AI video & content

Text-to-video best practices

Learn how to write stronger prompts, choose useful settings, and improve text-to-video results in Flikly.

6 min readLast updated: May 2026

Prompt

A close-up product video of a modern desk lamp turning on in a cozy home office at night, soft cinematic lighting, slow camera push-in, warm mood, designed for a short social post.

SubjectActionSceneStyleMotionAspect ratio

Prompt quality

Clear subject
One main action
Defined setting
Style direction

Preview

Draft readyRefine or regenerate

Text-to-video turns your written direction into motion. Strong prompts and thoughtful settings help you get closer to the result you want—especially when you match the workflow and model to your goal and plan.

What text-to-video is best for

Text-to-video is useful when you are starting from an idea, script, concept, campaign, product description, hook, or scene direction. It works best when the prompt gives the model enough visual guidance to understand subject, action, and style.

It is a strong fit for early drafts, creative exploration, and short-form concepts before you invest in higher-quality or longer output.

Start with a clear goal

Before writing the prompt, decide what the video should accomplish. For example:

Explain a product
Introduce a brand
Create a social hook
Visualize a scene
Test an ad concept
Create a short-form post idea
Support a campaign

A clear goal keeps the prompt focused and makes it easier to judge whether the result is worth refining.

Write a strong prompt

A practical structure to follow:

Subject + action + setting + style + camera/motion + mood + output goal.

Example prompt

A close-up product video of a modern desk lamp turning on in a cozy home office at night, soft cinematic lighting, slow camera push-in, warm mood, designed for a short social post.

Describe what should appear on screen in plain language. Avoid unrelated details, conflicting instructions, or requests that depend on specific real people or brands you do not have rights to depict.

Prompt details that improve results

Subject

Who or what appears—product, person, object, or scene focus.

Action

What happens—turns on, walks in, pours, reveals, opens, etc.

Setting

Where it happens—home office, street, studio, kitchen, outdoors.

Style

Realistic, cinematic, minimal, animated, documentary, premium, etc.

Motion

Slow pan, push-in, orbit, handheld, static, or gentle drift.

Mood

Calm, energetic, premium, playful, dramatic, warm, or bold.

Format

Vertical, square, or landscape depending on where you plan to publish.

Keep prompts focused

Avoid asking for too many unrelated things in one generation.
Keep one main subject and one main action.
Use short, direct sentences.
Remove conflicting style instructions.
If the first result is close, refine the prompt instead of starting over from scratch.

Choose settings carefully

Match aspect ratio to where you plan to publish.
Duration can affect complexity and how long generation may take.
Resolution and quality options may depend on the selected model and your plan.
Some available models handle motion, realism, or stylized output differently.

For more on picking a workflow and model, see Choosing a model: Runway, Luma, Pika, and more.

Review and improve the result

Is the subject clear?
Is the first second strong enough for social?
Does the motion match the prompt?
Is the style consistent?
Does the format fit the target platform?
Would the video benefit from captions, trimming, or splitting?
Should you regenerate, refine, or try another available model?

Common prompt patterns

Short templates you can adapt—replace the bracketed parts with your own details.

Product showcase

Close-up of [product] in [setting], [action], [lighting/style], [camera motion], designed for [platform/format].

Founder or storytelling hook

Person in [setting] delivering [moment/action], [mood], [style], [camera motion], short social hook feel.

Educational explainer

Clear visual of [concept/subject] with [simple action], clean [style], steady [camera], easy-to-follow scene.

Event or announcement teaser

[Subject] in [setting], energetic [mood], quick [motion], bold [style], teaser for upcoming [event/launch].

Lifestyle or social scene

[Subject] doing [everyday action] in [relatable setting], natural [lighting], [mood], vertical-friendly framing.

Credits, speed, and expectations

Text-to-video generations may use credits depending on workflow, selected model, duration, resolution, and your plan. Credits are usage units, not minutes of video.

Generation time can vary by model, provider availability, duration, and resolution. AI video output often benefits from refinement—regeneration, trimming, captions, or trying another available model when results are close but not quite right.

For credit details, see Understanding credits and plan limits.

Common issues

Why did the video not match my prompt?

Simplify the prompt, make the subject and action clearer, and remove conflicting style instructions. If your plan supports it, image-to-video can help when you need a specific look.

Why does the motion look strange?

Use clearer camera and motion language—such as slow push-in, static shot, or gentle pan—and avoid asking for too many movements at once.

Why did the subject change between frames?

Keep the prompt focused on one subject and one action. For stronger visual consistency, consider image-to-video with a reference image.

Why is generation taking longer than expected?

Duration, resolution, the selected model, and provider load can all affect timing. Longer or higher-quality output often takes more time.

Should I regenerate or edit?

Regenerate if the concept is wrong. Edit, refine the prompt, or trim in the video editor if the result is close to what you want.

How do I make videos better for social platforms?

Use a clear hook in the first second, match aspect ratio to the platform, add captions where helpful, and keep scenes short and focused.

Next steps

Continue with these AI video & content guides.

Still need help?

Visit the Help Center or contact support if you need help with text-to-video prompts or settings.