AIVIDEO / FIELD NOTES

Text-to-Video AI Prompts: Write One Clear Shot

Describe a subject, an action and a camera choice without asking one clip to tell an entire story.

Try the sample workspace ↗
Blank storyboard panels and a miniature astronaut on a white desk.
AI-created illustration

Build text-to-video AI prompts around one shot

Start with what a viewer could see: a subject, an action and a setting. Then add a camera choice if it changes the meaning of the shot. A short scene with a clear action is easier to evaluate than a paragraph that mixes a story summary, several locations and conflicting camera moves.

This page helps you write a brief. AIVideo does not send that brief to a model or generate a video.

Use a simple brief structure

Subject: Who or what matters in the frame?

Action: What happens during this shot?

Setting: Where is it taking place?

Camera: How does the viewer see the action?

Appearance: Which lighting or visual detail is important enough to specify?

You do not need to fill every category with adjectives. If the camera should stay still, say that once. If lighting is not central to the idea, leave room to experiment later.

Work through an example

A vague starting point might be: “An amazing cinematic adventure with a cyclist.” It does not tell you which action to look for.

A more testable brief is: “A cyclist rides across an empty bridge at sunrise. A fixed wide camera keeps the entire bicycle in view.” You can now check the subject, action and framing separately.

For a second version, change only the camera instruction: “A side-tracking camera follows the cyclist.” Keep both versions in your notes. If everything changes at once, it becomes harder to understand which instruction mattered.

Avoid conflicting directions

A camera cannot be both fixed and circling the subject in the same uncomplicated shot. A close-up and a wide establishing view may be better treated as separate shots. Split a sequence into a small shot list when the action has distinct stages.

Prompt conventions are model-specific. Runway publishes its own prompting guidance; read the instructions for the model you actually choose rather than assuming one phrasing works everywhere.

Review before spending generation time

Ask whether the brief has one main action, whether the subject fits the chosen frame and whether you can describe a successful result without using the word “better.” Do not include confidential details in a third-party tool without reviewing its data practices.

In AIVideo, enter your scene, choose a frame format and review the format summary. Copy your text before leaving the page because the workspace does not save it.

Put your next step into practice

Open the AIVideo workspace to try the sample controls. For another part of the decision, read image-to-video ai: prepare the starting frame.

Common questions

How long should my prompt be?

Long enough to describe one clear shot. Add details when they affect the intended result, rather than aiming for a fixed word count.

Does Preview creative brief generate a clip?

No. It confirms the selected format. Copy your scene text before leaving the page.

KEEP EXPLORING

Image-to-Video AI: Prepare the Starting Frame ↗Video Aspect Ratio: Choose the Frame for the Shot ↗