Prompting Basics

Good prompts give the AI a clear picture of what you want. Structure and specificity improve consistency and quality.

Core elements

• Subject — The main focus (person, object, scene).
• Action or state — What’s happening (for video) or how the subject appears (for images).
• Setting — Where it takes place (environment, location).
• Lighting — Quality, direction, and mood of light.
• Style — Look and feel (cinematic, commercial, realistic, illustrated, etc.).

Do’s

• Be specific where it matters (e.g. “soft window light from the left” instead of “nice lighting”).
• Use one main idea per sentence; add detail in short phrases.
• Include style and mood so the model matches your intent.
• For video, include motion and camera (e.g. “slow push-in”, “character turns toward camera”).

Don’ts

• Avoid long, run-on sentences with many unrelated ideas.
• Don’t overload a single prompt with too many subjects or actions.
• If results are off, change one variable at a time (lighting, framing, or style) to see what fixes it.

Using references

When you have a Character or Environment (or a reference image), select it in the tool. The model can use it to keep identity or location consistent. See References & Source Assets for more.

Different AI models may respond better to different wording; use the model’s info or detail page for prompt tips and examples.