and a video reference are different capabilities. Our photo guide explains what to check before choosing your starting portrait.
Flamingo Float AI prompt for the starting image
Copy this into an image tool that supports a portrait reference. Describe any clothing change separately if you want one; otherwise, begin with the appearance already visible in your photo.
Create a vertical cinematic pool scene using the uploaded portrait as the appearance reference for one adult rider. Preserve the person's recognizable facial features, hairstyle, and visible clothing characteristics without adding another person.
Seat the rider securely inside a large pink flamingo inflatable. Show the float's rounded body, curved neck, and head as a single connected object. The rider's hips rest inside the ring, with a relaxed upper body and believable leg placement. The rider casually looks at a phone held in one hand.
Place the float in a rooftop infinity pool above a city skyline. Keep the water surface, the pool's near boundary, and the distant buildings visually distinct. Position the float close enough to the boundary for its direction to be clear, while leaving visible water between the inflatable and the edge. Leave open water behind the rider for the return movement.
Frame the scene from a steady poolside viewpoint. Keep the rider's face readable and the float mostly visible. Use natural daylight, soft water reflections, and a coherent horizon. Do not add text, logos, extra riders, detached float parts, or an already falling pose.
What the image instructions are doing
The appearance clause assigns the portrait one job: informing the rider. It does not ask the image tool to transplant the entire background of your photo into the pool. Check the face and hair, but also check whether the shoulders, clothing, and seated body belong together.
The geography clause is equally important. An infinity edge can look like empty sky when the water and boundary blend together. Your viewer needs enough visual information to understand where the float is heading. Leave room for the return instead of filling the frame with the inflatable.
Finally, “securely inside” gives the movement a plausible starting position. If the rider is hovering above the ring in the still image, adding more motion instructions will not repair that starting pose reliably.
Flamingo Float AI prompt for continuous motion
Approve the still image before moving on. Use it as the starting frame in a tool that supports image-to-video creation, then give the animation instructions their own space.
Animate the supplied pool image as one continuous shot. Keep the same adult rider, pink flamingo inflatable, rooftop pool, skyline, and camera viewpoint throughout.
The float moves slowly toward the visible pool boundary. The rider stays relaxed and focused on the phone, keeping their body seated inside the ring. Let small ripples show the direction of travel.
A single helping hand enters from beside the camera. It reaches the outer side of the inflatable, makes clear contact with the pink ring, and gently pushes it back toward the open water. Show contact before the float changes direction. The hand releases the ring and withdraws without becoming a second visible character.
The rider gives a brief, natural reaction to the rescue, then settles back as the float continues drifting away from the boundary. Finish with the rider still seated and the inflatable fully inside the pool. Keep the movement continuous, with no sudden relocation, scene cut, falling over an edge, glass breakage, duplicate hands, or changing float shape.
Give each motion clause a specific job
The opening sentence holds the scene together. You already chose the clothing and skyline in the still; repeating a long visual description here can introduce competing details. Focus the second stage on changes over time.
Hand contact is the turning point. “A hand saves the rider” leaves the mechanism vague. This prompt names the contact surface, the order of events, and the direction of the push. Inspect that moment closely: the float should not reverse before the hand reaches it.
The reaction comes after the rescue, so it does not compete with the contact action. A glance and a brief response can read more clearly than several dramatic gestures happening together.
The ending clause defines this particular version. Other clips may use a different payoff, but this sequence finishes inside the pool. The negative constraints support that ending; they cannot guarantee that a model will obey every detail.
Optional: use a video reference you have permission to use
Some tools accept a motion or video reference as well as an appearance image. If yours does, use your own clip or a reference you have permission to adapt. You do not need footage of a public figure. Check the tool's input rules before uploading anything.
Assign a distinct role to each reference. This avoids asking the model to take the rider's appearance, camera position, and movement from two conflicting sources.
Use the uploaded portrait for the adult rider's appearance. Use the licensed reference video only to guide the pacing of the approach, the helping hand's contact, and the rider's reaction. Use the supplied pool image for the starting composition and setting.
Keep one continuous shot. Preserve a clearly visible pool boundary and open water for the return. The helping hand must touch the inflatable before pushing it away from the edge. End with the rider seated on the float, continuing to drift inside the pool. Do not transfer another person's identity, added captions, logos, or soundtrack from the reference video.
This is an instruction for assigning reference roles, not a rights clearance or a guarantee of matching motion. If the tool cannot accept all three inputs, choose a supported workflow rather than treating a filename mentioned in text as an actual uploaded reference.
Four small revisions to try
Keep the accepted instructions and change one issue at a time. Add the relevant line instead of rewriting the whole Flamingo Float AI prompt after every result.
- Unclear edge: “Keep the pool boundary visible beside the float throughout the approach and return.”
- Missed contact: “Hold the hand against the outer pink ring briefly before the float starts moving away.”
- Excessive reaction: “Limit the rider's reaction to a glance toward the helping hand and a small change in expression.”
- Wrong ending: “Finish the shot with open water between the float and the boundary; the rider remains seated.”
Review the sequence before generating again
Watch once at normal speed, then inspect the approach, contact, and final frame. Decide which single failure matters most. A convincing face does not compensate for a hand passing through the float, and attractive water does not explain a sudden jump across the pool.
For a composition problem, return to the still image. For a movement problem, revise the motion instructions. For a clean generation with awkward pacing, consider editing the result instead of regenerating everything. The visual repair guide helps separate these cases.
Does the exact phrase trigger the effect?
No special trigger phrase is established here. “Flamingo Float AI prompt” names the task; the scene and action descriptions communicate what you want.
What if there is no negative-prompt field?
Keep the short constraints in the main instructions. If a separate field exists, follow that tool's guidance and avoid contradictory directions across fields.
Can this reproduce an original video exactly?
Do not expect that from these prompts. Compare the result with your intended sequence, not an assumed guarantee. Use the storyboard guide to plan the beats, or return to the fixed preset when you want the site's demonstrated scene without writing prompts.