← Image prompts
ImageGenerate

Create an image in a chosen visual style

Turn a scene idea into a coherent illustration with one clear style, useful composition details and practical constraints. Customize the scene, visual style, materials and layout.

Tools: An image generator that accepts a detailed text prompt.

What to provide

  • A subject and setting.
  • The visual style and visible traits that communicate it.
  • Composition and palette requirements.
Make it yours

Fill in what you know. Empty fields stay as placeholders.

Name what the image should depict and where it takes place.

Choose one style such as papercraft, mosaic, watercolor or neonpunk; describe its visual traits.

Describe materials, edges, texture, lighting or other traits that make the style recognizable.

List the important objects and details to include.

Name a short, coordinated set of colours.

Specify orientation, framing and where the subject sits.

List unwanted objects or content.

Ready to copy

Example Recorded test output

Input

Subject: a small neighborhood greenhouse. Style: layered papercraft. Style details: hand-cut colored paper layers, subtle paper fibers, crisp edges and gentle cast shadows. Scene details: three glasshouse bays, leafy seedlings on shelves and a watering can by the door. Palette: sage green, warm cream and terracotta. Composition: square, centered greenhouse with ample margin. Exclude: extra buildings.

Output

Generated square editorial illustration of a small greenhouse with three visible glasshouse bays, shelves of seedlings and a watering can. Sage, cream and terracotta layers with crisp edges and soft shadows make the papercraft direction visible. No lettering or additional buildings. The result is a generated image, not a photograph or construction plan.

AI-generated square papercraft illustration of a neighborhood greenhouse with three bays and seedlings
Result

Tool settings

One square image. Exact output dimensions and model selection were not exposed in this built-in run.

Improve the result

  • Keep the subject and composition, but change the visual treatment to [style] and describe its defining materials and marks.
  • Reduce background detail while retaining all named objects and the same framing.

Tested with Codex built-in image_gen (model name not exposed) on .

Only the papercraft greenhouse example was tested. Other styles and subjects are untested variations. Results may combine visual cues or interpret unfamiliar style names inconsistently; describe the materials, texture, color and composition you want. Check object structure and any text before using the image.

Revision 1

Worked examples

See the related guide, 24 ChatGPT image prompt labels with generated examples, for a practical glossary of photo, lighting, style and layout directions. The examples are illustrative and do not establish that slash-prefixed labels are built-in commands.