patrickz.aiLet’s talk ↗

Create media / Guide

Direct AI images with a reusable visual brief

Create two images for different uses while preserving a consistent visual identity.

About 30 minutesBeginnerRead free · No sign-up

Before you start

An image tool you can access, a reference you have permission to use, and two intended placements.

Why this lesson exists

The image-model posts and this site’s artwork refresh show why purpose and reference quality matter. A hero image and a small card need related style but different compositions.

Do the exercise

  1. Define the visual contract

    List the subject’s stable traits, palette, medium and mood. Identify what must stay the same and what can change. Use your own or licensed references.

  2. Specify the placement

    For each asset record the aspect ratio, display size, focal action and crop-safe area. A thumbnail needs fewer competing details than a large illustration.

  3. Generate separately for each job

    Request one scene per asset. Carry the same reference and contract across prompts. Avoid asking a single collage to become every image on a website.

  4. Inspect at actual size

    Check identity, hands, objects, unwanted text and purpose. Compare at phone size. Save the prompt, reference and selected version, then export an appropriately sized web image.

A prompt to adapt

Replace the bracketed parts with your own practice details.

Create one [asset type] for [placement and aspect ratio]. Reference role: [identity/style]. Preserve [invariants]. Scene: [one focal action]. Use [palette/medium]. Keep important subjects inside the safe area. Avoid unwanted lettering and watermarks. The image must remain readable at [display size].

What this looks like

A hero can show two people building together. A debugging card should make one visible error the focus. Matching style does not require repeating the same composition.

Check your result

Use evidence from your output. A confident explanation from the AI is not enough.

  • Both images preserve the chosen identity and style.
  • Each image communicates its own job at actual size.
  • Prompts and selected versions are saved.

If it isn’t working

If a generation adds an extra hand or changes a key prop, request a targeted correction and inspect again. Fine-tuning is a separate advanced workflow; try clear references and prompting first.

Where this came from

Adapted from the archived LinkedIn theme “Train the Image Model Before It Trains Your Taste.” See the post coverage and editorial method.

Prepared September 2026. Tools and interfaces change; use current official setup instructions. Session lengths are estimates.

KEEP GOING

Your next useful step

Browse all 59 guides ↗