Skip to content
H3 Max
Use CasesGuidesPricing
  1. Home
  2. /Guides
  3. /Text to Video Prompts

Prompt workshop · Text mode

Text to Video Prompts with Practical Examples

Useful text to video prompts describe a scene someone could film: a subject, an observable action, a setting and a camera direction. Start with one shot. In H3 Max Turbo, that shot can run from 5 to 15 seconds.

Build the first frame before moving it

Imagine opening the result at its first instant. What occupies the frame? Where does it sit? What separates it from the background? Text mode has no photograph to answer these questions, so give it enough visual information before describing movement.

The annotated example keeps each decision small. The notebook identifies the subject; the window supplies a plausible light source; the sunlight gives the shot something to do. A slow push describes how the viewer approaches that action. These text to video prompts work as shot briefs, with room for the model to interpret details you have not specified.

A phrase such as “a beautiful, cinematic workspace” leaves most of those decisions open. That may suit an exploratory background clip. If a particular object must appear in a particular place, name the object and its position instead of adding another adjective.

A shot in four parts

Subject and setting
An open plain notebook and wooden pencil on an oak desk beside a window.
Visible change
A small patch of sunlight moves slightly across the paper.
Camera
Slow camera push, one continuous shot.
Output choice
Select the duration, resolution and frame shape in the generator. Writing “vertical” is not a substitute for choosing the ratio.

Replace a mood with something visible

Compare “an inspiring morning desk” with “sunlight moves across an open notebook.” The second version gives you a specific result to inspect. If the light never moves, you know which request was missed. If the camera circles the room, the motion direction needs tightening.

When revising text to video prompts, make a small edit with a reason. Replace “dynamic camera” with “the camera slowly approaches the notebook.” Replace “lots of activity” with one action that fits the available seconds. Keep the setting unchanged while checking that edit.

Prompt wording revisions and what each revision clarifies
Loose wordingMore specific wordingWhat to inspect
A dramatic city sceneA quiet residential street after rain, viewed from pavement levelStreet type, wet surfaces and viewpoint
The camera moves beautifullyThe camera tracks slowly forward along the pavementTravel direction and pace
Tell a story about discoveryA folded paper boat drifts into a patch of sunlightOne visible event with a clear endpoint

Six text to video prompts to adapt

Each draft below assigns a different job to the shot. Replace details that matter to your project, then read the whole prompt again. A copied camera direction may conflict with your new subject. These are editable starting points; only the desk and street examples shown later have accompanying generated results.

Everyday moment: morning desk

A quiet desk beside a window. An open cream notebook with completely blank, unmarked pages, a dark wooden pencil and a glass of water rest on an oak surface. Soft morning light. The camera moves slowly toward the notebook in one continuous shot. No people, legible writing, subtitles or speech.

Recorded revised desk request, shown below. Select 16:9 separately; inspect the paper for invented writing.

Use this prompt ↗

Nature: one branch after rain

A close view of a small green branch after rain, with dark foliage softly out of focus behind it. One droplet gathers at a leaf tip and falls. The camera stays fixed while the branch moves very slightly in the breeze. One continuous shot.

Draft example. The droplet is the main event; a strong wind would make that event harder to follow.

Use this prompt ↗

Still life: ceramic on linen

A plain cream ceramic bowl rests on folded natural linen, seen from a low three-quarter angle. Soft side light reveals the rim and fabric weave. The camera slowly moves closer while the bowl remains stationary. Keep the whole rim inside the frame.

Draft example. This invents a bowl; use an image when a specific product must appear.

Use this prompt ↗

Product atmosphere: an imagined bottle

An unbranded green glass bottle stands on pale stone near a window. The bottle stays upright as a soft reflection travels across its curved surface. A fixed camera holds the full bottle and its contact shadow in view. One quiet continuous shot.

Draft example. Suitable for an invented mood shot, not proof of a real brand or product.

Use this prompt ↗

City: after the rain

A quiet residential street after rain, wet pavement reflecting soft evening light. The camera tracks slowly forward at walking height. No people dominate the frame. One continuous, natural-looking shot with buildings and trees remaining coherent.

Draft variant of a street brief. The recorded street result below uses its own saved prompt.

