Genvizu

AI video prompt generator

AI video prompt generator for text-to-video shots

A video prompt is a shot description, not a picture description. Name the opening frame, one camera move, one subject action and the duration, then pick a model whose length and resolution match the brief.

Start creatingExplore prompts

Start free · no credit card

  • Video models from 2 credits per second up to fixed-price 4K clips.
  • Duration, resolution and reference inputs are set in the controls, and the cost updates before you generate.
  • Failed tasks return the credits through the existing refund-safe task lifecycle.

Try it now

Describe the shot you want and get a video prompt with camera, pacing and duration.

Free3 free runs left today, no sign-in needed

0/600

Quick human check

Loading human verification…

Three shot briefs written as prompts

Each input is the shot as a director would note it down. Each output is the prompt that survives contact with a video model.

Product hero shot, 5 seconds

Input

Shot: espresso machine on a cafe counter. Move: slow push-in. Duration: 5s. Ratio: 16:9.

Prompt output

Slow dolly push-in on a stainless espresso machine on a walnut cafe
counter, steam rising from the portafilter, warm morning light raking in
from a window on the left, shallow depth of field, background softly out
of focus, no camera shake, continuous take, 5 seconds, 16:9

One camera move plus one subject motion. Two moves in a single clip is where most video generations fall apart.

Character beat, 6 seconds

Input

Shot: woman looks up from a notebook and smiles. Move: static, then a slight tilt up. Duration: 6s. Ratio: 9:16.

Prompt output

Medium shot of a woman at a wooden desk writing in a notebook; she
pauses, looks up towards the camera and smiles. Camera holds static for
the first half then tilts up slightly, soft overcast window light, muted
palette, natural skin texture, no cuts, continuous take, 6 seconds, 9:16

Write "no cuts, continuous take". Without it, models often insert a cut at the moment the action changes.

Social loop, 4 seconds

Input

Shot: ink dropping into water. Move: locked-off macro. Duration: 4s, should feel loopable.

Prompt output

Locked-off macro shot of a single drop of black ink falling into clear
water, slow-motion bloom of tendrils, backlit against a white
background, high contrast, clean plate, no people, no camera movement,
4 seconds, 1:1

Locked-off macro shots are the cheapest way to test a model. A 4-second 720p clip on Gemini Omni Flash 1.1 is 79 credits; Seedance 2.5 starts at 5 seconds.

Scene and motion

Define subject action, camera move, environment, pacing, and continuity before the model call.

Model-ready constraints

Keep duration, aspect ratio, resolution, and style aligned with the selected video model.

Credit-aware planning

A better prompt reduces wasted runs while the existing credit and task system remains the source of truth.

Prompt structure for video

  1. 01

    Describe the opening frame and the final frame.

  2. 02

    Add one camera movement and one subject movement.

  3. 03

    Specify duration, aspect ratio, lighting, and style.

  4. 04

    Generate, review motion quality, then refine one variable at a time.

Video models and what each one is for

Some models bill per second and some per clip, so the same brief can cost very different amounts. The figure on the generate button is the one that counts.

ModelBest forCost and limits
Seedance 2.5Longer takes and shots built from reference video or audio5–30s; 480p / 720p = 35 / 79 credits per second, 5-second minimum
Gemini Omni Flash 1.1Fixed-length clips, reference images and first/last-frame control4 / 6 / 8 / 10s; 720p or 1080p 4s = 79 credits per clip, 4K 10s = 263
MiniMax H3 (Hailuo 03)Shots that need native stereo audio at 2K4–15s; 2K = 46 credits per second
Grok Imagine Video 1.5Cheap drafts while you are still testing the prompt1–15s; 480p / 720p = 2 / 4 credits per second, no minimum

Where video prompts usually go wrong

The clip contains a cut you never asked for

Add "no cuts, continuous take" to the prompt. Models tend to cut whenever the described action changes, because that is what the training footage does.

The camera drifts or shakes on its own

Name exactly one camera move and then rule the rest out: "slow dolly in, no handheld shake, no zoom". Silence is read as permission to move.

The duration in the prompt is ignored

Duration comes from the control next to the generate button, not from the prompt text, and it drives the credit cost. Keep the prompt duration only as a hint to the pacing.

Text or a logo in the footage flickers

Keep type out of generated footage entirely. Generate a clean plate, then composite the logo and captions in an editor where they stay stable.

Testing burns credits faster than expected

Draft at 480p on a per-second model, lock the prompt, then re-run the winning version once at the resolution you actually need.

Video generation requires a signed-in account and deducts credits per second or per clip, shown before you generate. Reference images, video and audio you upload are stored by Genvizu for the run, results are kept in your generation history, and failed tasks are refunded by the existing task lifecycle.

Not the page you need?

Image to prompt

You have a still reference and want the prompt behind it before animating anything.

Prompt tools

You want the full list of prompt tools before picking one.

Questions people ask

Related pages

Prompt toolsAI prompt generatorImage to promptMidjourney prompt helperModel pricingExplore galleryVideo creations

How long can a generated clip be?

Seedance 2.5 runs 5 to 30 seconds, MiniMax H3 runs 4 to 15 and Grok Imagine Video 1.5 runs 1 to 15. Gemini Omni Flash 1.1 is fixed at 4, 6, 8 or 10 seconds, and Veo 3.1 is fixed at 8.

How is a video prompt different from an image prompt?

An image prompt describes one frame. A video prompt has to describe change over time: what the subject does, how the camera moves, and how long that takes. Add those three and the same scene description starts producing usable footage.

What does one test clip cost?

The exact figure appears on the generate button first. As a reference point, a 5-second Seedance 2.5 clip at 480p is 175 credits, a 4-second Gemini Omni Flash 1.1 clip at 720p is 79, and a 10-second Grok Imagine Video 1.5 clip at 720p is 40.