A reusable GPT Image Prompt is built from decisions, not adjectives. The prompt should tell the model what role the image plays and which visual choices matter most.
The prompt stack
Use this stack when writing:
- Purpose: where the image will be used.
- Subject: the main object, person, layout, or concept.
- Scene: background, props, environment, and context.
- Visual system: style, lighting, color, materials, and mood.
- Composition: framing, camera angle, aspect ratio, and text space.
- Constraints: what must not appear or change.
Reference prompt example
Create a product listing image for a ceramic coffee dripper. Match the reference image for product shape and matte texture. Place it on a clean white kitchen counter with soft morning light, a three-quarter angle, realistic shadows, and a square crop. Avoid extra logos, added text, distorted rim shape, and busy props.
Keep a prompt library
Save prompts by use case. A good library lets you reuse the same structure for future products, campaigns, or design systems while changing only the subject and constraints.