Use this prompt ↗

Narrative insert: a paper boat

A small folded paper boat drifts slowly along a shallow rainwater channel beside a stone path. It passes from shade into a patch of warm sunlight. The camera follows gently at water level, keeping the boat visible until it settles near a small pebble.

Draft example. The move into light supplies a beginning and ending without requesting another scene.

Use this prompt ↗

Save the version you actually submit. Text to video prompts often change while you are comparing ideas, and a result cannot tell you which words were present in its request. A short note containing the prompt, ratio, duration and resolution makes a later comparison much more useful.

What happened in the desk examples

The original desk generation added “The First Chapter” to the notebook even though its request excluded text. A second request added “completely blank, unmarked pages” to the notebook description. The revised output's sampled frames show blank pages. Both are retained because text to video prompts can still be interpreted differently from what you intended.

Both requests used five seconds, 768P and 16:9. They are independent generations with different compositions, so the pair does not prove that the added phrase will always prevent writing. Inspect the paper yourself. If blank pages are essential, reject a version with invented words.

Your browser does not support video playback.
AI-generated example · Text to Video · 5 seconds · 768PWide desk trial. The generated notebook contains an added title despite the request to avoid text.
Your browser does not support video playback.
AI-generated example · Text to Video · 5 seconds · 768PRevised desk request: “completely blank, unmarked pages.” The sampled frames remain blank; this is an independent generation.

Street trial: forward travel or a fixed view

The second pair changes the camera sentence from slow forward movement to a locked view. Both text to video prompts retain the street description and allow a few leaves to move. The fixed-view trial stays near one viewpoint, but it also invents a different street composition.

Again, five seconds, 768P and 16:9 were requested for both. The difference is useful to inspect, not proof of a repeatable camera path. Normal playback shows pace; pausing exposes changing windows, malformed objects or added text.

Your browser does not support video playback.
AI-generated example · Text to Video · 5 seconds · 768POriginal street request: the camera moves slowly forward. Watch how the view approaches the street.
Your browser does not support video playback.
AI-generated example · Text to Video · 5 seconds · 768PRevised camera sentence: “The camera stays locked in place.” The generated view is stable, with a different street composition.

Adapt text to video prompts to the available seconds

For five seconds, start with one continuous movement that can settle. Ten seconds can give a slower action more room. Fifteen seconds allows a longer progression, but a larger duration does not require more scenes. A single quiet shot may be exactly what your timeline needs.

Long text to video prompts are useful only when the extra words remove ambiguity. Describing six locations, multiple actors and a camera move for each location creates more competing requests than a short clip can clearly communicate. Generate separate shots if the edit genuinely needs separate places.

Does a longer prompt improve quality?

Length alone does not tell you that. Add the missing visual fact, remove a contradictory instruction, and judge the next output. A precise forty-word description may be easier to evaluate than a paragraph full of style labels.

Where should a beginner start?

Choose one of the text to video prompts above, set five seconds and select the frame shape your edit needs. Review the full text before generating. Use Text to Video for a scene created from words, or read the motion guide for existing photos if the first frame already exists.

For further terminology, Runway's text prompting guide also distinguishes visual description from motion description. Its model-specific controls differ from H3 Max; the examples and limits on this page refer to this generator.

Keep working on your video

  • Text to Video↗
  • B-Roll Generator↗
  • Camera Movement Prompts↗
  • Choose a Generation Mode↗
H3 Max

From an idea or an image to your next video.
H3 Max Turbo by fal.ai, based on MiniMax H3.

[email protected]

Video tools

  • Text to Video
  • Image to Video
  • Product Video Generator
  • Batch Product Videos
  • Video Ad Clips
  • B-Roll Generator

Use cases

  • Use Cases
  • Shopify Product Videos
  • Amazon Product Videos
  • Etsy Listing Videos
  • Skincare Videos
  • Jewelry Videos

Learn

  • Guides
  • Text to Video Prompts
  • Image to Video Prompts
  • Choose Your Starting Image
  • 480P vs 768P
  • Choose a Generation Mode

H3 Max

  • Pricing
  • Contact
  • Privacy
  • Terms
  • Report prohibited content
  • Cookie Policy

© 2026 H3 Max. All rights reserved.